Patent Application
20050256007
This invention relates to superconducting circuitry. More specifically, this invention relates to devices for quantum computation.
Research on what is now called quantum computing may have begun with a paper published by Richard Feynman. See Feynman, 1982, International Journal of Theoretical Physics 21, pp. 467-488, which is hereby incorporated by reference in its entirety. Feynman noted that a quantum system is inherently difficult to simulate with conventional computers but that observation of the evolution of an analogous quantum system could provide an exponentially faster way to solve the mathematical model of the quantum system of interest. In particular, solving a mathematical model for the behavior of a quantum system commonly involves solving a differential equation related to the Hamiltonian of the quantum system. David Deutsch noted that a quantum system could be used to yield a time savings, later shown to include exponential time savings, in certain computations. If one had a problem modeled in the form of an equation that represented the Hamiltonian of a quantum system, the behavior of the system could provide information regarding the solutions to the equation. See Deutsch, 1985, Proceedings of the Royal Society of London A 400, pp. 97-117, which is hereby incorporated by reference in its entirety.
A major activity in the quantum computing art is the identification of physical systems that can support quantum computation. This activity includes finding suitable qubits as well as developing systems and methods for controlling such qubits. As detailed in the following sections, a qubit serves as the basis for performing quantum computation.
The physical systems that are used in quantum computing are quantum computers. A quantum bit or “qubit” is the building block of a quantum computer in the same way that a conventional binary bit is a building block of a classical computer. A qubit is a quantum bit, the counterpart in quantum computing to the binary digit or bit of classical computing. Just as a bit is the basic unit of information in a classical computer, a qubit is the basic unit of information in a quantum computer. A qubit is conventionally a system having two or more discrete energy states. The energy states of a qubit are generally referred to as the basis states of the qubit. The basis states of a qubit are termed the |0> and |1> basis states. In the mathematical modeling of these basis states, each state is associated with an eigenstate of the sigma-z (σZ) Pauli matrix. See Nielsen and Chuang, 2000, Quantum Computation and Quantum Information, Cambridge University Press, which is hereby incorporated by reference in its entirety.
The state of a qubit can be in any superposition of two basis states, making it fundamentally different from a bit in an ordinary digital computer. A superposition of basis states arises in a qubit when there is a non-zero probability that the system occupies more than one of the basis states at a given time. Qualitatively, a superposition of basis states means that the qubit can be in both basis states |0> and |1> at the same time. Mathematically, a superposition of basis states means that the wave function that characterizes the overall state of the qubit, denoted |Ψ>, has the form
|Ψ>=a|0>+b|1>
where a and b are amplitudes respectively corresponding to probabilities |a|2 and |b|2. The amplitudes a and b each have real and imaginary components, which allows the phase of qubit to be modeled. The quantum nature of a qubit is largely derived from its ability to exist in a superposition of basis states, and for the state of the qubit to have a phase.
To complete a quantum computation using a qubit, the state of the qubit is typically measured (e.g., read out). When the state of the qubit is measured the quantum nature of the qubit is temporarily lost and the superposition of basis states collapses to either the |0> basis state or the |1> basis state, thus regaining its similarity to a conventional bit. The actual state of the qubit after it has collapsed depends on the amplitudes a and b immediately prior to the readout operation.
A survey of exemplary physical systems from which qubits can be formed is found in Braunstein and Lo (eds.), Scalable Quantum Computers, Wiley-VCH Verlag GmbH, Berlin (2001), which is hereby incorporated by reference in its entirety. Of the various physical systems surveyed, the systems that appear to be most suited for scaling (e.g., combined in such a manner such that they entangle with each other) are those physical systems that include superconducting structures such as superconducting qubits.
Superconducting qubits generally fall into two categories; phase qubits and charge qubits. Phase qubits store and manipulate information in the phase states of the device. Charge qubits store and manipulate information in the elementary charge states of the device. In superconducting materials, phase is a property of the material whereas elementary charges are represented by pairs of electrons called Cooper pairs. The division of such devices into two classes is outlined in Makhlin et al., 2001, “Quantum-State Engineering with Josephson-Junction Devices,” Reviews of Modern Physics 73, pp. 357-401 which is hereby incorporated by reference in its entirety.
Phase and charge are related values in superconductors and, at energy scales where quantum effects dominate, the Heisenberg uncertainty principle causes certainty in phase to lead to uncertainty in charge and, conversely, causes certainty in charge to lead to uncertainty in the phase of the system. Superconducting phase qubits are devices formed out of superconducting materials having a small number of distinct phase states and many charge states, such that when the charge of the device is certain, information stored in the phase states becomes delocalized and evolves quantum mechanically. Therefore, fixing the charge of a phase qubit leads to delocalization of the phase states of the qubit and subsequent useful quantum behavior in accordance with well-known principles of quantum mechanics.
Experimental realization of superconducting devices as qubits was made by Nakamura et al., 1999, Nature 398, p. 786, which is hereby incorporated by reference in its entirety. Nakamura et al. developed a charge qubit that demonstrates the basic operational requirements for a qubit. However, the Nakamura et al. charge qubits have unsatisfactorily short decoherence times and stringent control parameters. Decoherence time is the duration of time that it takes for a qubit to lose some of its quantum mechanical properties, e.g., the state of the qubit no longer has a definite phase. When the qubit loses it quantum mechanical properties, the phase of the qubit is no longer characterized by a superposition of basis states and the qubit is no longer capable of supporting all types of quantum computation.
Superconducting qubits have two modes of operation related to localization of the states in which information is stored. When the qubit is initialized or measured, the information is classical, 0 or 1, and the states representing that classical information are also classical in order to provide reliable state preparation. Thus, a first mode of operation of a qubit is to permit state preparation and measurement of classical information. A second mode of operation occurs during quantum computation, where the information states of the device become dominated by quantum effects such that the qubit can evolve controllably as a coherent superposition of those states and, in some instances, even become entangled with other qubits in the quantum computer. Thus, qubit devices provide a mechanism to localize the information states for initialization and readout operations, and de-localize the information states during computation. Efficient functionality of both of these modes and, in particular, the transition between them in superconducting qubits is a challenge that has not been satisfactorily resolved in the prior art.
A proposal to build a quantum computer from superconducting qubits was published in 1997. See Bocko et al., 1997, IEEE Trans. Appl. Supercon. 7, p. 3638, which is hereby incorporated by reference in its entirety. See also Makhlin et al., 2001, Rev. Mod. Phys. 73, p. 357 which is hereby incorporated by reference in its entirety. Since then, designs based on many other types of qubits have been introduced. One such design is based on the use of superconducting phase qubits. See Mooij et al., 1999, Science 285, 1036; and Orlando et al., 1999, Phys. Rev. B 60, 15398, which are hereby incorporated by reference in their entireties. In particular, quantum computers based on persistent current qubits, which are one type of superconducting phase qubit, have been proposed.
The superconducting phase qubit is well known and has demonstrated long coherence times. See, for example, Orlando et al., 1999, Phys. Rev. B 60, 15398, and Il'ichev et al., 2003, Phys. Rev. Lett. 91, 097906, which are hereby incorporated by reference in their entireties. Some other types of superconducting phase qubits comprise superconducting loops interrupted by more or less than three Josephson junctions. See, e.g., Blatter et al., 2001, Phys. Rev. B 63, 174511, and Friedman et al., 2000, Nature, 406, 43, which are hereby incorporated by reference in their entireties.
The difference between any two Josephson junctions in the persistent current qubit is characterized by a coefficient, termed α, which typically ranges from about 0.5 to about 1.3. In some instances, the term a for a pair of Josephson junctions in the persistent current qubit is the ratio of the critical current between the two Josephson junctions in the pair. The critical current of a Josephson junction is the minimum current through the junction at which the junction is no longer superconducting. That is, below the critical current, the junction is superconducting whereas above the critical current, the junction is not superconducting. Thus, for example, the term a for junctions 101-1 and 101-2 is defined as the ratio between the critical current of junction 101-1 and the critical current of junction 101-2.
Referring to
Persistent current qubit 101 has a two-dimensional potential with respect to the phase across Josephson junctions 101-1, 101-2, and 101-3. In some instances, persistent current qubit 101 is biased with a magnetic flux Φx, such that the two-dimensional potential profile includes regions of local energy minima, where the local energy minima are separated from each other by small energy barriers and are separated from other regions by large energy barriers. In some instances, this potential has the shape of double well potential 100B (
When wells 160-0 and 160-1 are at or near degeneracy, meaning that they are at the same or nearly the same energy potential as illustrated in
There are broad classes of condensed matter systems that have states defined by the presence and absence of extra charge, or the excess charge exists in either a ground or an excited state. Such systems are diverse and have long held theoretical and experimental interest, e.g. Millikan, 1911, Phys. Rev. 32, pp. 349-397, which is hereby incorporated by reference in its entirety. There has been attention directed to semiconductor systems such as quantum dots. Research has been conducted using single particle electronics because they hold promise for conventional computers. Subsequently, proposals were made for these systems as quantum computers. While these systems collectively could be called charge qubits, this term is reserved herein for superconducting qubits. Specifically, a superconducting charge qubit has as basis states the presence (charge=2e, or some multiple thereof of 2e) or absence (charge=0) of charge on a small superconducting island. For charge qubits, the Coulomb energy EC=e2/2C exceeds the Josephson energy EJ of the qubit.
A charge qubit is a small (mesoscopic) island of superconductor separated by a Josephson junction from a large superconductor (reservoir), see
where the dimensionless gate charge ng≈CgVg/2e is determined in operation by the gate voltage Vg, and in fabrication by the capacitance Cg. Here it is assumed that ng≈½. The bias term for the charge qubit is proportional to σZ. A finite Josephson energy allows transition between the states with tunnel splitting Δ proportional to EJ. The Josephson energy of the Josephson junction connecting the island to the reservoir can be made tunable. Tuning the dimensionless gate and the Josephson energy allows one to independently bias the charge and vary the tunneling rate of the charge qubit. See Nakamura et al., 1999, Nature 398, pp. 786-788; and Makhlin et al., 2001, Rev. Mod. Phys. 73, pp. 357-401, each of which is hereby incorporated by reference in its entirety.
Computer scientists concerned with complexity routinely use the definitions of different complexity classes. The number of complexity classes is ever changing, as new ones are defined and existing ones merge through advancements made in computer science. The complexity classes known as non-deterministic polynomial-time (NP), NP-complete (NPC), and NP-hard (NPH) are all classes of decision problems. Decision problems have binary outcomes.
Problems in NP are computational problems for which there exists polynomial time verification. That is, it takes no more than polynomial time (class P) in the size of the problem to verify a potential solution. It may take more than polynomial time to create a potential solution. NP-hard problems take longer to verify a potential solution. For each NP-hard problem, there is an NP-complete problem that can be reduced to the NP-hard problem. However, NP-complete problems that can be reduced to a NP-hard problem do not enjoy polynomial time verification.
Problems in NPC can be defined as problems in NP that have been shown to be equivalent to, or harder to solve, than a known problem in NPC. Equivalently, the problems in NPC are problems in NP that are also in NPH. This can be expressed as NPC=NP∩NPH.
A problem is equivalent, or harder to solve, than a known problem in NPC if there exists a polynomial time reduction to the instant problem from the known problem in NPC. Reduction can be regarded as a generalization of mapping. The mappings can be one to one functions, many to one functions, or make use of oracles, etc. The concepts of complexity classes and how they define the intractability of certain computational problems is found in, for example, Garey and Johnson, 1979, Computers and Intractability: A Guide to the Theory of NP-Completeness, Freeman, San Francisco, ISBN: 0716710455, which is hereby incorporated by reference in its entirety. Also see, Cormen, Leiserson, and Rivest, 1990, Introduction to Algorithms, MIT Press, Cambridge, ISBN: 0262530910.
Analogous to the way a classical computer is built using wires and logic gates, a quantum computer can be built using quantum circuits comprised of “wires” and “unitary gates.” Here, the wire is not a physical entity. Rather, it represents the state of the qubit in time. The “unitary gates” are applied at precise times to specific qubits to effect evolution of the qubit in accordance with the circuit model for quantum computing. The circuit model of quantum computing is a standard and universal model used by many practitioners in the art. The circuit model is universal in the sense that it is able to convert any input state into any output state. The elements of the circuit model are that a small set of one- and two-qubit unitary gates are applied to the qubits with precise timing. The circuit model of quantum computing can implement algorithms such as Shor's algorithm for factoring numbers or Grover's algorithm for searching databases. Shor's algorithm provides an exponential speedup relative to classical (non-quantum) computers for factoring numbers. Grover's application provides a polynomial speed up relative to classical computers for searching databases. See, for example, Nielsen and Chuang, 2000, Quantum Computation and Quantum Information, Cambridge University Press, which is hereby incorporated by reference in its entirety.
An example of the circuit model is shown in
The following subsections discuss the adiabatic theorem of quantum mechanics and introduce adiabatic quantum computing.
One definition of an adiabatic process is a process that occurs in a system without heat entering or leaving the system. There exists a theorem in quantum mechanics that provides a suitable framework for such processes. The adiabatic theorem of quantum mechanics has several versions but a notable element of many such versions is as follows. A quantum system prepared in its ground state will remain in the ground state of the various instantaneous Hamiltonians through which it passes, provided the changes are made sufficiently slowly. This form of change is termed adiabatic change. Such a system is adiabatic because the population of the various states of the quantum system has not been altered as a result of the change. Hence, if the populations have not changed, the temperature of the system has not changed, and therefore no heat has entered or left the system.
In 2000, a form of quantum computing, termed adiabatic quantum computing, was proposed. See, for example, Farhi et al., 2001, Science 292, pp. 472-475, which is hereby incorporated by reference in its entirety. In adiabatic quantum computing (AQC), the problem to be solved is encoded into a physical system such that departures from the solution to the problem incur a net energy cost to the system. AQC is universal in that it is able to convert any input state into any output state. However, unlike the circuit model of quantum computing, there is no application of a predetermined set of one- and two-qubit unitary gates at precise times. It is believed that AQC can be used to find solutions to some problems with greater efficiency than the circuit model. Such problems include problems contained in, and related to, the NP, NP-hard, and NP-complete classes.
