WO2023243011A1 - 情報処理プログラム、情報処理方法、および情報処理装置 - Google Patents
情報処理プログラム、情報処理方法、および情報処理装置 Download PDFInfo
- Publication number
- WO2023243011A1 WO2023243011A1 PCT/JP2022/024000 JP2022024000W WO2023243011A1 WO 2023243011 A1 WO2023243011 A1 WO 2023243011A1 JP 2022024000 W JP2022024000 W JP 2022024000W WO 2023243011 A1 WO2023243011 A1 WO 2023243011A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- value
- update process
- parameter
- coefficient
- information processing
- Prior art date
Links
- 230000010365 information processing Effects 0.000 title claims abstract description 31
- 238000003672 processing method Methods 0.000 title claims description 8
- 238000000034 method Methods 0.000 claims abstract description 113
- 230000008569 process Effects 0.000 claims abstract description 106
- 238000004364 calculation method Methods 0.000 claims abstract description 103
- 230000008859 change Effects 0.000 claims abstract description 44
- 238000012545 processing Methods 0.000 claims description 41
- 238000005457 optimization Methods 0.000 description 89
- 239000002096 quantum dot Substances 0.000 description 39
- 238000010586 diagram Methods 0.000 description 14
- 230000006870 function Effects 0.000 description 11
- 230000003287 optical effect Effects 0.000 description 7
- 230000007423 decrease Effects 0.000 description 6
- UFHFLCQGNIYNRP-UHFFFAOYSA-N Hydrogen Chemical compound [H][H] UFHFLCQGNIYNRP-UHFFFAOYSA-N 0.000 description 5
- 238000004891 communication Methods 0.000 description 4
- 230000000694 effects Effects 0.000 description 4
- 229910052739 hydrogen Inorganic materials 0.000 description 4
- 239000001257 hydrogen Substances 0.000 description 4
- 238000003077 quantum chemistry computational method Methods 0.000 description 4
- 238000005259 measurement Methods 0.000 description 3
- 230000002093 peripheral effect Effects 0.000 description 3
- 238000005401 electroluminescence Methods 0.000 description 2
- 230000005283 ground state Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000010276 construction Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000011478 gradient descent method Methods 0.000 description 1
- 238000012804 iterative process Methods 0.000 description 1
- 239000004973 liquid crystal related substance Substances 0.000 description 1
- 230000010355 oscillation Effects 0.000 description 1
- 239000004065 semiconductor Substances 0.000 description 1
- 239000007787 solid Substances 0.000 description 1
- 239000000126 substance Substances 0.000 description 1
- 230000007704 transition Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N10/00—Quantum computing, i.e. information processing based on quantum-mechanical phenomena
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N99/00—Subject matter not provided for in other groups of this subclass
Definitions
- the present invention relates to an information processing program, an information processing method, and an information processing device.
- VQE Variational Quantum Eigensolver
- a quantum computer measures the expected value of a quantum state based on a variational quantum circuit parameterized by a plurality of parameters ⁇ .
- the expected value of energy can be obtained from the expected value of the quantum state.
- a classical computer adjusts the parameter ⁇ so that the energy is lower.
- Such a process for adjusting the parameter ⁇ is called an optimization process.
- the quantum computer generates a quantum state using the optimized parameter ⁇ and measures the expected value again.
- Quantum computers and classical computers repeatedly measure the expected value of the quantum state and optimize parameters until the energy converges.
- a quantum computing device As a technology related to quantum computation, for example, a quantum computing device has been proposed that can execute a quantum variational algorithm even if the error rate of the quantum computing device is not a sufficiently small value. Techniques regarding generation of trial states for VQE have also been proposed. In addition, classical computers estimate the expected value of a Hamiltonian that can be expressed as a linear combination of observables based on the expected value of the observables for a quantum state, and convert one or both of the Hamiltonian and the quantum state. Optimization methods have also been proposed. Furthermore, techniques have been proposed to facilitate quantum computation of Monte Carlo minimization.
- the present invention aims to reduce VQE computation time.
- One proposal provides an information processing program that causes a computer to multiple times update the values of parameters applied to a variational quantum circuit used for VQE calculations.
- the computer periodically changes the value for each update process of the coefficient used in the update process of the parameter value to a value higher and lower than a predetermined reference value as the number of updates indicating the number of update processes increases. Decide on a value that changes over time.