As shown in
One computational problem that can be solved with adiabatic quantum computing is the
The
Consider an instance of a positive number K and a graph G=(V, E), having a set of vertices V={v1, . . . , v|V|}, and a set of edges E={e1, . . . , ei, . . . , e|E|}, where ei=(vj,vk) for all 1<j,k<|V|. The graph's edges have weights w(ei), w(vj,vk), or wjk that are positive. The explicit decision problem is whether there is a partition of V, i.e., V1⊂V; V2⊂V, and V1∪V2=V, such that the sum of the weights of the edges crossing the partition is equal or greater than some given predefined positive integer K, e.g.,
An optional addition to the definitions above is the graph may have vertex weights that are also positive w(vi) or wi. Using this alternative,
Mathematically, solving
One computational problem that can be solved with adiabatic quantum computing is the
Mathematically, solving
Mathematically, solving
The question of whether superconducting qubits can be used to implement adiabatic quantum computing (AQC) has been posed in the art. However, such proposals are unsatisfactory because they either lack enabling details on the physical systems on which AQC would be implemented or they rely on qubits that have not been shown to successfully perform an n-qubit quantum computation, where n is greater than 1 and the quantum computation requires entanglement of qubits. For example, Kaminsky and Lloyd, 2002, “Scalable Architecture for Adiabatic Quantum Computing of NP-Hard Problems,” in Quantum Computing & Quantum Bits in Mesoscopic Systems, Kluwer Academic, Dordrecht, Netherlands, also published as arXiv.org: quant-ph/0211152, which is hereby incorporated by reference in its entirety, suggests that AQC can be performed with persistent current qubits, without explicitly stating how. As another example, W. M. Kaminsky, S. Lloyd, T. P. Orlando, 2004, “Scalable Superconducting Architecture for Adiabatic Quantum Computation,” arXiv.org: quant-ph/0403090, hereby incorporated by reference, describes a method and structure for AQC for a type of persistent current qubit that has not been shown to support multi-qubit quantum computation. This reference shows a system of logical qubits without giving explicit construction of the logical qubits from physical qubits or an explicit coupling between more than two qubits.
Accordingly, given the above background, there is a need in the art for improved systems-and methods for adiabatic quantum computing. Discussion or citation of a reference herein shall not be construed as an admission that such reference is prior art to the present invention.
The present invention addresses the need in the art for improved systems and methods for adiabatic quantum computing. In some embodiments of the present invention, a graph based computing problem, such as
In one aspect of the present invention, the plurality of qubits that represents the graph is initialized to a first state that does not permit the qubits to quantum tunnel. Then, the plurality of qubits is set to an intermediate state in which quantum tunneling between individual basis states within each qubit in the plurality of qubits can occur. In preferred embodiments, the change to the intermediate state occurs adiabatically. In other words, for any given instant t that occurs during the change to the intermediate state or while the qubits are in the intermediate state, the plurality of qubits are in the ground state of an instantaneous Hamiltonian that describes the plurality of qubits at the instant t. The qubits remain in the intermediate state that permits quantum tunneling between basis states for a period of time that is sufficiently long enough to allow the plurality of qubits to reach a solution for the computation problem represented by the plurality of qubits.
Once the qubits have been permitted to quantum tunnel for a sufficient period of time, the state of the qubits is adjusted such that they reach some final state that either does not permit quantum tunneling or, at least, does not permit rapid quantum tunneling. In preferred embodiments, the change to the final state occurs adiabatically. In other words, for any given instant t that occurs during the change to the final state, the plurality of qubits are in the ground state of an instantaneous Hamiltonian that describes the plurality of qubits at the instant t.
In other examples of the systems and methods of the present invention, the plurality of qubits that represents the graph is initialized to a first state that does permit the qubits to quantum tunnel. The state of the quantum system is changed once the qubits have been permitted to quantum tunnel for a sufficient period of time. The state of the qubits is adjusted such that they reach some final state that either does not permit quantum tunneling or, at least, does not permit rapid quantum tunneling. In preferred embodiments, the change to the final state occurs adiabatically.
Some embodiments of the present invention are universal quantum computers in the adiabatic quantum computing model. Some embodiments of the present invention include qubits with single-qubit Hamiltonian terms and at least one two-qubit Hamiltonian term.
A first aspect of the invention provides a method for quantum computing using a quantum system comprising a plurality of superconducting qubits. The quantum system is characterized by an impedance. Also, the quantum system is capable of being in any one of at least two configurations at any given time. These at least two configurations include a first configuration characterized by an initialization Hamiltonian HO as well as a second configuration characterized by a problem Hamiltonian HP. The problem Hamiltonian has a ground state. Each respective first superconducting qubit in the plurality of superconducting qubits is arranged with respect to a respective second superconducting qubit in the plurality of superconducting qubits such that the first respective superconducting qubit and the corresponding second respective superconducting qubit define a predetermined coupling strength. The predetermined coupling strengths between each of the first respective superconducting qubit and corresponding second respective superconducting qubit collectively define a computational problem to be solved. In this first aspect of the invention, the method comprises initializing the quantum system to the initialization Hamiltonian HO. The quantum system is then adiabatically changed until it is described by the ground state of the problem Hamiltonian HP. The state of the quantum system is then read out by probing an observable of the σX Pauli matrix operator.
In some embodiments in accordance with the first aspect of the invention, the reading step comprises measuring an impedance of the quantum system. In some embodiments the reading step comprises determining a state of a superconducting qubit in the plurality of superconducting qubits. In some embodiments, the reading step differentiates a ground state of the superconducting qubit from an excited state of the superconducting qubit. In some embodiments, a superconducting qubit in the plurality of superconducting qubits is a persistent current qubit. In some embodiments, the reading step measures a quantum state of the superconducting qubit as a presence or an absence of a voltage. In some embodiments, a superconducting qubit in the plurality of superconducting qubits is capable of tunneling between a first stable state and a second stable state when the quantum system is in the first configuration.
In some embodiments, a superconducting qubit in the plurality of superconducting qubits is capable of tunneling between a first stable state and a second stable state during the adiabatic changing step. In some embodiments, the adiabatic changing step occurs during a time period that is between 1 nanosecond and 100 microseconds. In some embodiments, the initializing step includes applying a magnetic field to the plurality of superconducting qubits in the direction of a vector that is perpendicular to a plane defined by the plurality of superconducting qubits. In some embodiments, a superconducting qubit in the plurality of superconducting qubits is a persistent current qubit.
A second aspect of the invention provides a method for quantum computing using a quantum system that comprises a plurality of superconducting qubits. The quantum system is coupled to an impedance readout device. The quantum system is capable of being in any one of at least two configurations at any given time. The at least two configurations include a first configuration characterized by an initialization Hamiltonian H0, and a second Hamiltonian characterized by a problem Hamiltonian HP. The problem Hamiltonian HP has a ground state. Each respective first superconducting qubit in the plurality of superconducting qubits is arranged with respect to a respective second superconducting qubit in the plurality of superconducting qubits such that the first respective superconducting qubit and the second respective superconducting qubit define a predetermined coupling strength. The predetermined coupling strength between each said first respective superconducting qubit and corresponding second respective superconducting qubit collectively define a computational problem to be solved. In this second aspect of the invention, method comprises initializing the quantum system to the initialization Hamiltonian HO. Then the quantum system is adiabatically changed until it is described by the ground state of the problem Hamiltonian HP. The state of the quantum system is then read out through the impedance readout device thereby solving the computational problem.
In some embodiments in accordance with this second aspect of the invention, the reading step measures a quantum state of a superconducting qubit in the plurality of superconducting qubits as a presence or an absence of a voltage. In some embodiments, the reading step differentiates a ground state of the superconducting qubit from an excited state of the superconducting qubit. In some embodiments, a superconducting qubit in the plurality of superconducting qubits is (i) a phase qubit in the charge regime or (ii) a persistent current qubit. In some embodiments, a superconducting qubit in the plurality of superconducting qubits is capable of tunneling between a first stable state and a second stable state when the quantum system is in the first configuration. In some embodiments, a superconducting qubit in the plurality of superconducting qubits is capable of tunneling between a first stable state and a second stable state during the adiabatic changing step. In some embodiments, the adiabatic changing step occurs during a time period that is greater than 1 nanosecond and less than 100 microseconds. In some embodiments, the initializing step includes applying a magnetic field to the plurality of superconducting qubits in the direction of a vector that is perpendicular to a plane defined by the plurality of superconducting qubits. In some embodiments, a superconducting qubit in the plurality of superconducting qubits is a persistent current qubit.
A third aspect of the invention provides a method of determining a quantum state of a first target superconducting qubit. The method comprises presenting a plurality of superconducting qubits including a first target superconducting qubit in the plurality of superconducting qubits. A problem Hamiltonian describes (i) the quantum state of the plurality of superconducting qubits and (ii) each coupling energy between qubits in the plurality of qubits. The problem Hamiltonian is at or near a ground state. An rf-flux is added to the first target superconducting qubit. The rf-flux has an amplitude that is less than one flux quantum. An amount of an additional flux in the first target superconducting qubit is adiabatically varied. A presence or an absence of a dip in a voltage response of a tank circuit that is inductively coupled with the first target superconducting qubit during the adiabatically varying step is observed thereby determining the quantum state of the first target superconducting qubit.
In some embodiments in accordance with this third aspect of the invention, each superconducting qubit in the plurality of superconducting qubits is in a quantum ground state during all or a portion of the adding step, the adiabatically varying step, and the observing step. In some embodiments, the problem Hamiltonian corresponds to a terminus of an adiabatic evolution of the plurality of superconducting qubits. In some embodiments, the method further comprises biasing all or a portion of the superconducting qubits in the plurality of superconducting qubits. The problem Hamiltonian further describes a biasing on the first target superconducting qubit. In some embodiments, an energy of the biasing step exceeds the tunneling energy of a tunneling element of the Hamiltonian of the first target superconducting qubit, thereby causing tunneling to be suppressed in the first target superconducting qubit during an instance of the biasing step, adding step and the adiabatically varying step.
In some embodiments in accordance with this third aspect of the invention, the method further comprises adiabatically removing additional flux that was added to the first target superconducting qubit during the adiabatically varying step. In some embodiments, the adiabatically varying step comprises adiabatically varying the additional flux in accordance with a waveform selected from the group consisting of periodic, sinusoidal, triangular, and trapezoidal. In some embodiments, the adiabatically varying step comprises adiabatically varying the additional flux in accordance with a low harmonic Fourier approximation of a waveform selected from the group consisting of periodic, sinusoidal, triangular, and trapezoidal. In some embodiments, the additional flux has a direction that is deemed positive or negative. In some embodiments, the adiabatically varying step is characterized by a waveform that has an amplitude that grows with time. The amplitude of the waveform corresponds to an amount of additional flux that is added to the first target superconducting qubit during the adiabatically varying step. In some embodiments, the additional flux has an equilibrium point that varies with time. In some embodiments, the additional flux is either unidirectional or bidirectional. In some embodiments, the additional flux has a frequency of oscillation between about 1 cycle per second and about 100 kilocycles per second.
In some embodiments in accordance with the third aspect of the invention, the adding step comprises adding the rf-flux using (i) an excitation device that is inductively coupled to the first target superconducting qubit or (ii) a the tank circuit. In some embodiments, the method further comprises repeating the adding step and the adiabatically varying step between 1 time and 100 times. In such embodiments, the presence or absence of the dip in the voltage response of the tank circuit is observed as an average of the voltage response of the tank circuit across each instance of the adiabatically varying step.
In some embodiments in accordance with the third aspect of the invention, the first target superconducting qubit is flipped from an original basis state to an alternate basis state during the adiabatically varying step. The method further comprises returning the first target superconducting qubit to its original basis state by adiabatically removing additional flux in the qubit after the adiabatically varying step. In some embodiments, the adiabatically varying step does not alter the quantum state of each of superconducting qubits in the plurality of superconducting qubits other than the first target superconducting qubit. In some embodiments, the method further comprises recording a presence or an absence of the dip in the voltage response of the tank circuit.
In some embodiments in accordance with the third aspect of the invention, the method further comprises adding a second rf-flux to a second target superconducting qubit in the plurality of superconducting qubits. The second rf-flux has an amplitude that is less than one flux quantum. Then an amount of a second additional flux in the second target superconducting qubit is adiabatically varied. A presence or an absence of a second dip in a voltage response of a tank circuit that is inductively coupled with the second target superconducting qubit during said adiabatically varying is observed, thereby determining the quantum state of the second target superconducting qubit.
In some embodiments in accordance with the third aspect of the invention, the method further comprises designating a different superconducting qubit in the plurality of superconducting qubits as the first target superconducting qubit. The adding step and the adiabatically varying step are then reperformed with the different superconducting qubit as the first target superconducting qubit. The designating and reperforming are repeated until all or a portion (e.g., most, almost all, at least eighty percent) of the superconducting qubits in the plurality of superconducting qubits have been designated as the first target superconducting qubit.
In some embodiments in accordance with the third aspect of the invention, a tank circuit is inductively coupled with the first target superconducting qubit. The method further comprises performing an adiabatic quantum computation step for an amount of time with the plurality of superconducting qubits prior to the adding step. The amount of time is determined by a factor the magnitude of which is a function of a number of qubits in the plurality of superconducting qubits. An amount of an additional flux in the first target superconducting qubit is adiabatically varied. Then, a presence or an absence of a dip in the voltage response of a tank circuit during the adiabatically varying step is observed, thereby determining the quantum state of the first target superconducting qubit. In some embodiments, the presence of a dip in the voltage response of the tank circuit corresponds to the first target superconducting qubit being in a first basis state. The absence of a dip in the voltage response of the tank circuit corresponds to the target superconducting qubit being in a second basis state.
In some embodiments in accordance with the third aspect of the invention, the adiabatically varying step further comprises identifying an equilibrium point for the additional flux using an approximate evaluation method. In some embodiments, the method further comprises classifying the state of the first target qubit as being in the first basis state when the dip in the voltage across the tank circuit occurs to the left of the equilibrium point and classifying the state of the first target qubit as being in the second basis state when the dip in the voltage across the tank circuit occurs to the right of the equilibrium point.
A fourth aspect of the present invention comprises a method for adiabatic quantum computing using a quantum system comprising a plurality of superconducting qubits. The quantum system is capable of being in any one of at least two quantum configurations at any give time. The at least two quantum configurations include a first configuration described by an initialization Hamiltonian HO and a second configuration described by a problem Hamiltonian HP. The Hamiltonian HP has a ground state. The method comprises initializing the quantum system to the first configuration. Then the quantum system is adiabatically changed until it is described by the ground state of the problem Hamiltonian HP. Then the state of the quantum system is read out.
In some embodiments in accordance with the fourth aspect of the invention, each respective first superconducting qubit in the plurality of superconducting qubits is arranged with respect to a respective second superconducting qubit in the plurality of superconducting qubits such that the first respective superconducting qubit and the corresponding second respective superconducting qubit define a predetermined coupling strength. The predetermined coupling strength between each of the first respective superconducting qubits and corresponding second respective superconducting qubits in the plurality of superconducting qubits collectively define a computational problem to be solved. In some instances, the problem Hamiltonian HP comprises a tunneling term for each of the respective superconducting qubits in the plurality of superconducting qubits. The energy of the tunneling term for each respective superconducting qubit in the plurality of superconducting qubits is less than the average of the predetermined coupling strengths between each of the first respective superconducting qubits and second respective superconducting qubits in the plurality of superconducting qubits.
In some embodiments in accordance with the fourth aspect of the invention, the reading out step comprises probing an observable of the σX Pauli matrix operator or σZ Pauli matrix operator. In some embodiments, a tank circuit is in inductive communication with all or a portion of the superconducting qubits in the plurality of superconducting qubits. In such embodiments, the reading out step comprises measuring a voltage across the tank circuit. In some embodiments, the superconducting qubit in the plurality of superconducting qubits is a persistent current qubit.
A fifth aspect of the present invention provides a structure for adiabatic quantum computing comprising a plurality of superconducting qubits. The plurality of superconducting qubits is capable of being in any one of at least two configurations at any give time. The at least two configurations include a first configuration characterized by an initialization Hamiltonian H0 and a second Hamiltonian characterized by a problem Hamiltonian HP. The problem Hamiltonian has a ground state. Each respective first superconducting qubit in the plurality of superconducting qubits is coupled with a respective second superconducting qubit in the plurality of superconducting qubits such that the first respective superconducting qubit and the corresponding second respective superconducting qubit define a predetermined coupling strength. The predetermined coupling strength between each of the first respective superconducting qubits and the corresponding second respective superconducting qubits collectively define a computational problem to be solved. A tank circuit is inductively coupled to all or a portion of the plurality of superconducting qubits.
In some embodiments in accordance with the fifth aspect of the invention, a superconducting qubit in the plurality of superconducting qubits is a persistent current qubit. In some embodiments, the tank circuit has a quality factor that is greater than 1000. In some embodiments, the tank circuit comprises an inductive element. The inductive element comprises a pancake coil of superconducting material. In some embodiments, the pancake coil of a superconducting material comprising a first turn and a second turn. The superconducting material of the pancake coil is niobium. Furthermore, there is a spacing of 1 about micrometer between the first turn and the second turn of the pancake coil.
In some embodiments in accordance with the fifth aspect of the invention, the tank circuit comprises an inductive element and a capacitive element that are arranged in parallel or in series with respect to each other. In some embodiments, the tank circuit comprises an inductive element and a capacitive element that are arranged in parallel with respect to each other and the tank circuit has an inductance between about 50 nanohenries and about 250 nanohenries. In some embodiments, the tank circuit comprises an inductive element and a capacitive element that are arranged in parallel with respect to each other and the tank circuit has a capacitance between about 50 picofarads and about 2000 picofarads. In some embodiments, the tank circuit comprises an inductive element and a capacitive element that are arranged in parallel with respect to each other and the tank circuit has a resonance frequency between about 10 megahertz and about 20 megahertz. In some embodiments, the tank circuit has a resonance frequency fT that is determined by the equality:
fT=ωT/2π=1/{square root}{overscore (LTCT)}
such that
In some embodiments in accordance with the fifth aspect of the invention, the tank circuit comprises one or more Josephson junctions. In some embodiments, the structure further comprises means for biasing the one or more Josephson junctions of the tank circuit. In some embodiments, the structure further comprises an amplifier connected across the tank circuit in such a manner that the amplifier can detect a change in voltage across the tank circuit. In some embodiments, the amplifier comprises a high electron mobility field-effect transistor (HEMT) or a pseudomorphic high electron mobility field-effect transistor (PHEMT). In some embodiments, the amplifier comprises a multi-stage amplifier. In some embodiments, the multi-stage amplifier comprises two, three, or four transistors. In some embodiments, structure further comprises a helium-3 pot of a dilution refrigerator that is thermally coupled to all or a portion of the plurality of superconducting qubits. to.
In some embodiments in accordance with the fifth aspect of the invention, the structure further comprising means for driving the tank circuit by a direct bias current IDC. In some embodiments, the structure further comprises means for driving the tank circuit by an alternating current IRF of a frequency ω close to the resonance frequency ω0 of the tank circuit. In some embodiments, the total externally applied magnetic flux to a superconducting qubit in the plurality of superconducting qubits, ΦE, is
ΦE=ΦDC+ΦRF
In some embodiments in accordance with the fifth aspect of the invention, the structure further comprises an amplifier connected across the tank circuit and means for measuring a total impedance of the tank circuit, expressed through the phase angle % between driving current IRF and the tank voltage. In some embodiments, the means for measuring a total impedance of the tank circuit is an oscilloscope.
A sixth aspect of the invention provides a computer program product for use in conjunction with a computer system. The computer program product comprises a computer readable storage medium and a computer program mechanism embedded therein. The computer program mechanism comprises instructions for initializing a quantum system comprising a plurality of superconducting qubits to an initialization Hamiltonian HO. The quantum system is capable of being in one of at least two configurations at any give time. The at least two configurations include a first configuration characterized by the initialization Hamiltonian HO and a second configuration characterized by a problem Hamiltonian HP. Each respective first superconducting qubit in the plurality of superconducting qubits is arranged with respect to a respective second superconducting qubit in the plurality of superconducting qubits such that the first respective superconducting qubit and the second respective superconducting qubit define a predetermined coupling strength. The predetermined coupling strengths between each of the first respective superconducting qubits and the second respective superconducting qubits collectively define a computational problem to be solved. The computer program mechanism further comprises instructions for adiabatically changing the quantum system until it is described by the ground state of the problem Hamiltonian HP and instructions for reading out the state of the quantum system.
In some embodiments in accordance with this sixth aspect of the invention, the computer program mechanism further comprises instructions for repeating the instructions for biasing, instructions for adding, and instructions for adiabatically varying between 1 time and 100 times inclusive. The presence or absence of the voltage response of the tank circuit is observed as an average of the voltage response of the tank circuit to each instance of the instructions for adiabatically changing that are executed by the instructions for repeating.
A seventh aspect of the invention comprises a computer program product for use in conjunction with a computer system. The computer program product comprises a computer readable storage medium and a computer program mechanism embedded therein. The computer program mechanism determines a quantum state of a first target superconducting qubit in a plurality of superconducting qubits. The computer program mechanism comprises instructions for initializing a plurality of superconducting qubits so that they are described by a problem Hamiltonian. The problem Hamiltonian describes (i) the quantum state of the plurality of superconducting qubits and (ii) each coupling energy between qubits in the plurality of qubits. The problem Hamiltonian is at or near a ground state. The computer program mechanism further comprises instructions for adding an rf-flux to the first target superconducting qubit. The rf-flux has an amplitude that is less than one flux quantum. The computer program mechanism further comprises instructions for adiabatically varying an amount of an additional flux in the first target superconducting qubit and observing a presence or an absence of a dip in a voltage response of a tank circuit that is inductively coupled with the first target superconducting qubit during the adiabatically varying step.
In some embodiments in accordance with this seventh aspect of the invention, each superconducting qubit in the plurality of superconducting qubits is in a quantum ground state during all or a portion of the instructions for initializing, instructions for adding, and the instructions for adiabatically varying. In some embodiments, the problem Hamiltonian corresponds to a terminus of an adiabatic evolution of the plurality of superconducting qubits. In some embodiments, the computer program product further comprises instructions for biasing all or a portion of the superconducting qubits in the plurality of superconducting qubits. In such embodiments, the problem Hamiltonian additionally describes the biasing on the qubits in the plurality of superconducting qubits. In some embodiments, an energy of the biasing exceeds the tunneling energy of a tunneling element of the Hamiltonian of a superconducting qubit in the plurality of superconducting qubits thereby causing tunneling to be suppressed in the superconducting qubit during an instance of the instructions for biasing, instructions for adding and the instructions for adiabatically varying.
In some embodiments in accordance with the seventh aspect of the invention, the computer program mechanism further comprises instructions for adiabatically removing additional flux that was added to the first target superconducting qubit during the instructions for adiabatically varying. In some embodiments, the instructions for adiabatically varying comprise instructions for adiabatically varying the additional flux in accordance with a waveform selected from the group consisting of periodic, sinusoidal, triangular, and trapezoidal. In some embodiments, the instructions for adiabatically varying comprise instructions for adiabatically varying the additional flux in accordance with a low harmonic Fourier approximation of a waveform selected from the group consisting of periodic, sinusoidal, triangular, and trapezoidal. In some embodiments, the additional flux has a direction that is deemed positive or negative. In some embodiments, the instructions for adiabatically varying are characterized by a waveform that has an amplitude that grows with time and such that the amplitude of the waveform corresponds to an amount of additional flux that is added to the first target superconducting qubit during an instance of the instructions for adiabatically varying.
In some embodiments in accordance with the seventh aspect of the invention, the additional flux has an equilibrium point that varies with time. In some embodiments, the additional flux is either unidirectional or bidirectional. In some embodiments, the additional flux has a frequency of oscillation between about 1 cycle per second and about 100 kilocycles per second. In some embodiments, the instructions for adding comprise instructions for adding the rf-flux using (i) an excitation device that is inductively coupled to the first target superconducting qubit or (ii) the tank circuit. In some embodiments, the computer program mechanism further comprises instructions for repeating the instructions for adding and the instructions for adiabatically varying between 1 time and 100 times. In such embodiments, the presence or absence of the voltage response of the tank circuit is observed as an average of the voltage response of the tank circuit across each instance of the instructions for adiabatically varying that is executed by the instructions for repeating.
An eight aspect of the invention comprises a computer system for determining a quantum state of a first target superconducting qubit in a plurality of superconducting qubits. The computer system comprises a central processing unit and a memory, coupled to the central processing unit. The memory stores instructions for biasing all or a portion of the qubits in the plurality of superconducting qubits other than the first target superconducting qubit. A problem Hamiltonian describes (i) the biasing on the qubits in the plurality of superconducting qubits and (ii) each coupling energy between respective superconducting qubit pairs in the plurality of superconducting qubits. The problem Hamiltonian is at or near a ground state. The memory further stores instructions for adding an rf-flux to the first target superconducting qubit. The rf-flux has an amplitude that is less than one flux quantum. The memory further stores instructions for adiabatically varying an amount of an additional flux in the first target superconducting qubit and observing a presence or an absence of a dip in a voltage response of a tank circuit that is inductively coupled with the first target superconducting qubit during a time when the instructions for adiabatically varying are executed.
A ninth aspect of the present invention provides a computation device for adiabatic quantum computing comprising a plurality of superconducting qubits. Each superconducting qubit in the plurality of superconducting qubits comprises two basis states associated with the eigenstates of a σZ Pauli matrix that can be biased. The quantum computation device further comprises a plurality of couplings. Each coupling in the plurality of couplings is disposed between a superconducting qubit pair in the plurality of superconducting qubits. Each term Hamiltonian for a coupling in the plurality of couplings has a principal component proportional to σZ{circle over (x)}σZ. The sign for at least one principal component proportional to σZ{circle over (x)}σZ for a coupling in the plurality of couplings is antiferromagnetic. The superconducting qubits and the plurality of couplings are collectively capable of being in any one of at least two configurations. The at least two configurations include a first configuration characterized by an initialization Hamiltonian H0 and a second Hamiltonian characterized by a problem Hamiltonian HP. The problem Hamiltonian has a ground state. Each respective first superconducting qubit in the plurality of superconducting qubits is coupled with a respective second superconducting qubit in the plurality of superconducting qubits such that the first respective superconducting qubit and the corresponding second respective superconducting qubit define a predetermined coupling strength. The predetermined coupling strength between each of the first respective superconducting qubits and the corresponding second respective superconducting qubits collectively define a computational problem to be solved. The computation device further comprises a read out circuit coupled to at least one superconducting qubit in the plurality of superconducting qubits.
A tenth aspect of the invention comprises an apparatus comprising a plurality of superconducting charge qubits. Each respective first superconducting charge qubit in the plurality of superconducting charge qubits is coupled with a respective second superconducting charge qubit in the plurality of superconducting charge qubits such that the first respective superconducting charge qubit and the second respective superconducting charge qubit define a predetermined coupling strength. The predetermined coupling strength between each of the first respective superconducting charge qubits and each of the second respective superconducting charge qubits in the plurality of superconducting charge qubits collectively define a computational problem to be solved. Each superconducting charge qubit in the plurality of superconducting charge qubits is capable of being in one of at least two configurations. These at least two configurations include a first configuration in accordance with an initialization Hamiltonian H0 and a second configuration in accordance with a problem Hamiltonian HP. The apparatus further comprises an electrometer coupled to a superconducting charge qubit in the plurality of superconducting charge qubits.
In some embodiments in accordance with this tenth aspect of the invention, a superconducting charge qubit in the plurality of superconducting charge qubits comprises (i) a mesoscopic island made of superconducting material, (ii) superconducting reservoir, and (iii) a Josephson junction connecting the mesoscopic island to the superconducting reservoir. In some embodiments, the Josephson junction is a split Josephson junction. In some embodiments, the superconducting charge qubit further comprises a flux source configured to apply flux to the split Josephson junction.
In some embodiments in accordance with the tenth aspect of the invention, the apparatus further comprises a generator capacitively coupled to a superconducting charge qubit in the plurality of superconducting charge qubits by a capacitor. In some embodiments, the generator is configured to apply a plurality of electrostatic pulses to the superconducting charge qubit. The plurality of electrostatic pulses additionally define the computational problem.
In some embodiments in accordance with the tenth aspect of the invention, the apparatus further comprises a variable electrostatic transformer disposed between a first superconducting charge qubit and a second superconducting charge qubit in the plurality of superconducting charge qubits such that the predetermined coupling strength between the first superconducting charge qubit and the second superconducting charge qubit is tunable. In some embodiments, each respective first superconducting charge qubit in the plurality of superconducting charge qubits is arranged with respect to a respective second superconducting charge qubit in the plurality of superconducting charge qubits such that the plurality of superconducting charge qubits collectively form a non-planar graph.
An eleventh aspect of the invention provides a method for computing using a quantum system comprising a plurality of superconducting charge qubits. The quantum system is coupled to an electrometer and the quantum system is capable of being in any one of at least two configurations. The at least two configurations includes a first configuration characterized by an initialization Hamiltonian H0 and a second configuration characterized by a problem Hamiltonian HP. The problem Hamiltonian has a ground state. The plurality of superconducting charge qubits are arranged with respect to one another, with a predetermined number of couplings between respective pairs of superconducting charge qubits in the plurality of charge qubits, such that the plurality of superconducting charge qubits, coupled by the predetermined number of couplings, collectively define a computational problem to be solved. The method comprises initializing the quantum system to the initialization Hamiltonian HO. Then the quantum system is adiabatically changed until it is described by the ground state of the problem Hamiltonian HP. Next the quantum state of each superconducting charge qubit in the quantum system is read out through the electrometer, thereby solving the computational problem to be solved.
In some embodiments in accordance with the eleventh aspect of the invention, a first superconducting charge qubit in the plurality of superconducting charge qubits is coupled to a second superconducting charge qubit in the plurality of superconducting charge qubits by a capacitor such that the predetermined coupling strength between the first superconducting charge qubit and the second superconducting charge qubit is predetermined and is a function of the physical properties of the capacitor.