- the computer changes the value of the parameter from the pre-update value to the changed value by the amount of change according to the value of the coefficient determined for the number of updates of the update process to be executed. , update the value of the parameter.
- VQE computation time is reduced.
- FIG. 3 is a diagram illustrating an example of an information processing method according to the first embodiment. It is a figure showing an example of system configuration of a 2nd embodiment.
- 1 is a diagram showing an example of classical computer hardware;
- FIG. It is a diagram showing an example of a variational quantum circuit.
- FIG. 2 is a block diagram illustrating an example of the functionality of a classical computer for VQE calculations.
- 3 is a flowchart illustrating an example of a procedure for VQE calculation processing. It is a figure which shows an example of VQE calculation of the energy of a hydrogen molecule.
- FIG. 3 is a diagram illustrating an example of a change in step size.
- FIG. 3 is a diagram illustrating an example of the relationship between m and the number of optimizations.
- the present embodiment will be described below with reference to the drawings. Note that each embodiment can be implemented by combining a plurality of embodiments within a consistent range.
- the first embodiment is an information processing method that reduces the number of optimization iterations and shortens calculation time by accelerating energy convergence in VQE calculations.
- FIG. 1 is a diagram illustrating an example of an information processing method according to the first embodiment.
- FIG. 1 shows an information processing apparatus 10 that implements an information processing method.
- the information processing device 10 can implement the information processing method by executing an information processing program, for example.
- the information processing device 10 includes a storage section 11 and a processing section 12.
- the storage unit 11 is, for example, a memory or a storage device included in the information processing device 10.
- the processing unit 12 is, for example, a processor or an arithmetic circuit included in the information processing device 10.
- the storage unit 11 stores a variational quantum circuit 1 corresponding to a quantum many-body system to be solved by VQE calculation.
- the variational quantum circuit 1 is parameterized by, for example, a plurality of parameter sets ⁇ ( ⁇ 1 , ⁇ 2 , . . . ).
- the processing unit 12 performs VQE calculation.
- the processing unit 12 uses, for example, the quantum computer 2 to measure the expected value of the quantum state by the variational quantum circuit 1 to which the value of the parameter at that time is applied.
- the processing unit 12 calculates the energy of the quantum many-body system based on the expected value of the quantum state.
- the processing unit 12 determines whether the calculated energy satisfies a predetermined convergence condition, and if it does not, updates the value of the parameter set ⁇ in a direction that reduces the energy. Such updating of the values of the set ⁇ of a plurality of parameters is called parameter optimization.
- the processing unit 12 repeatedly performs expected value measurement and parameter optimization using the quantum computer 2 until the energy satisfies the convergence condition.
- the processing unit 12 dynamically adjusts a coefficient that adjusts the magnitude of the change in the value of the set ⁇ of a plurality of parameters in parameter optimization as the VQE calculation progresses.
- Change to The coefficient is, for example, the step size ⁇ .
- the step size ⁇ is, for example, the learning rate in the gradient descent method.
- the processing unit 12 changes the value for each update process of the coefficient used in the update process of the value of the parameter applied to the variational quantum circuit 1 to a predetermined value as the number of updates indicating the number of update processes increases. A value that periodically changes between higher and lower values than the reference value is determined.
- the number of updates is k (k is a natural number)
- the value of the coefficient is determined for each value of the number of updates k.
- the reference value ( ⁇ 0 ) is “0.05”.
- the change period of the coefficient value may be, for example, one period with a fixed number of updates.
- two update processes constitute one cycle. In this case, each time the number of updates increases by 1, values higher and lower than the reference value are alternately repeated.
- the processing unit 12 can dynamically calculate the value of the coefficient for each update process during the VQE calculation process. For example, when the kth first update process of the parameter value is performed in the process of VQE calculation, the processing unit 12 updates the k+1st update process based on the first change amount of the parameter value in the first update process. Calculate the value of the coefficient used in the second update process.
- the processing unit 12 updates the next update process based on the average value of the first change amount of each parameter value of the plurality of parameter sets ⁇ ( ⁇ 1 , ⁇ 2 , . . . ) in the first update process. Calculate the values of coefficients commonly used in determining the updated values of a plurality of parameters in the update process of step 2.