In some embodiments in accordance with the eleventh aspect of the invention, a first superconducting charge qubit in the plurality of superconducting charge qubits is coupled to a generator by a device configured to provide a tunable effective charging energy. The device comprises a capacitor and the method further comprises: tuning the value of the effective charging energy of the first superconducting charge qubit by varying the gate voltage on the capacitor of said device. In some embodiments, a superconducting charge qubit in the plurality of superconducting charge qubits comprises a split Josephson junction having a variable effective Josephson energy. In such embodiments, the method further comprises tuning the value of the effective Josephson energy of the superconducting charge qubit by varying a flux applied to the split Josephson junction. In some embodiments, the first configuration is reached by setting the effective Josephson energy of the superconducting charge qubit to a maximum value.
In some embodiments in accordance with the eleventh aspect of the invention, the adiabatically changing step comprises changing the configuration of the system from the first configuration characterized by the initialization Hamiltonian H0, to the second Hamiltonian characterized by a problem Hamiltonian HP in the presence of tunneling on a superconducting charge qubit in the plurality of superconducting charge qubits.
In some embodiments in accordance with a eleventh aspect of the invention, a first superconducting charge qubit in the plurality of superconducting charge qubits is characterized by (i) an effective Josephson energy that is tunable and (ii) an effective charging energy that is tunable. A minimum value of the effective Josephson energy is less than the effective charging energy of the first superconducting charge qubit A minimum value of the effective Josephson energy is less than a strength of a coupling between the first superconducting charge qubit and a second superconducting charge qubit in the plurality of superconducting charge qubits. The effective charging energy is, at most, equal to a maximum value of the effective Josephson energy of the first superconducting charge qubit. Furthermore, a strength of a coupling between the first superconducting charge qubit and a second superconducting charge qubit in the plurality of superconducting charge qubits is, at most, equal to a maximum value of the effective Josephson energy of the first superconducting charge qubit.
In still another embodiment in accordance with the eleventh aspect of the invention, a first superconducting charge qubit in the plurality of superconducting charge qubits is characterized by (i) an effective Josephson energy that is tunable and (ii) an effective charging energy that is tunable. In such embodiments, the adiabatically changing step comprises adiabatically tuning the effective Josephson energy of the first superconducting charge qubit such that the effective Josephson energy of the first superconducting charge qubit reaches a minimum value when the quantum system is described by the ground state of the problem Hamiltonian HP.
In some embodiments in accordance with the eleventh aspect of the invention, a first superconducting charge qubit in the plurality of superconducting charge qubits has a first basis state and a second basis state and, when the quantum system is described by the ground state of the problem Hamiltonian HP, tunneling between the first basis state and the second basis state of the first superconducting charge qubit does not occur.
In some embodiments in accordance with the eleventh aspect of the invention, a first superconducting charge qubit in the plurality of superconducting charge qubits has a first basis state and a second basis state and, when the quantum system is described by the ground state of the problem Hamiltonian HP, the tunneling between the first basis state and the second basis state of the first superconducting charge qubit does occur. Furthermore, the reading out step comprises probing an observable of the sigma-x Pauli matrix σX.
In some embodiments in accordance with the eleventh aspect of the invention, a first superconducting charge qubit in the plurality of superconducting charge qubits is characterized by (i) an effective Josephson energy that is tunable and (ii) an effective charging energy that is tunable. In such embodiments, a minimum value of the effective Josephson energy is less than the effective charging energy of the first superconducting charge qubit; a minimum value of effective Josephson energy is less than a strength of a coupling between the first superconducting charge qubit and a second superconducting charge qubit in the plurality of superconducting charge qubits; the effective charging energy is greater than a maximum value of the effective Josephson energy of the first superconducting charge qubit; and a strength of a coupling between the first superconducting charge qubit and a second superconducting charge qubit in the plurality of superconducting charge qubits is, at most, equal to the maximum effective Josephson energy of the first superconducting charge qubit. In some such embodiments, the initializing step comprises setting the effective charging energy of the first superconducting charge qubit to a minimum value. In some such embodiments, the adiabatically changing step comprises adiabatically tuning the effective Josephson energy of the first superconducting charge qubit such that the effective Josephson energy is at a minimum value when the quantum system is described by the ground state of the problem Hamiltonian HP, and adiabatically increasing the effective charging energy of the first superconducting charge qubit.
In some embodiments in accordance with the eleventh aspect of the invention, a first superconducting charge qubit in the plurality of superconducting charge qubits is characterized by an effective Josephson energy that is tunable. The initializing step comprises setting the effective Josephson energy of the first superconducting charge qubit to a minimum value, and the adiabatically changing step comprises (i) adiabatically tuning the effective Josephson energy of the first superconducting charge qubit such that the effective Josephson energy is greater than a minimum value for a period of time before the quantum system is described by the ground state of the problem Hamiltonian HP, and adiabatically tuning the effective Josephson energy of the first superconducting charge qubit such that the effective Josephson energy is at a minimum value when the quantum system is described by the ground state of the problem Hamiltonian HP.
Like reference numerals refer to corresponding parts throughout the several views of the drawings.
The present invention comprises systems and methods for adiabatic quantum computing using superconducting qubits. In various embodiments of the present invention, adiabatic quantum computing is performed on registers of superconducting qubits that have demonstrated quantum computing functionality. Adiabatic quantum computing is a model of quantum computing that can be used to attempt to find solutions for computationally difficult problems.
When choosing a candidate system for adiabatic quantum computing there are a few criteria that can be observed. These criteria can be drawn from those described herein below. However, some embodiments of the present invention may not adhere to all of these criteria. One criterion is that the readout device should a Stern-Gerlach σZ type observation. A second criterion is that the tunneling term in the problem Hamiltonian should be about zero. For HP=ΔσX+εσZ then Δ≈0. A third criterion is that the magnitude of the tunneling term in the problem, initial, or extra Hamiltonian (HP, H0, HE) should be tunable. A fourth criterion is that the qubit-qubit coupling should be diagonal in the basis of final qubit states, i.e., σZ{circle over (x)}σZ. Because an Ising model with ferromagnetic couplings has a trivial ground state, all spins aligned, a fifth criterion is that the system have some antiferromagnetic coupling between qubits. Some AFM couplings include the case where all are antiferromagnetic. Also, ferromagnetic couplings have a negative sign −JσZ{circle over (x)}σZ, and antiferromagnetic couplings have a positive sign JσZ{circle over (x)}σZ.
Some embodiments of the present invention adhere to the above criteria. Other embodiments of the present invention do not. For instance, in the case of the phase qubit, it is possible to have the tunneling term in the problem Hamiltonian be, not zero, but weak, e.g., for HP=ΔσX+εσZ then Δ<<ε. In such a case it is possible for the readout device to probe a Stern-Gerlach σX type observable. Other embodiments of superconducting adiabatic quantum computers of the present invention do not adhere to the third criterion described above. For example, the magnitude of the tunneling term in the problem, initial, or extra Hamiltonian (HP, H0, HE) is fixed but the contribution of the problem, initial, or extra Hamiltonian to the instant Hamiltonian is tunable in such embodiments. Specific embodiments of the present invention are described below.
In accordance with embodiments of the present invention, the general procedure of adiabatic quantum computing is shown in
In step 403, the quantum system is initialized to the ground state of the time-independent Hamiltonian, H0, which initially describes the quantum system. It is assumed that the ground state of H0 is a state to which the system can be reliably and reproducibly set. As will be disclosed in further detail below, this assumption is reasonable for quantum systems comprising, at a minimum, specific types of qubits and specific types of arrangements of such qubits.
In transition 404 between steps 403 and 405, the quantum system is acted upon in an adiabatic manner in order to alter the system. The system changes from being described by Hamiltonian H0 to a description under HP. This change is adiabatic, as defined above, and occurs in a period of time T. In other words, the operator of an adiabatic quantum computer causes the system, and Hamiltonian H describing the system, to change from H0 to a final form HP in time T. The change is an interpolation between H0 and HP. The change can be a linear interpolation:
H(t)=(1−γ(t))H0+γ(t)HP
where the adiabatic evolution parameter, γ(t), is a continuous function with γ(t=0)=0, and γ(t=T)=1. The change can be a linear interpolation, γ(t)=t/T such that
In accordance with the adiabatic theorem of quantum mechanics, a system will remain in the ground state of H at every instance the system is changed and after the change is complete, provided the change is adiabatic. In some embodiments of the present invention, the quantum system starts in an initial state H0 that does not permit quantum tunneling, is perturbed in an adiabatic manner to an intermediate state that permits quantum tunneling, and then is perturbed in an adiabatic manner to the final state described above.
In step 405, the quantum system has been altered to one that is described by the final Hamiltonian. The final Hamiltonian HP can encode the constraints of a computational problem such that the ground state of HP corresponds to a solution to this problem. Hence, the final Hamiltonian is also called the problem Hamiltonian HP. If the system is not in the ground state of HP, the state is an approximate solution to the computational problem. Approximate solutions to many computational problems are useful and such embodiments are fully within the scope of the present invention.
In step 407, the system described by the final Hamiltonian HP is read out. The read out can be in the σZ basis of the qubits. If the read out basis commutes with the terms of the problem Hamiltonian Hp, then performing a read out operation does not disturb the ground state of the system. The read out method can take many forms. The object of the read out step is to determine exactly or approximately the ground state of the system. The states of all qubits are represented by the vector {right arrow over (O)}, which gives a concise image of the ground state or approximate ground state of the system. The read out method can compare the energies of various states of the system. More examples are given below, making use of specific qubits for better description.
In one embodiment of the present invention, the natural quantum mechanical evolution of the quantum system under the slowly changing Hamiltonian H(t) carries the initial state H0 of the quantum system into a final state, the ground state of HP, corresponding to the solution of the problem defined by the quantum system. A measurement of the final state of the quantum system reveals the solution to the computational problem encoded in the problem Hamiltonian. In such embodiments, an aspect that can define the success of the process is how quickly (or slowly) the change between the initial Hamiltonian and problem Hamiltonian occurs. How quickly one can drive the interpolation between H0 and HP, while keeping the system in the ground state of the instantaneous Hamiltonians that the quantum system traverses through in route to HP, is a determination that can be made using the adiabatic theorem of quantum mechanics. This section provides a detailed explanation of the adiabatic theorem of quantum mechanics. More particularly, this section describes how quantum mechanics imposes constraints on which quantum systems can be used in accordance with the present invention and how such quantum systems can be used to solve computational problems using the methods of the present invention.
A quantum system evolves under the Schrödinger equation:
where Θ(t) and H(t) are respectively the time ordering operator and Hamiltonian of the system. Adiabatic evolution is a special case where H(t) is a slowly varying function. The time dependent basis states and energy levels of the Hamiltonian are:
H(t)|l;t>=El(t)|l;t>
where lε[0, N−1], and N is the dimension of the Hilbert space for the quantum system described by the Schrödinger equation. The energy levels, energies, or energy spectra of the quantum system E1(t) are a set of energies that the system can occupy. The energies of the states are a strictly increasing set.
A general example of the adiabatic evolution of a quantum system is as follows. The states |0;t> and |1;t> are respectively the ground and first excited states of Hamiltonian H(t), with energies E0(t) and E1(t). Gap g(t) is the difference between energies of the ground and first excited states as follows:
g(t)=E1(t)−E0(t).
If the quantum system is initialized in the ground state and evolved under H(t), where H(t) is slowly varying, and if the gap is greater than zero, then for 0≦t≦T the quantum system will remain in the ground state. In other words:
|<E0;T|Ψ(T)>|2≧1−ε2.
Without intending to be limited to any particular theory, it is believed that the existence of the gap means that the quantum system under the Schrödinger equation remains in the ground state with high fidelity, e.g., 1−ε2(ε<<±1). The fidelity of the operation can be found quantitatively.
The minimum energy gap between the ground state E0 and first excited state E1 of the instantaneous Hamiltonian is given by gmin where:
Also relevant is the matrix element:
The adiabatic theorem asserts fidelity of the quantum system will be close to unity provided that:
If this criterion is met, the quantum system will remain in the ground state.
In an embodiment of the present invention, T is the time taken to vary a control parameter of a charge qubit, for example induced gate charge or flux for a charge qubit with split Josephson junction. See Section 5.4. In an embodiment of the present invention, time T is a value between about 0.1 nanosecond and about 500 microseconds. In other words, the amount of time between when the quantum system is allowed to begin adiabatically changing from the initial state H0 to when the quantum system first reaches the final state HP is between about 0.1 nanosecond and about 500 microseconds. In another embodiment of the present invention, time T is a value between about 10 nanosecond and about 100 microsecond. In an embodiment of the present invention, time T is a value less than the inverse of the characteristic frequency of the physical system comprising superconducting qubits. The characteristic frequency of a qubit is the energy difference between a ground state and an excited state of a qubit, converted to energy units. Thus, the characteristic frequency of a physical system comprising qubits is the characteristic frequency of one or more qubits within the physical system.
In an embodiment of the present invention, T is the time taken to vary a control parameter of a phase qubit, for example flux in a persistent current qubit. See Section 5.3. In an embodiment of the present invention, time T is a value between about 1 nanosecond and about 100 microseconds. In other words, the amount of time between when the quantum system is allowed to begin adiabatically changing from the initial state H0 to when the quantum system first reaches the final state HP is between about 1 nanosecond and about 100 microseconds. In an embodiment of the present invention, time T is a value between about 4 nanoseconds and about 10 microseconds. In some embodiments, the time T is calculated as the time at which a Landau-Zener transition is likely not to occur. For more information on Landau-Zener transitions see, for example, Garanin et al., 2002, Phys. Rev. B 66, 174438, which is hereby incorporated by reference in its entirety.
Most analyses of adiabatic algorithms emphasize how the gap, g(t), and the ratio of the matrix element <dH/dt>1,0 to the square of the minimum of the gap, scale with increasing problem size. It is believed that, by examining these metrics, the validity of adiabatic algorithms, and other adiabatic processes, can be determined. Some embodiments of the present invention make use of an alternative analysis that looks at the probability of transition 404, or some other process, being diabatic (i.e. a process that involves heat transfer as opposed to an adiabatic process that involves no heat transfer). Rather than calculating the minimum gap, which is the difference between the energy of the ground and the first excited state of the quantum system 850 that models the problem to be solved, this additional analysis calculates the probability of a transition out of the ground state by the quantum system. In examples of the present invention, this probability calculation can be a more relevant metric for assessing the failure rate of the adiabatic algorithm or process. To perform the computation, it is assumed that any diabatic transition (any transition characterized by the transfer of heat) is a Landau-Zener transition, e.g., a transition confined to adjacent levels at anticrossings. A description of anticrossing levels is provided below in conjunction with
The values for the parameters gmin, and Δm will vary with the specific instance of the algorithm being run on the adiabatic quantum computer.
Other embodiments of the present invention can be constructed and operated with a different estimate for the probability of a diabatic transition at step 404. For instance, in some embodiments the second estimate of the Landau-Zener transition probability is computed. This probability has the form:
where k is a constant that is about 1, EJ is the Josephson energy of the qubit (or maximum Josephson energy of the Josephson energies of a plurality of qubits), and fP is frequency of oscillation of an additional flux that is added to the superconducting qubit. The values for the parameters gmin, and EJ will vary with the specific instance of the algorithm being run on the adiabatic quantum computer.
In many embodiments of the present invention, quantum systems for adiabatic quantum computation are designed such that the minimum of the energy gap, the difference in the asymptotic slopes, and the rate of change of the adiabatic evolution parameter ensure that the probability of diabatic transition at step 404 is small, e.g. PLZ is much smaller than 1. In an embodiment of the present invention, PLZ is 1×10−4 or less. In another embodiment of the present invention, PLZ is 1×10−3 or less. The probability of transition from the ground state of the quantum system to the first excited state of the quantum system is exponentially sensitive to the rate of change of the adiabatic evolution and provides a lower bound to that rate. The probability of transition from the ground state of the quantum system to the first excited state of the quantum system also provides an upper limit on rate of change of the adiabatic evolution parameter. The duration of an adiabatic algorithm, or process, should be less than the time it takes for a Landau-Zener transition to occur. If PLZ is the probability per anticrossing, then the quantum system (e.g., quantum system 850 of
In some embodiments of the present invention, the amount of time required to perform the readout process should be engineered so that the probability that quantum system 850 will transition from the ground state to the first excited state is small (e.g., a probability less than one percent). For a readout process, the following should hold: PLZ×m×r<<1, or τ<<fA×(PLZ×m)−1, where m is the number of qubits, and r is the average number of cycles used to readout the qubits, fA is the frequency of the cycles used to readout the qubits, and τ is the time for the readout of one qubit in a plurality of m qubits.
In an embodiment of the present invention this requisite small probability for transition PLZ is dependent on the process performed. In the case of a readout process that applies additional flux to a superconducting qubit and measures the state through a tank circuit, such as the embodiment described in detail below in conjunction with
Section 5.3 describes quantum computing systems of the present invention that make use of phase qubits. Section 5.4, below, describes quantum computing systems of the present invention that make use of charge qubits.
Referring to
The two stable states of a qubit 101 will have equal energy, meaning that they will be degenerate, and will therefore support quantum tunneling between the two equal energy states (basis states) when the amount of flux trapped in the qubit is 0.5Φ0. The amount of flux required to trap 0.5Φ0 in a qubit 101 is directly proportional to the area of the qubit. Here, the area of a qubit is defined as the area enclosed by the superconducting loop of the qubit. If the amount of flux needed to achieve a trapped flux of 0.5Φ0 in a first qubit 101 having area A1 is B1, then the amount of flux that is needed to trap 0.5Φ0 of flux in a second qubit having area A2 is (A2/A1)B1. Advantageously, in system 500, each qubit 101 has the same total surface area so that an external mechanism (e.g., a tank circuit) can cause each respective qubit 101 in system 500 to trap 0.5Φ0 of flux at approximately or exactly the same time.
In preferred embodiments, the three persistent current qubits, 101-1, 101-2, and 101-3 in structure 500 are inductively coupled to a tank circuit (not fully shown in
In some embodiments of the present invention, qubit parameters are chosen to satisfy the requirements of the problem to be solved by adiabatic quantum computation and the restrictions of the qubits. For instance, in the case of
In one particular embodiment of a qubit 101 in quantum system 500 (
Various embodiments of the present invention provide different values for the persistent current that circulates in persistent current qubits 101. These persistent currents range from about 100 nanoamperes to about 2 milliamperes. The persistent current values change the slope of the asymptotes at anticrossing 915 (
Step 401 (Preparation). An overview of an apparatus 500 used to solve a computational problem in accordance of the present invention has been detailed in Section 5.3.1.1. In this section, the general adiabatic quantum computing process set forth in
In accordance with the present invention, the problem of determining the ground state of the three node frustrated ring is encoded into system 500 by customizing the coupling constants between neighboring qubits 101 using two variables: (i) the distance between the qubits and (ii) the amount of surface area common to such qubits. The lengths and widths of qubits 101, as well as the spatial separation between such qubits, is adjusted to customize inter-qubit inductive coupling strengths in such a way that these coupling strengths correspond to a computational problem to be solved (e.g., the ground state of a three member frustrated ring). In preferred embodiments, qubit 101 length and width choices are subject to the constraint that each qubit 101 have the same or approximately the same total surface area so that the qubits can be adjusted to a state where they each trap half a flux quantum at the same time.
As shown in
All three qubits 101-1, 101-2, and 101-3, have the same area subject to manufacturing tolerances. In some embodiments of the present invention, such tolerance allows for a ±25% deviation from the mean qubit surface area, in other embodiments such tolerance is ±15%, ±5%, ±2%, or less than ±1% deviation from the mean qubit surface area. Qubits 101-1, 101-2, and 101-3 are coupled to each other asymmetrically. In other words, the total surface area common to qubit 101-3 and 101-1 (or 101-2) is less than the total surface area common to qubit 101-1 and 101-2. Embodiments of the present invention, such as those that include a system like 500, have a Hamiltonian given by:
where N is the number of qubits. The quantity Δi is the tunneling rate, or energy, of the ith qubit expressed in units of frequency, or energy. These unit scales differ by a factor of h or Plank's constant. The quantity εi is the amount of bias applied to the qubit, or equivalently, the amount of flux in the loops of the qubits. The quantity Jij is the strength of the coupling between the ith and the jth qubit. The coupling energy is a function of the mutual inductance between qubits Jij=MijIiIj. In order to encode the three node frustrated ring problem, the three coupling energies are arranged such that J12>>J13≈J23.
In some embodiments of the present invention, the coupling strength between any two abutting qubits is a product of the mutual inductance and the currents in the coupled qubits. The mutual inductance is a function of common surface area and distance. The greater the common surface area the stronger the mutual inductance. The greater the distance the less the mutual inductance. The current in each qubit is a function of the Josephson energy of the qubit, EJ. EJ depends on the type of Josephson junctions interrupting the superconducting loop within the qubit. Inductance calculations can be performed using programs such as FASTHENRY, a numeric three-dimensional inductance calculation program, is freely distributed by the Research Laboratory of Electronics of the Massachusetts Institute of Technology, Cambridge, Mass. See, Kamon, et al., 1994, “FASTHENRY: A Multipole-Accelerated 3-D Inductance Extraction Program,” IEEE Trans. on Microwave Theory and Techniques, 42, pp. 1750-1758, which is hereby incorporated by reference in its entirety. Alternatively, mutual inductance can be calculated analytically using techniques known in the art. See Grover, Inductance Calculations: Working Formulas and Tables, Dover Publications, Inc., New York, 1946, which is hereby incorporated by reference in its entirety.
Step 403 (Initialization of system 500 to HO). Qubits 101 of system 500 of
where Q represents the strength of the external magnetic field. In some embodiments, this external magnetic field has a strength such that ε>>Δ or ε>>J. That is, the magnetic field is on an energy scale that is large relative to other terms in the Hamiltonian. As an example, for a qubit having the dimensions of 50 to 100 micrometers squared, the magnetic field is between about 10−8 teslas and about 10−6 teslas. The energy of a magnetic field B in a persistent current qubit of area S, in terms of MKS units is
where μ0 is 4π×10−7 Wb/(A·m) (webers per ampere meter). The magnetic field also controls the bias ε applied to each qubit. In step 403, the large external magnetic field is used to initialize system 500 to the starting state characterized by the Hamiltonian H0. System 500 is in the ground state |000> of H0 when the flux that is trapped in each qubit 101 is aligned with the external magnetic field.
Once system 500 is in the ground state |000> of the starting state H0, it can be used to solve the computational problem hard-coded into the system through the engineered inter-qubit coupling constants. To accomplish this, the applied magnetic field is removed at a rate that is sufficiently slow to cause the system to change adiabatically. At any given instant during the removal of the externally applied initialization magnetic field, the state of system 500 is described by the instantaneous Hamiltonian:
where Δ is a tunneling term that is a function of the bias ε applied by coil 502. As described above, it is preferred that the bias ε applied by coil 502 be such that a half flux quantum (0.5Φ0) be trapped in each qubit 101 so that the two stable states of each qubit have the same energy (are degenerate) to achieve optimal quantum tunneling. Furthermore, it is preferred that each qubit have the same total surface area so that quantum tunneling in each qubit starts and stops at the same time. In less preferred embodiments, qubits 101 are biased by coil 502 such that the flux trapped in each qubit is an odd multiple of 0.5Φ0 (e.g., 1.5, 2.5, 3.5, etc., or in other words, N·0.5Φ0, where N is 1, 2, 3, . . . ) since the bistable states of the qubits 101 will have equal energy (will be degenerate) in these situations as well. However, such higher amounts of trapped flux can have undesirable side effects on the properties of the superconducting current in the qubits. For instance, large amounts of trapped flux in qubits 101 may quench the superconducting current altogether that flows through the superconducting loop of each said loop. In some embodiments, the qubits are biased by coil 502 such that the amount of flux trapped in each qubit is ζ·N·0.5Φ0 where ζ is between 0.7 and 1.3, between 0.8 and 1.2, or between 0.9 and 1.1 and where N is a positive integer. However, qubits 101 tunnel best when the bias in each corresponds to half-flux quantum of flux.
Step 405 (reaching the problem state HP). As the external perpendicularly applied field is adiabatically turned off, the problem Hamiltonian HP is arrived at:
Note that, unlike the instantaneous Hamiltonian, the problem Hamiltonian does not include the tunneling term Δ. Thus, when system 500 reaches the final problem state, the quantum states of each respective qubit 101 can no longer tunnel. Typically, the final state is one in which no external magnetic field is applied to qubits 101 and, consequently, qubits no longer trap flux.
Step 407 (Measurement). In step 407, the quantum system is measured. In the problem addressed in the present system, there eight possible solutions {000, 001, 010, 100, 011, 110, 101, and 111}. Measurement involves determining which of the eight solutions was adopted by system 500. An advantage of the present invention is that this solution will represent the actual solution of the quantum system.
In an embodiment of the present invention, the result of the adiabatic quantum computation is determined using individual qubit magnetometers. Referring to
In an embodiment of the present invention, the result of the adiabatic quantum computation is determined using individual qubit magnetometers laid adjacent to respective qubits in the quantum system. Referring to
In an embodiment of the present invention, the qubits 501 used in the adiabatic quantum computation are coupled by Josephson junctions. Referring to
In an embodiment of the present invention, qubits 501 used in the adiabatic quantum computation have tunable tunneling energies. Referring to
where ΦX is the external flux applied to the split Josephson junction, and EJ0 is the Josephson energy of one of the Josephson junctions in the split junction. When the magnetic flux through split junction 528 is one half a flux quantum the tunneling energy for the corresponding qubit 501 is zero. The magnetic flux in the split Josephson junctions 528 may be applied by a global magnetic field.
In an embodiment of the present invention the tunneling of qubits 501 is suppressed by applying a flux of one half a flux quantum to the split Josephson junction 528 of one or more qubits 501. In an embodiment of the present invention the split junction loop is orientated perpendicular to the plane of the corresponding (adjacent) qubit 501 such that flux applied to it is transverse in the field of the qubits.
Referring to
Referring to
In other embodiments of the present invention, the state of each qubit 101 of system 500 is not individually read out. Rather, such embodiments make use of the profile of the energy level diagram of system 500 as a whole over a range of probing biasing currents. In such embodiments, when system 500 reaches HP the system is probed with a range of biasing currents using, for example, a tank circuit that includes coil 502 (
As described above, for many embodiments of the present invention, the energy levels of the system for adiabatic quantum computing HP across a range of probing biasing values are differentiable from each other.
Considering
In an embodiment of the present invention, the system is read out by differentiating various energy levels (solutions) to identify the ground state. In an example of the persistent current qubit, the dephasing rate is currently recorded as being 2.5 microseconds or less. Once the system is in a state of HP the state, e.g. 551 or 550 can be determined by locally probing the energy level structure by low frequency applications of a biasing magnetic field.
The states of energy level diagram like that of
Knowledge of the initial energy level and the corresponding voltage response can allow the ground state to be determined provided that a sampling of some of the lowest energy levels voltage response has been made. Accordingly, in an embodiment of the present invention, the system is not initialized in the ground state of the initial Hamiltonian. Rather the system is initialized in an excited state of the system such as 551. Then the interpolation between initial and final Hamiltonians occurs at a rate that is adiabatic but is faster than the relaxation rate out of state 551. In one example of a persistent current qubit, the relaxation rate is about 1 to about 10 microseconds. The ground state with have a greater curvature and the lowest number inflection points of the energy levels of the system.
In operation, when an external signal, such as a magnetic flux in an inductively coupled device, is applied through tank circuit 810, a resonance is induced. This resonance affects the magnitude of impedance 804. In particular, impedance 804 is affected by the direction of the magnetic flux in devices inductively coupled to it, such as a superconducting qubit (e.g., quantum system 850). Because of the state dependency of impedance 804, the tank voltage or current response is likewise state dependent. Therefore, the tank voltage or current response of tank circuit 810 can be used to measure the direction of the magnetic flux in devices inductively coupled to the tank circuit. In an embodiment of the present invention, tank circuit 810 is inductively coupled to a superconducting qubit (an example of quantum system 850) having quantum states such that the resonant frequency of the tank circuit is correlated with the state of the superconducting qubit. In operation, characteristics of the superconducting qubit can be probed by tank circuit 810 by applying signals through tank circuit 810 and measuring the response of the tank circuit. The characteristics of tank circuit 810 can be observed by amplifier 809, for example. See, for example, U.S. Patent Publication 2003/0224944 A1, entitled “Characterization and measurement of superconducting structures,” which is hereby incorporated by reference in its entirety.
Some embodiments of the present invention make use of the high electron mobility field-effect transistors (HEMT) and their improved pseudomorphic variants (PHEMT) to measure the signal from tank circuit 810. Such embodiments can make use of multistage amplifiers 809, such as two, three, or four transistors. Some embodiments of the present invention use the commercially available transistor ATF-35143 from Agilent Technologies, Inc. (Palo Alto, Calif.) as amplifier 809. In some embodiments, amplifier 809 comprises a three transistor amplifier having a noise temperature of about 100 millikelvin. The power consumption of the transistors is decreased by reducing both the transistors' drain voltage and the drain current to about two percent of the average ratings for the drain source voltage, and about 0.3% of the current rating. This reduces the power dissipation of the first transistor to about 20 milliwatts. In some embodiments, all active resistances in the amplifier's signal channel are replaced by inductors and third stage is used to match the amplifier output with cable impedance. In some embodiments, amplifier 809 is assembled on a printed board. In a specific example, amplifier 809 is a cold amplifier, built as described above, and is thermally connected to a helium-3 pot of a dilution refrigerator that houses quantum system 850.