- ” is the amount of change in the parameter value in the k-th update process (the difference in the parameter value before and after the update).
- the processing unit 12 determines the amount of change in the value of the parameter in the update process repeatedly executed during the VQE calculation process based on the value of the coefficient determined for the number of updates of the update process to be executed. For example, the processing unit 12 increases the amount of change in the parameter value as the coefficient value increases. Then, the processing unit 12 updates the value of the parameter from the value of the parameter before updating to the value changed by the determined amount of change.
- the processing unit 12 determines that the coefficient value is the product ( ⁇ k ( ⁇ f( ⁇ )/ ⁇ p ) of the slope of the cost function (f( ⁇ )) corresponding to the parameter value before updating and the coefficient value ) is determined as the amount of change.
- the cost function is, for example, a function whose value decreases as the energy of the quantum many-body system to be solved decreases.
- the processing unit 12 periodically changes the value of the coefficient between a value larger and a value smaller than the reference value. This speeds up energy convergence when the coefficient value becomes larger than the reference value, and reduces the number of iterative processes (including parameter optimization and expected value measurement).
- the amount of change in the parameter can be appropriately adjusted by changing the value of the coefficient.
- the processing unit 12 calculates the value of the coefficient to be used in the (k+1)th update process based on the amount of change in the value of the parameter in the k-th update process. For example, the processing unit 12 decreases the value of the coefficient used in the (k+1)th update process as the amount of change in the value of the k-th update process parameter increases. Thereby, it is possible to dynamically calculate the value of the coefficient that periodically changes between higher and lower values than the reference value during the VQE calculation process.
- the second embodiment shortens the VQE calculation time by accelerating energy convergence in VQE calculation using a quantum computer.
- the process of updating the value of the set of parameters ⁇ so as to reduce the energy of the quantum many-body system is called an optimization process.
- the number of times optimization processing is executed during the VQE calculation process is called the number of optimization times.
- a coefficient that adjusts the amount of change in the set ⁇ of a plurality of parameters is called a step size.
- FIG. 2 is a diagram showing an example of the system configuration of the second embodiment.
- the classical computer 100 and the quantum computer 200 are connected via a network.
- Classical computer 100 is a Neumann type computer.
- the classical computer 100 performs processing such as parameter optimization calculation in VQE calculation.
- the quantum computer 200 is a quantum gate type quantum computer that performs desired calculations by manipulating the states of quantum bits based on quantum circuits. In the VQE calculation, the quantum computer 200 obtains the expected value of the quantum state indicated by the variational quantum circuit based on the variational quantum circuit according to the value of the specified parameter.
- FIG. 3 is a diagram showing an example of classical computer hardware.
- the classical computer 100 is entirely controlled by a processor 101.
- a memory 102 and a plurality of peripheral devices are connected to the processor 101 via a bus 109.
- Processor 101 may be a multiprocessor.
- the processor 101 is, for example, a CPU (Central Processing Unit), an MPU (Micro Processing Unit), or a DSP (Digital Signal Processor).
- At least a part of the functions realized by the processor 101 executing a program may be realized by an electronic circuit such as an ASIC (Application Specific Integrated Circuit) or a PLD (Programmable Logic Device).
- ASIC Application Specific Integrated Circuit
- PLD Programmable Logic Device
- the memory 102 is used as the main storage device of the classical computer 100.
- the memory 102 temporarily stores at least a portion of OS (Operating System) programs and application programs to be executed by the processor 101.
- the memory 102 also stores various data used for processing by the processor 101.
- a volatile semiconductor storage device such as a RAM (Random Access Memory) is used, for example.
- Peripheral devices connected to the bus 109 include a storage device 103, a GPU (Graphics Processing Unit) 104, an input interface 105, an optical drive device 106, a device connection interface 107, and a network interface 108.
- a storage device 103 a graphics processing unit
- a GPU Graphics Processing Unit
- the storage device 103 electrically or magnetically writes and reads data to and from a built-in recording medium.
- the storage device 103 is used as an auxiliary storage device for the classical computer 100.
- the storage device 103 stores OS programs, application programs, and various data.
- an HDD Hard Disk Drive
- an SSD Solid State Drive
- the GPU 104 is an arithmetic unit that performs image processing, and is also called a graphics controller.
- a monitor 21 is connected to the GPU 104.