In one embodiment of the present invention, the response voltage of the tank circuit is amplified by a cold amplifier thermally coupled to the helium-3 pot of the dilution refrigerator. The response signal is once more amplified by a room temperature amplifier and detected by an rf lock-in voltmeter. The rf lock-in voltmeter can be used to average the response of the tank circuit over many cycles of the response of the tank. For more information on such an amplifier arrangement, see Oukhanski et al., 2003, Rev. Sci. Instr. 72, 1145, which is hereby incorporated by reference in its entirety.
Some embodiments of the present invention make use of tank circuit 810 to read out the state of quantum system 850. By making use of a radio-frequency technique, called the impedance measurement technique (IMT), the main parameters of the circuit described by quantum system 850 can be reconstructed. With the IMT, a qubit in quantum system 850 to be read out is coupled through a mutual inductance M to a tank circuit 810 (
Tank circuit 810 is driven by a direct bias current IDC and an alternating current IRF of a frequency ω close to the resonance frequency of the tank circuit ω0. The total externally applied magnetic flux to the qubit, ΦE, is therefore determined by the tank circuit where ΦE=ΦDC+ΦRF. Here, ΦDC=ΦA+f(t)Φ0.
In one embodiment of the present invention, the direct bias current IDC and an alternating current IRF in the tank circuit are supplied along filtered lines by commercially available current sources. An example of a DC-current source for supplying the direct bias current IDC is the Agilent 33220A. An example of an AC-current source for supplying the alternating current IRF is the Agilent E8247C microwave generator. Both devices are available from Agilent Technologies, Inc. (Palo Alto, Calif.).
The terms ΦA and f(t)Φ0 are constant or are slowly varying and are generated by the direct bias current IDC. The magnitude of ΦA is discussed below, f(t) has a value between 0 and 1 inclusive, and Φ0 is one flux quantum (2.07×10−15 webers (or volt seconds)). Further, ΦRF is a small amplitude oscillating rf flux generated with ac current IRF, having a magnitude of about 10−5Φ0 to about 10−1Φ0. In an embodiment of the present invention, the ΦRF amplitude is about 10−3Φ0. In an embodiment of the present invention, the term “about” in relation to small magnetic fluxes, such as ΦRF, means a variance of ±10%, ±50%, ±100%, and ±500% of the magnetic flux.
If the amplitude of ΦRF is small, meaning that it has a negligible value relative to Φ0(ΦRF<<Φ0), then the approximate equality ΦE≅ΦDC is valid. In an embodiment of the present invention f(t), ΦA, and ΦRF are varied on progressively shorter time scales because f(t) and ΦA are the qubit bias, and ΦRF is a small amplitude fast function that is used to determine the state of quantum system 850 (e.g., the state of a qubit in quantum system 850). By monitoring the effective impedance of tank circuit 810 (
The imaginary part of the total impedance of tank circuit 810, expressed through the phase angle χ between driving current IRF and tank voltage, can be expressed as
at ω=ω0. Here, F′(Φ(ΦDC))=I′(Φ(ΦDC))/Ic is the derivative of the normalized supercurrent in the qubit with respect to the total magnetic flux, Φ=ΦE−β×F(Φ(ΦDC)) in the superconducting qubit. The parameter β=2πLIc/Φ0 is the normalized inductance of the superconducting qubit, having inductance L, and critical current IC.
This section describes techniques for reading out the state of a quantum system 850. In some embodiments, this is accomplished by reading out the state of each qubit within quantum system 850 on an individual basis. Individualized readout of qubits in a quantum system 850 can be accomplished, for example, by making use of individual bias wires or individual excitation devices 820 for each qubit within quantum system 850 as described in this section and as illustrated in
In accordance with an embodiment of the present invention, a method for reading out the state of a superconducting qubit within quantum system 850 involves using tank circuit 810 to apply a range of fluxes to the superconducting qubit over a period of time and detecting a change in the properties of the tank circuit coupled to the superconducting circuit during the sweep. In the case where the qubit is a superconducting flux qubit, a superconducting loop with low inductance L interrupted by three Josephson junctions, the flux applied during the sweep can range from, for example, 0.49Φ0 to 0.51Φ0. In other embodiments, the flux applied during the sweep can range between 0.3Φ0 and 0.7Φ0. In still other embodiments, the flux applied during the sweep can range across several multiples of Φ0 (e.g., between 0.3Φ0 and 5Φ0 or more). In preferred embodiments of the present invention, the flux sweep is performed adiabatically to ensure that the qubit remains in its ground state during the transition. Under this adiabatic sweep, tank circuit 810 (
The readout of a superconducting qubit involves sweeping the flux applied to the superconducting qubit by an amount and in a direction designed to detect the anticrossing 915. This method makes use of the fact that a tank circuit can detect the anticrossing 915 by the change in the curvature of the ground state energy level 909 at anticrossing 915. In other words, the inventive method measures the magnetic susceptibility of the qubit. See, for example, U.S. Patent Publication 2003/0224944 A1, entitled “Characterization and measurement of superconducting structures,” which is hereby incorporated by reference in its entirety. In some embodiments, the measurable quantity is a dip in the voltage across tank circuit 810, or a change in the phase of tank circuit 810, or both. If a dip in the voltage results from the readout operation then the state of the qubit within quantum system 850 transitioned through the anticrossing. If no dip is observed, the qubit did not transition through the anticrossing during measurement sweep 407. These state-dependent observables are used to perform the readout operation.
In some embodiments, when the superconducting qubit is in the |0> ground state, e.g. in region 910, and the flux applied to the superconducting qubit is increased, then the state of the qubit will transition through anticrossing 915, and a dip in the voltage across the tank circuit is observed. Similarly, when the superconducting qubit is in the |1> ground state, e.g. in region 911, and the flux applied to the superconducting qubit is decreased, then the state of the qubit will transition through anticrossing 915, and a dip in the tank circuit voltage 810 is observed. Alternatively, when the superconducting qubit is in the |0> ground state, e.g. in region 910, and the flux applied to the superconducting qubit is decreased, the quantum state of the qubit will not transition through anticrossing 915, and no dip in the tank circuit voltage 810 is observed. Likewise, if the superconducting qubit is in the |1> ground state, e.g. in region 911, and the flux applied to the superconducting qubit is increased, the state of the qubit will not transition through the anticrossing 915, and no dip in the tank circuit voltage is observed.
At termini 952 and 953, the states corresponding to energy levels disappear and the state of the qubit falls to the first available energy level. As illustrated in
The readout 407 (
Referring to
Now consider the case in which the superconducting qubit is in the |0> state, e.g. at energy level 950, and the flux applied to the superconducting qubit is decreased. In this case, the qubit will remain in the |0> state until the flux is decreased to just before the point corresponding to terminus 952, at which point a wide dip will occur in the voltage of tank circuit 810 due to slight curvature of level 950 in the vicinity of terminus 952. After the flux is decreased past terminus 952, the state transitions from |0> to |1> because state |0> no longer exists. Consequently, there is an abrupt rise in voltage across tank circuit 810.
Further consider the case in which a qubit within quantum system 850 of
In some embodiments, there is an additional magnetic field BA applied to the qubit within quantum system 850 of
The objective of the qubit measurements described in the preceding paragraphs is not to establish the state (ground or excited) that a qubit in quantum system 850 is in, but rather what is the direction of the qubit's circulating current (e.g. clockwise 102-0 or counterclockwise 101-2 as defined in
HQ=½ΔσX−½εσZ
where Δ and ε are respectively the tunneling and bias energies. The states |0> and |1> are assigned to direction of current circulation in the qubit, for example in the manner illustrated in
The addition of an external magnetic field that is applied against the superconducting qubit induces a third additional flux ΦA in the qubit. The total flux applied to the qubit or bias is described as:
ΦE=ΦDC+ΦRF,
where ΦDC=ΦA+f(t)Φ0.
In some embodiments, of the present invention, the total flux in the qubit is varied in a triangular pattern, with the magnitude and sign of the flux chosen such that the states of the qubit can be differentiated and the distance to anticrossings or crossings can be found. Details on how such waveforms can accomplish this task are described in Section 5.3.2.2, below. In an embodiment of the present invention, the flux is applied to a superconducting qubit through individual bias wires. In such an embodiment each qubit, or all qubits except for one qubit, in quantum system 850 has a bias wire.
There are independent parameters that can be altered for each waveform. These parameters include, for example, waveform shape, direction, period, amplitude, amplitude growth function, and equilibrium point. The waveforms can oscillate in both directions about an equilibrium point with a mean of zero (bidirectional), or oscillate away from the equilibrium point in one direction with a non-zero mean (unidirectional). The amplitude for the additional flux ΦA can be constant, or grow with time. The growth can be continuous or punctuated. The equilibrium point can be a fixed value of flux or vary with time. The shape of the waveform is an oscillatory function that can be selected from a variety of shapes including, for example, sinusoidal, triangular, and trapezoidal, and the low harmonic Fourier approximations thereof. Exemplary waveforms that can be used in accordance with the present invention are illustrated in FIGS. 10A-D. The waveforms, in FIGS. 10A-D are plotted as the additional flux ΦA against time, in arbitrary units.
In an embodiment of the present invention, the maximum amplitude of the additional flux ΦA in a superconducting qubit is between about 0.01 Φ0 and about 1 Φ0. In another embodiment of the present invention, the maximum amplitude of the additional flux in a superconducting qubit is between about 0.1 Φ0 and about 0.5 Φ0. In another embodiment of the present invention, the maximum amplitude of the additional flux in a superconducting qubit varies but has a mean of about 0.25 Φ0 and can deviate from this value by as much as 0.125 Φ0. For a superconducting qubit such as qubit 101 of
The frequency of oscillation, fA, is such that the change in flux ΦDC in quantum system 850 remains adiabatic. In embodiments of the present invention, the period ranges from one cycle per second to about 100 kilocycles per second. Failure of the process to remain adiabatic results in a Landau-Zener transition. For more information on Landau-Zener transitions see, for example, Garanin et al., 2002, Phys. Rev. B 66, 174438, which is hereby incorporated by reference in its entirety.
In some embodiments, rather than sweeping through a range of applied fluxes in accordance with a single waveform (oscillation, cycle), the waveform is repeated a plurality of time (plurality of oscillations, plurality of cycles) and voltage response across each of these oscillations (cycles) is averaged. The number of oscillations needed to perform readout, in such embodiments, depends on the architecture of quantum system 850. In one embodiment of the present invention, between 1 cycle and 100 cycles are used to readout the state of a qubit. In one embodiment of the present invention, a single cycle that transitions through the Landau-Zener anticrossing (
As shown in
In accordance with some embodiments of the present invention,
Voltage dips 1180 and 1181 in
In an embodiment of the present invention, the voltage dips are in fact peaks because the polarity of the leads to tank circuit 810 wires have been reversed. In an embodiment of the present invention, a qubit in quantum system 850 does not have a crossing or an anticrossing in the region that is probed. The signal from the readout of such a qubit is illustrated as element 1199 in
The fidelity of the readout (step 407 of
In general, and especially during adiabatic evolution for adiabatic quantum computation (transition 404 of
One embodiment of the present invention makes use of a negative feedback loop technique to ensure that Landau-Zener transitions do not occur during adiabatic evolution 404 (
In some embodiments of the present invention, the change in magnitude of the response of tank circuit, χ, ranges from 0.01 radians to about 6 radians for the phase signal. In some embodiments of the present invention, the change in magnitude of the response of the tank circuit, tan(χ), ranges from 0.02 microvolt (μV) to about 1 μV for the amplitude signal.
Embodiments of the present invention can make use of an adiabatic process to readout the state of the superconducting qubit during measurement step 407. Additional flux ΦA and rf flux ΦRF are added to the superconducting qubit and are modulated in accordance with the adiabatic processes described above. In instances where an additional flux generates a dip in the voltage of tank circuit 810 and this additional flux exceeds the amount of flux needed to reach the equilibrium point of the qubit, the qubit is deemed to have been in the |0> quantum state at the beginning of measurement step 407, in accordance with the conventions of
In contrast to the example provided in Section 5.3.2.4, embodiments of the present invention in accordance with this section make use of a diabatic process to readout the state of a superconducting qubit after the adiabatic quantum computation has been completed. In typical embodiments, this is the only part of the process illustrated in
Embodiments of the present invention can make use of repeated adiabatic processes to readout the states of a plurality of superconducting qubits in quantum system 850 (
In preferred embodiments, each qubit in quantum system 850 (
As part of measurement step 407, each qubit, when it is the target qubit, in the plurality of superconducting qubits in quantum system 850 (
Readout using tank circuit 810, unlike readout using a DC-SQUID or a magnetic force microscope MFM, is a quantum non-demolition read out (QND). The readout is quantum non-demolition in nature because the qubit remains in its ground state for the entire duration of the readout. That is, the qubit remains in the ground state (lower energy level 909 of
The detailed exact calculation of energy spectra of an instantaneous Hamiltonian H(t) can be intractable due to exponential growth of the problems size as a function of the number of qubits used in an adiabatic computation. Therefore, approximate evaluation techniques are useful as a best guess of the location of the crossings and anticrossings. Accordingly, some embodiments of the present invention make use of an approximate evaluation method to locate anticrossing of the energies, or energy spectra, of the qubits in quantum system 850. Doing so lessens the time needed to probe the anticrossing and crossings with additional flux.
In an embodiment of the present invention, techniques collectively known as random matrix theory (RMT) are applied to analyze the quantum adiabatic algorithm during readout. See Bougerol and Lacroix, 1985, Random Products of Matrices with Applications to Schrödinger Operators, Birkhäuser, Basel, Switzerland; and Brody et al., 1981, Rev. Mod. Phys. 53, p. 385, each of which is hereby incorporated by reference in its entirety.
In another embodiment of the present invention, spin density functional theory (SDFT) is used as an approximate evaluation method to locate anticrossing of the energies of the system. The probing of the energy spectra by the additional flux can then be used to locate the crossings and anticrossing and perform a readout of the state of superconducting qubits.
In another embodiment of the present invention, the approximate evaluation technique comprises a classical approximation algorithm in order to solve NP-Hard problems. When there is a specific instance of a problem to be solved, the problem is mapped to a description of the qubits being used to solve the NP-Hard problem. This process involves finding an approximation algorithm to the problem being solved by the quantum computer and running the approximation algorithm. The approximate solution is then mapped to the quantum computer's state using the mapping that was used to encode the instance of the NP-Hard problem. This provides a good estimate for the state of the superconducting qubits and lessens the requirements on probing the energy levels for crossings and anticrossings. Such a mapping typically involves setting the coupling energies between the qubits being used to solve the NP-Hard problem so that the qubits approximately represent the problem to be solved. For examples of approximation algorithms useful for the present invention see Goemans and Williamson, “0.878-approximation algorithms for MAX CUT and MAX 2SAT,” In Proceedings of the Twenty-Sixth Annual ACM Symposium on the Theory of Computing, pages 422-431, Montréal, Québec, Canada, 23-25 May 1994; Hochbaum, 1996, Approximation algorithms for NP-hard problems, PWS Publishing Co., Boston; and Cormen et al., 1990, Introduction to Algorithms, MIT Press, Cambridge, each of which is hereby incorporated by reference it its entirety.
In another specific embodiment of the present invention, separations 501 each independently range from about 0.01 microns to about 1 micron. The mutual inductance between qubits 101-4 and 101-5 is about 18.7 picohenries. The mutual inductance between qubits 101-5 and 101-6 is about 13.3 picohenries and the mutual inductance between qubits 101-4 and 101-6 is about 2.5 picohenries.
In an embodiment of the present invention, each qubit 101 in
The adiabatic quantum computing method of the present invention progresses using the system described in
A specific adiabatic quantum computation that can be performed with system 510, an optimization based on an instance of
Step 401 (preparation). Mathematically, to solve MAXCUT as an optimization, the maximum of the following edge pay off function is sought:
The vector |s>={s1, . . . ,si, . . . , s|V|}, where si is either 0, or 1, is a labeling of all the vertices denoting which side of the partition a given vertex is on. The quantity wij is the weight of the edge. In an alternative case of the
In some embodiments of the present invention, the
In an embodiment of the present invention, an instance of
Step 403 (Initialization to H0). The partition of a graph can be found using adiabatic quantum computing. Beginning in the state |000>, which is the ground state of:
the adiabatic quantum computation begins. H0 corresponds to a large magnetic field being applied to all the qubits.
Step 405 (Transition to HP). By slowly reducing the magnetic field while still applying bias to individual qubits proportional to the weights assigned to the qubits, the system is changed to one described by:
Step 407 (Measurement). In step 407, the state of system 510 is determined. To appreciate how measurement of the state of HP occurs using system 510, consider
Measurement of the state of system 510 can occur by repeating the adiabatic quantum computing steps once (steps 401, 403, 405, and 407) for each respective target qubit 101 in system 510. At step 407, the target qubit (the one to be measured) is left in the state determined by HP but all other qubits 101 in the system are biased to a definite state, e.g. all up. The biasing of all but the target qubit occurs by use of the bias lines 507. The state of all the qubits can then be readout and the state of the target qubit can be inferred from the fact that its state either contributes to or against the field created by all the other qubits. The process is repeated and a new target qubit is selected.
In some embodiments, steps 401, 403, 405 and 407 do not have to be repeated for each qubit in system 510. If the number of tank circuits used is greater than one, (e.g., at least one biasing element 507 is on an independent circuit) the process can be parallelized with one target per tank circuit.
In accordance with an aspect of the present invention, superconducting charge qubits can be used in adiabatic quantum computation devices (e.g., in quantum systems 850). In an embodiment of the present invention, capacitively coupled superconducting charge qubits can be used for adiabatic quantum computing. In an embodiment of the present invention, the charge qubits have a fixed or tunable tunneling term. In an embodiment of the present invention, the couplings between charge qubits have a fixed or tunable sign and/or magnitude of coupling. Some embodiments of the present invention are operated in a dilution refrigerator, where the temperature is about 10 milikelvin. Some quantum systems 850 of the present invention are operated at a temperature below 50 milikelvin. The quantum systems 850 of other embodiments of the invention are operated below 200 milikelvin. Still others are operated below 4.2 Kelvin.
A superconducting charge qubit suitable for use in embodiments of the present invention is shown in
In some embodiments, qubit 1220 is made of a film of superconducting material deposited on top of an insulating layer over a ground plane. In an embodiment of the present invention, the film of superconducting material is a layer of aluminum, a thin (about 400 nanometer) layer of Si3N3 is useful insulating layer above a gold ground plane. In such an example, a suitable insulating material is aluminum oxide, used alone or in combination with other insulating materials. Reservoir 1210 can be made of similar materials but is larger than island 1214. In other embodiments of the present invention, the superconducting material is niobium and the insulating material is aluminum oxide.
In an embodiment of the present invention, island 1214 has an effective capacitance to ground of about 600 attofarads and is coupled to the reservoir 1210 through a Josephson junction 1211 with the Josephson energy of about 20 micro-electronvolts (0.232 kelvin). Reservoir 1210 is a big island with about 0.1 nanofarad or larger capacitance relative to the ground plane and is otherwise galvanically isolated from the external environment.
The qubit states are controlled by electrostatic control pulses. The control pulses can be direct current or alternating current created by generator 1216. The control pulses can be used to induce a gate charge on superconducting island 1214.
The energy states of the qubit are controlled by electrostatic control pulses. The control pulses can be direct current or alternating current. The control pulses can be used to induce a gate charge on superconducting island 1214. In some embodiments, the pulses are applied to the superconducting charge qubit islands, to the SET islands and to the trap islands in the various embodiments of the invention. The pulses are applied across the capacitances coupled to the islands. Some embodiments of the present invention require a pulse generator for qubit manipulation and readout. Suitable generators include an Agilent E8247C microwave generator (Agilent Technologies, Inc., Palo Alto, Calif.) a Anritsu MP1763C (Anritsu Corporation, Kanagawa, Japan, or Anritsu Company, Morgan Hill, Calif.).
As shown in
where ΦX is the external flux applied from coil 1227 to DC-SQUID 1212, and EJ0 is the Josephson energy of one of the Josephson junctions in 1212. The DC-SQUID 1212 can also be refereed to as a split Josephson junction, or a split junction. If one half a flux quantum is applied to the split junction the effective Josephson energy is zero, when the Josephson energies of the Josephson junctions are equal. If a flux quantum, or no flux, is applied to the split junction, the effective Josephson energy is 2EJ0, or EJ, MAX. The magnetic flux in the split Josephson junctions can be applied by a global magnetic field.
The Josephson energies of the two Josephson junctions in DC-SQUID 1212 are ideally equal but need not be perfectly equal. The coupling through the DC-SQUID is proportional to the difference between the Josephson energies of the Josephson junctions divided by their sum. With some fabrication practices it is possible to make this ratio one percent or less. Other fabrication processes can make this ratio one tenth of one percent or less. For adiabatic quantum computing there is no need for absolute suppression. Because the superconducting charge qubit are away from the degeneracy point in their final states, the error introduced by unequal Josephson junctions will be small enough to be ignored. This is unlike the circuit model of quantum computing where the mismatch of the two Josephson junctions that make up the DC-SQUID leads to errors in computation. The minimum effective Josephson energy of split Josephson junction is EJ, MIN.
As illustrated in
At readout, resonator 1225 is exited to its resonance frequency. The electric field produced by resonator 1225 has an antinode at the center of the resonator to which the qubit is strongly coupled by capacitance 1228. It has been demonstrated that the coupling between resonator 1225 and the qubit can be made so large that a single photon in the transmission line can resonantly drive Rabi oscillations in the superconducting charge qubit at frequencies in excess of 10 megahertz. See, for example, Wallraff et al, 2004, Nature 431, p. 162, which is hereby incorporated by reference in its entirety.
The rate of coherent exchange of a single excitation between the superconducting charge qubit and resonator 1225 is much larger than the rate at which the superconducting charge qubit decoheres or the rate photons in resonator 1125 are lost. The qubit transitions frequency, the difference between the ground and first energy level in the qubit, expressed in frequency units, can be detuned from the resonator frequency. With this detuning and with a strong coupling between resonator 1225 and the qubit, there is a qubit state-dependent frequency shift in the resonator transition frequency of resonator 1225. This frequency shift can be used to perform a quantum non-demolition (QND) measurement of the qubit state. In this QND measurement, the amplitude and phase of a probe microwave is measured. This probe microwave is at the resonance frequency transmitted through resonator 1225 from port 1222 to port 1223. In the detuned case, the resonator also enhances the qubit radiative lifetime by providing an impedance transformation that effectively filters out the noise of the electromagnetic environment at the qubit transition frequency. See, Devoret, 2004, arXiv.org: cond-mat/0411174, which is hereby incorporated by reference in its entirety.
A single shot readout device is shown in
The island of trap 1233 is coupled to a voltage source 1231 by capacitance 1232. SET island 1239 is coupled to voltage source 1237 by capacitance 1238. SET island 1239 is isolated from the surrounding circuitry by Josephson junctions 1241 and 1242. SET 1299 can be biased by optional current source 1244. The signal from the SET is amplified by amplifier 1246. In some embodiments, amplifier 1246 is the same as amplifier 809 of
The use of trap 1233 enables the user of the system 1250 to separate in time the state manipulation of the qubit's state and readout processes. In addition, qubit island 1214 becomes electrostatically decoupled from SET 1299. The qubit relaxation rate induced by the SET voltage noise is suppressed by a factor of about 10−5 for the capacitance values given above. However, since the qubit's ground state is read out, the relaxation time of the system is not a limiting factor. In other words, a long time is available for readout and this can significantly improve the fidelity. Current SETs have a demonstrated charge sensitivity of as low as a few 10−5 e/{square root}{overscore (Hz)}, which means that the charge variation of 10−5 e can be detected in a measurement time of about 1 second. The measurement precision improves as the square root of the measurement time. Because the final states of adiabatic quantum computing are classical, the operator of an adiabatic quantum computer using superconducting charge qubits has a long time for read out. This aspect alone makes a charge qubit suitable for use in quantum systems that are used for adiabatic quantum computing in accordance with the systems and methods of the present invention.
During operation of the qubit, e.g., the steps of adiabatic quantum computing, the island of trap 1233 is kept unbiased, prohibiting charge relaxation that would arise were charges on qubit island 1214 to tunnel onto the trap. To measure the charge state of island 1214, trap 1233 is biased by a readout pulse applied to capacitance 1232 by source 1231, so that if the superconducting charge qubit is in the excited state, an extra Cooper-pair charge tunnels into the trap in a sequential two-quasiparticle process. These quasiparticles, or electrons, are the constituents of the Cooper-pair and create an excess charge on the island of trap 1233. This excess charge is then detected by SET 1299. System 1299 can be constructed so that the quasiparticle charge on the island of trap 1233 does tunnel back to island 1214 of the qubit with appreciable probability. The quasiparticle charge on the island of trap 1233 influences island 1239 of SET 1299 through capacitor 1234. See, for example, Devoret and Schoelkopf, 2000, Nature 406, pp. 1039-1046; Astafiev et al., Phys. Rev. Lett. 93, 267007 (2004); and Astafiev et al., 2004, Phys. Rev. B 69, 180507, each of which is hereby incorporated by reference in its entirety.
Embodiments of the present invention can be operated without trap 1233, capacitor 1234. In such embodiments, island 1239 of SET 1299 is coupled directly to the superconducting charge qubit island 1214. In such embodiments the single shot readout may not be possible but repeated measurements can establish the final state of the superconducting charge qubit. The rate at which measurements can be cycled in such an embodiment is limited by the RC constant of the read out system. For resistance values through SET island 1239 across junction 1241 and 1242 of about 100 kΩ (kiloohms) and signal cable capacitance 1 nanofarad, the corresponding RC constant limits the measurement cycle to a few kilohertz. However, this is not a significant limitation because the adiabatic evolution that created the final state for readout is a slower process.
A radio frequency single electron transistor coupled to a superconducting charge qubit is shown in
SET island 1259 is coupled to voltage source 1257 by capacitance 1258. In an embodiment of the present invention, the capacitance of capacitor 1258 is about 20 attofarads. In some embodiments of the present invention, the capacitance of capacitor ranges between about 1 attofarad and 100 attofarads. SET island 1259 is isolated from the surrounding circuit by resistive tunnel junctions 1255 and 1256. RF-SET 1297 can be voltage biased by a voltage applied on terminal 1264 relative to ground. SET island 1259 is coupled to inductance 1263 and capacitance 1262 of RF-SET 1297. The value of inductance 1263 and capacitance 1262 is chosen to create a resonance frequency of RF-SET 1297. For example, if inductance 1263 is 600 nanoheneries and capacitance 1262 is about a picofarad, the resonance frequency is about 330 megahertz. The measurement cycle time of system 1270 is proportional to the RC constant of RF-SET 1297. For numerical values given above, and an overall RF-SET resistance of 40 kiloohms, the RC time is about a tenth of a microsecond, or a cycle of seven megahertz. The value of the inductance 1263 can be easily and widely varied in fabrication.
In operation, for superconducting charge qubit read out at the end of the adiabatic quantum computation using system 1270 of
The readout scheme specified above for system 1270 has many advantages. It operates at a speed higher than a normal SET. By using a low-impedance and matched impedance high-frequency amplifier, the stray capacitance of the cabling between the SET island 1259 and the amplifier (not shown) becomes unimportant. The amplifier is physically and thermally separated from RF-SET 1298, and can be optimized without the constraint of very low power dissipation required for operation at millikelvin temperatures. Finally, because the readout is performed at a high frequency, there is no amplifier contribution to the 1/f noise. See, for example, Duty et al., 2004, Phys. Rev. B 69, 140503; Schoelkopf et al., 1998, Science 280, pp. 1238-1241; and Aassime et al., 2001, Appl. Phys. Lett 79, pp. 4031-4033, each of which is hereby incorporated by reference in its entirety.
The values of capacitance, inductance, and Josephson energy given in relation to superconducting charge qubits above, is provided in order to describe operable embodiments, but these are not the only values that lead to operable embodiments. The values can be changed with and without the need to make corresponding changes to the cooperative components. There are freely available and commercially available software that can aid the designer of superconducting charge qubits. Such software includes, but is not limited to, FASTCAP a tool designed to calculate capacitances of a given layout of superconducting devices. FASTCAP is freely distributed by the Research Laboratory of Electronics of the Massachusetts Institute of Technology, Cambridge, Mass. See also Nabors et al., 1992, IEEE Trans. On Microwave Theory and Techniques 40, pp. 1496-1507; and Nabors, September 1993, “Fast Three-Dimensional Capacitance Calculation,” Thesis MIT, each of which is hereby incorporated by reference in its entirety. Another useful tool for varying the capacitances, inductances, and Josephson energies in the examples above is JSPICE. JSPICE is a simulator for superconductor and semiconductor circuits with Josephson junction, and is based on the general-purpose circuit simulation program SPICE. SPICE originates from the EECS Department of the University of California at Berkeley and provides links to the download website for this free software. See also Whiteley, 1991, IEEE Trans. On Magnetics 27, pp. 2908-2905; and Gaj, et al., 1999, IEEE Trans. Appl. Supercond. 9, pp. 18-38, each of which is hereby incorporated by reference in its entirety. In an embodiment of the present invention, the thermal energy (kBT) of the charge qubit is less than the charge energy of the superconducting charge qubit (EC). In an embodiment of the present invention, the Josephson energy (EJ) of the charge qubit is greater than the superconducting material energy gap of the charge qubit. In an embodiment of the present invention, the charge energy of the charge qubit is about equal to the Josephson energy of the charge qubit.