- the GPU 104 displays an image on the screen of the monitor 21 according to instructions from the processor 101. Examples of the monitor 21 include a display device using organic EL (Electro Luminescence) and a liquid crystal display device.
- a keyboard 22 and a mouse 23 are connected to the input interface 105.
- the input interface 105 transmits signals sent from the keyboard 22 and mouse 23 to the processor 101.
- the mouse 23 is an example of a pointing device, and other pointing devices can also be used.
- Other pointing devices include touch panels, tablets, touch pads, trackballs, and the like.
- the optical drive device 106 reads data recorded on the optical disc 24 or writes data to the optical disc 24 using laser light or the like.
- the optical disc 24 is a portable recording medium on which data is recorded so as to be readable by reflection of light. Examples of the optical disc 24 include a DVD (Digital Versatile Disc), a DVD-RAM, a CD-ROM (Compact Disc Read Only Memory), and a CD-R (Recordable)/RW (ReWritable).
- the device connection interface 107 is a communication interface for connecting peripheral devices to the classical computer 100.
- a memory device 25 or a memory reader/writer 26 can be connected to the device connection interface 107.
- the memory device 25 is a recording medium equipped with a communication function with the device connection interface 107.
- the memory reader/writer 26 is a device that writes data to or reads data from the memory card 27.
- the memory card 27 is a card-type recording medium.
- the network interface 108 is connected to the quantum computer 200 via a network.
- the network interface 108 transmits information such as a request for quantum calculation to the quantum computer 200 and receives information indicating a calculation result from the quantum computer 200.
- the network interface 108 is a wired communication interface that is connected to a wired communication device such as a switch or a router using a cable.
- the classical computer 100 can implement the processing functions of the second embodiment using the hardware described above. Note that the apparatus shown in the first embodiment can also be realized by the same hardware as the classical computer 100 shown in FIG. 3.
- the classical computer 100 realizes the processing functions of the second embodiment by, for example, executing a program recorded on a computer-readable recording medium.
- a program that describes the processing content to be executed by the classical computer 100 can be recorded on various recording media.
- a program to be executed by the classical computer 100 can be stored in the storage device 103.
- the processor 101 loads at least part of the program in the storage device 103 into the memory 102 and executes the program.
- the program to be executed by the classical computer 100 may be recorded on a portable recording medium such as the optical disk 24, the memory device 25, or the memory card 27.
- the program stored in the portable recording medium becomes executable after being installed in the storage device 103 under the control of the processor 101, for example.
- the processor 101 can also directly read and execute a program from a portable recording medium.
- the classical computer 100 and the quantum computer 200 cooperate to execute VQE calculations.
- a plurality of parameter sets ⁇ are used in the variational quantum circuit used for VQE calculation.
- FIG. 4 is a diagram showing an example of a variational quantum circuit.
- FIG. 4 shows an example of a variational quantum circuit 30 for determining the base value of the energy of hydrogen molecules.
- operations on four quantum bits (qubits 0 to 3) are shown.
- a horizontal line associated with each quantum bit indicates a gate operation for that quantum bit.
- gate operations set for each quantum bit are executed in order from the left.
- the one-qubit gates 31a to 31l are quantum gates that perform a rotation operation around a predetermined axis by a specified angle.
- the variational quantum circuit 30 is shown to sequentially perform a rotation operation around the z-axis, a rotation operation around the y-axis, and a rotation operation around the z-axis for each quantum bit.
- the rotation angles of the one-qubit gates 31a to 31c that act on the first quantum bit (qubit 0) are ⁇ 0 , ⁇ 1 , and ⁇ 2 , respectively.
- the rotation angles of the one-qubit gates 31d to 31f acting on the second quantum bit (qubit 1) are ⁇ 3 , ⁇ 4 , and ⁇ 5 , respectively.
- the rotation angles of the one-qubit gates 31g to 31i acting on the third qubit (qubit 2) are ⁇ 6 , ⁇ 7 , and ⁇ 8 , respectively.
- the rotation angles of the one-qubit gates 31j to 31l that act on the fourth qubit (qubit 3) are ⁇ 9 , ⁇ 10 , and ⁇ 11 , respectively.
- the two-qubit gate 32a is a CNOT gate that indicates a CNOT operation between the third and fourth qubits.
- the third qubit is the control qubit
- the fourth qubit is the target qubit.