Some embodiments of the present invention comprise a lattice of interconnected superconducting charge qubits that are capacitively coupled to each other. This is a difference between these charge qubits and the systems of
As shown in
In the embodiment illustrated in
The Hamiltonian for lattices of superconducting charge qubits 1220, 1230, 1310, and 1330, etc. is:
which is similar in form to Hamiltonians given above for coupled phase qubits. All terms in the Hamiltonian are potentially in the initial, final, or instant Hamiltonian of a computational device that is performing an adiabatic quantum computation in accordance with
The biasing term proportional to the σZ Pauli matrix is tunable. The bias is applied to the superconducting charge qubit via capacitor coupled to the qubit. Mathematically the term is:
E′C=4EC(1−2ng)
where the dimensionless gate charge ng≈CgVg/(2e) is determined in operation by the gate voltage Vg and in fabrication by the gate capacitance Cg. The gate capacitance is capacitance 1215. Here it is assumed that ng≈½, but in parts of the computation the dimensional charge and hence the state of the qubit is biased to either 1 or 0.
The tunneling term proportional to the σX Pauli matrix is always present except for embodiments where the Josephson energy term EJ is tunable. In an embodiment of the present invention, the charge qubits with a tunable tunneling term have a Josephson junction connected to the island of the charge qubit replaced by split Josephson junction. In these embodiments, tunneling is suppressed by tuning the flux in the split Josephson junction. The suppression of tunneling is absolute when the Josephson energies of junction are equal and the flux in the DC-SQUID loop is a half flux quantum. The tunneling is maximal (at zero bias) when the flux in the split Josephson junction is a whole number multiple of a flux quantum (Φ0), including zero. Traditionally, the tunneling of the qubit states is suppressed when the qubits are in the final state of an adiabatic quantum computation. Flux through split Josephson junctions can be tuned by a global magnetic field applied perpendicular to the plane of superconducting charge qubits. The global field can have local corrections applied through flux coils like 1227 of
In embodiments of the present invention, the coupling term proportional to the tensor product of σZ Pauli matrix is tunable. In other embodiments, the coupling term is fixed. Coupling between two superconducting charge qubits can have many forms. If the Hamiltonian term for the coupling has a principal component proportional to σZ×σZ, then the coupling is suitable for adiabatic quantum computing. An example of a coupling between superconducting charge qubits that has a principal component proportional to σZ×σZ, is capacitive coupling. The sign of this coupling is positive making it an antiferromagnetic coupling. The coupling sign and coefficient (Jij) are not tunable for a simple capacitance coupling unless a VET is used. The coupling in terms of physical parameters is
where qi and qj is the excess charge on the ith and jth qubit, and
where CkΣ is the total capacitance for the kth qubit. The coupling sign and coefficient (Jij) are tunable if a variable electrostatic transformer (VET) is used. The VET can be used to create ferromagnetic and antiferromagnetic interactions between superconducting qubits. The VET is described in Averin and Bruder, 2003, Phys. Rev. Lett. 91, 057003, which is hereby incorporated by reference in its entirety.
In accordance with the general procedure of adiabatic quantum computing as shown in
In step 403, the plurality superconducting charge qubits is initialized to the ground state of the time-independent Hamiltonian, H0, which initially describes the states of the qubits. It is assumed that the ground state of H0 is a state to which the plurality superconducting charge qubits can be reliably and reproducibly set. This assumption is reasonable for plurality superconducting charge qubits including qubit bias terms.
In transition 404 between steps 403 and 405, the plurality superconducting charge qubits are acted upon in an adiabatic manner in order to alter the states of the plurality superconducting charge qubits. The plurality superconducting charge qubits change from being described by Hamiltonian H0 to a description under HP. This change is adiabatic, as defined above, and occurs in a time T. In other words, the operator of an adiabatic quantum computer causes the plurality superconducting charge qubits, and Hamiltonian H describing the plurality superconducting charge qubits, to change from H0 to a final form HP in time T. The change is an interpolation between H0 and HP.
In accordance with the adiabatic theorem of quantum mechanics, the plurality of superconducting charge qubits will remain in the ground state of H at every instance the qubits are changed and after the change is complete, provided the change is adiabatic. In some embodiments of the present invention, the plurality of superconducting charge qubits start in an initial state H0 that does not permit quantum tunneling, are perturbed in an adiabatic manner to an intermediate state that permits quantum tunneling, and then are perturbed in an adiabatic manner to the final state described above.
In step 405, the plurality of superconducting charge qubits has been altered to a state that is described by the final Hamiltonian. The final Hamiltonian HP can encode the constraints of a computational problem such that the state of the plurality superconducting charge qubits under HP corresponds to a solution to this problem. Hence, the final Hamiltonian is also called the problem Hamiltonian HP. If the plurality superconducting charge qubits is not in the ground state of HP, the state is an approximate solution to the computational problem. Approximate solutions for many computational problems are useful and such embodiments are fully within the scope of the present invention.
An aspect of the problem Hamiltonian HP is the energy of the tunneling terms in the Hamiltonian, which are either weak or zero. The energy of the tunneling terms is suppressed by applying a half flux quantum through the split Josephson junctions that separate the superconducting charge qubit island from the reservoir.
In step 407, the plurality superconducting charge qubits described by the final Hamiltonian HP is read out. The read out can be in the σZ basis of the qubits. If the read out basis commutes with the terms of the problem Hamiltonian Hp, then performing a read out operation does not disturb the ground state of the system. The read out method can take many forms. The object of the read out step is to determine exactly or approximately the ground state of the system. The states of all qubits are represented by the vector {right arrow over (O)}, which gives a concise image of the ground state or approximate ground state of the system. The plurality superconducting charge qubits can be readout using the devices and techniques described hereinabove.
This example describes an embodiment of the invention where the superconducting charge qubits have fixed couplings. To perform adiabatic quantum computation with superconducting charge qubits, in step 401, the operator chooses E′Ci and Jij in such a way as to simulate the problem Hamiltonian Hp. The charging energy E′Ci can be controlled by the gate voltages, and qubit-qubit coupling energy Jij is fixed in fabrication by the capacitance of the coupling capacitances unless a variable electrostatic transformer is used. In an embodiment of the present invention, these values are chosen to satisfy the constraint EJ,MIN<<E′Ci and EJ,MIN<<Jij while E′Ci≦EJ,MAX and Jij≦EJ,MAX, for every qubit or pair of coupled qubits. At the beginning of the computation, step 403, EJi are set to EJ,MAX. In that case, the tunneling terms dominate causing a decay of the system to the ground state. Then, in transition 404, the EJi terms are adiabatically reduced to EJ,MIN. At the end of the adiabatic evolution, the tunneling terms are negligible and the effective Hamiltonian is HP. Then, in step 405, the qubits can be read out. Suitable techniques for single shot readout include the use of a trap island, an RF-SET, QED read out circuits, etc.
This example describes an embodiment of the invention where the superconducting charge qubits have fixed couplings. In an embodiment of the present invention where E′Ci is greater than EJ,MAX, but not much greater, the operator can artificially reduce E′Ci. To perform adiabatic quantum computation with superconducting charge qubits, step 401, the operator chooses Jij in such a way as to simulate the problem Hamiltonian Hp. In an embodiment of the present invention, these values are chosen to satisfy EJ,MIN<<E′Ci and EJ,MIN<<Jij while E′Ci≧EJ,MAX and Jij≦EJ,MAX, for every qubit or pair of coupled qubits. At the beginning of the computation, step 403, each EJi is set to EJ,MAX. The gate voltages on the superconducting charge qubits are tuned to bring the qubits closer to degeneracy [dimensionless gate charge equal to one half (ng=½)]. This reduces the energy in the bias term E′Ci. Then, in transition 404, the EJi terms are adiabatically reduced to EJ,MIN. Also, in the adiabatic evolution the bias on the superconducting charge qubit is adiabatically increased at the same time as each EJi is decreased. The bias on the superconducting charge qubits is changed such that, at the end of the adiabatic evolution, the bias terms are appropriate for the problem Hamiltonian is HP. Then, in step 405, the qubits are read out.
Further embodiments for adiabatic quantum computing with superconducting charge qubits can be constructed from the following protocols for the biasing, tunneling, and coupling of qubits.
For the tunneling terms of a plurality of superconducting charge qubits the following protocols can be used. In an embodiment where the tunneling term is fixed, the bias and coupling energies for the initial Hamiltonian are set to be less that the tunneling energies. In the same examples, the bias and coupling energies for the problem Hamiltonian are set to be greater that the tunneling energies. In such a case, the tunneling is weak compared with the final Hamiltonian. In embodiments where the tunneling term is tunable, the tunneling energies for the initial Hamiltonian are set to values approaching and including EJ,MAX but away from zero. After the adiabatic evolution the tunneling energies are set to EJ,MIN, which can be zero. Alternatively, the tunneling energies for the initial Hamiltonian are set to EJ,MIN, which can be zero. During the adiabatic evolution the tunneling energies are to values approaching and including EJ,MAX but away from zero. After the adiabatic evolution the tunneling energies are set to EJ,MIN, which can be zero.
For the coupling terms of a plurality of superconducting charge qubits the following protocols can be used. In an embodiment where the coupling terms are fixed at fabrication time, the couplings are set such that they describe the couplings for the problem Hamiltonian. In embodiments where the coupling terms are tunable, the coupling energies for the initial Hamiltonian are set to zero. During the adiabatic evolution the coupling energies are increased such that they describe the couplings for the problem Hamiltonian. If the tunneling energies are at zero during readout, the coupling energies can be set to any value without the final state of the superconducting charge qubits charging.
System 1500 further includes an adiabatic quantum computing device 1540 that includes those adiabatic quantum computing devices shown above. Exemplary examples of adiabatic quantum computing devices 1540 include, but are not limited to, devices 101 of
System 1500 further includes a readout control system 1560. In some embodiments, readout control system 1560 comprises a plurality of magnetometers, or electrometers, where each magnetometer or electrometer is inductively coupled, or capacitively coupled, respectively, to a different qubit in quantum computing device 1540. In such embodiments, controller 1516 receives a signal, by way of readout control system 1560, from each magnetometer or electrometer in readout device 1560. System 1500 optionally comprises a qubit control system 1565 for the qubits in quantum computing device 1540. In some embodiments, qubit control system 1565 comprises a magnetic field source or electric field source that is inductively coupled or capacitively coupled, respectively, to a qubit in quantum computing device 1540. System 1500 optionally comprises a coupling device control system 1570 to control the couplings between qubits in adiabatic quantum computing device 1540. A coupling that is controllable includes, but is not limited to, coupling 1325 of
In some embodiments, memory 1520 includes a number of modules and data structures. It will be appreciated that at any one time during operation of the system, all or a portion of the modules and/or data structures stored in memory 1520 are resident in random access memory (RAM) and all or a portion of the modules and/or data structures are stored in non-volatile storage 1515. Furthermore, although memory 1520, including non-volatile memory 1515, is shown as housed within computer 1501, the present invention is not so limited. Memory 1520 is any memory housed within computer 1501 or that is housed within one or more external digital computers (not shown) that are addressable by digital computer 1501 across a network (e.g., a wide area network such as the internet).
In some embodiments, memory 1520 includes an operating system 1521. Operating system 1521 includes procedures for handling various system services, such as file services, and for performing hardware dependant tasks. In some embodiments of the present invention, the programs and data stored in system memory 1520 further include an adiabatic quantum computing device interface module 1523 for defining and executing a problem to be solved on an adiabatic quantum computing device. In some embodiments, memory 1520 includes a driver module 1527. Driver module 1527 includes procedures for interfacing with and handling the various peripheral units to computer 1501, such as controller 1516 and control systems 1560, qubit control system 1565, coupling device control system 1570, and adiabatic quantum computing device 1540. In some embodiments of the present invention, the programs and data stored in system memory 1520 further include a readout module 1530 for interpreting the data from controller 1516 and readout control system 1560.
The functionality of controller 1516 can be divided into two classes of functionality: data acquisition and control. In some embodiments, two different types of chips are used to handle each of these discrete functional classes. Data acquisition can be used to measure physical properties of the qubits in adiabatic quantum computing device 1540 after adiabatic evolution has been completed. Such data can be measured using any number of customized or commercially available data acquisition microcontrollers including, but not limited to, data acquisition cards manufactured by Elan Digital Systems (Fareham, UK) including, but are not limited to, the AD132, AD136, MF232, MF236, AD142, AD218, and CF241. In some embodiments, data acquisition and control is handled by a single type of microprocessor, such as the Elan D403C or D480C. In some embodiments, there are multiple interface cards 1516 in order to provide sufficient control over the qubits in a computation 1540 and in order to measure the results of an adiabatic quantum computation on 1540.
When introducing elements of the present invention or the embodiment(s) thereof, the articles “a,” “an,” “the,” and “said” are intended to mean that there are one or more of the elements. The terms “comprising,” “including,” and “having” are intended to be inclusive and to mean that there may be additional elements other than the listed elements. Moreover, the term “about” has been used to describe specific parameters. In many instances, specific ranges for the term “about” have been provided. However, when no specific range has been provided for a particular usage of the term “about” herein, than either of two definitions can be used. In the first definition, the term “about” is the typical range of values about the stated value that one of skill in the art would expect for the physical parameter represented by the stated value. For example, a typical range of values about a specified value can be defined as the typical error that would be expected in measuring or observing the physical parameter that the specified value represents. In the second definition of about, the term “about” means the stated value ±0.10 of the stated value. As used herein, the term “instance” means the execution of a step. For example, in a multistep method, a particular step may be repeated. Each repetition of this step is referred to herein as an “instance” of the step.
All references cited herein are incorporated herein by reference in their entirety and for all purposes to the same extent as if each individual publication or patent or patent application was specifically and individually indicated to be incorporated by reference in its entirety for all purposes.
The present invention can be implemented as a computer program product that comprises a computer program mechanism embedded in a computer readable storage medium. For instance, the computer program product could contain program modules, such as those illustrated in
Many modifications and variations of this invention can be made without departing from its spirit and scope, as will be apparent to those skilled in the art. The specific embodiments described herein are offered by way of example only, and the invention is to be limited only by the terms of the appended claims, along with the full scope of equivalents to which such claims are entitled.
This application claims benefit, under 35 U.S.C. § 119(e), of U.S. Provisional Patent Application No. 60/557,748, filed on Mar. 29, 2004, which is hereby incorporated by reference in its entirety. This application also claims benefit, under 35 U.S.C. § 119(e), of U.S. Provisional Patent Application No. 60/588,002, filed on Jul. 13, 2004, which is hereby incorporated by reference in its entirety. This application is further related to concurrently filed application Ser. No. ______, Attorney Docket No. 706700-999193, entitled “Adiabatic Quantum Computation with Superconducting Qubits,” and application Ser. No. ______, Attorney Docket No. 706700-999208, entitled “Adiabatic Quantum Computation with Superconducting Qubits,” each of which is hereby incorporated by reference in its entirety.
| Number | Date | Country | |
|---|---|---|---|
| 60557748 | Mar 2004 | US | |
| 60588002 | Jul 2004 | US |