- the two-qubit gate 32b is a CNOT gate that indicates a CNOT operation between the third and first qubit.
- the third qubit is the control qubit
- the first qubit is the target qubit.
- the two-qubit gate 32c is a CNOT gate that indicates a CNOT operation between the fourth and second quantum bits.
- the fourth qubit is the control qubit
- the second qubit is the target qubit.
- Symbols 33a to 33d shown at the right end of the line corresponding to each quantum bit indicate quantum state measurement operations.
- a gradient is used, for example, in optimizing the set ⁇ of a plurality of parameters.
- the value of the p-th (p is a natural number) parameter ⁇ p in the k-th (k is a natural number) optimization process is assumed to be ⁇ p,k .
- ⁇ p,k+1 in the k+1st optimization process can be calculated using the following equation (1) using ⁇ p,k .
- ⁇ is a parameter (step size) that determines the weight of numerical values updated in one optimization process.
- ⁇ is also called the learning rate.
- f( ⁇ ) is a cost function representing energy.
- ⁇ f( ⁇ )/ ⁇ p is the axial gradient of the parameter ⁇ p .
- the gradient is the partial differential coefficient with respect to the parameter ⁇ p at the point ( ⁇ 1,k , ⁇ 2,k , . . . ) of f( ⁇ ).
- the value of step size ⁇ is fixed and optimization is performed.
- the ⁇ value is too large, especially in the early stages of optimization, the values of multiple parameter sets ⁇ may change too much in a single optimization process, and the optimization path that should originally be followed may be changed too much. There is a possibility that the optimization will not go well due to deviation.
- the value of ⁇ is too small at the final stage of optimization, the amount of change in each value of a plurality of parameter sets ⁇ may be underestimated, resulting in an increase in the number of optimizations.
- the classical computer 100 changes the value of the step size ⁇ for each optimization process, as shown in equations (2) and (3).
- ⁇ k is the step size in the k-th optimization process.
- D 0 is a preset constant (real number).
- D k is the average value of the amount of change in the parameter ⁇ p in the set ⁇ of a plurality of parameters in the k-th optimization process.
- D k is determined according to equation (4).
- N is the number of parameters (N is a natural number).
- m in equation (3) is a preset constant (real number). Note that the value of m is selected so that the number of optimization iterations is as small as possible.
- FIG. 5 is a block diagram illustrating an example of the functionality of a classical computer for VQE calculations.
- the classical computer 100 includes a quantum calculation management section 110 and an optimization calculation section 120.
- the quantum calculation management unit 110 generates a variational quantum circuit for calculating the energy of a quantum many-body system such as a molecule, and instructs the quantum computer 200 to perform energy calculation based on the variational quantum circuit.
- the quantum calculation management unit 110 generates a variational quantum circuit for quantum chemical calculation, and sets a set ⁇ of a plurality of parameters related to gate operations at quantum gates in the variational quantum circuit.
- the quantum calculation management unit 110 sets initial values to the values of a plurality of parameter sets ⁇ before the first energy calculation based on the variational quantum circuit.
- the initial value of each parameter is, for example, a value specified in advance by the user. Furthermore, random values may be used as the initial values of each parameter.
- the quantum calculation management unit 110 obtains from the quantum computer 200 an energy calculation result based on a variational quantum circuit parameterized by a plurality of parameter sets ⁇ . Upon acquiring the energy calculation results, the quantum calculation management unit 110 determines whether the energy has converged. If the energy has not converged, the quantum calculation management unit 110 instructs the optimization calculation unit 120 to optimize the parameters.
- the optimization calculation unit 120 optimizes a plurality of parameter sets ⁇ for each optimization process. For example, the optimization calculation unit 120 updates the value of the set ⁇ of a plurality of parameters in a direction that decreases the energy value. When the optimization calculation is completed, the optimization calculation unit 120 notifies the quantum calculation management unit 110 of the updated value of the set of parameters ⁇ .
- each element shown in FIG. 5 can be realized, for example, by causing a computer to execute a program module corresponding to the element.
- the procedure of VQE calculation processing will be explained in detail.
- FIG. 6 is a flowchart illustrating an example of a procedure for VQE calculation processing. The process shown in FIG. 6 will be described below in accordance with step numbers.
- the quantum calculation management unit 110 uses, for example, a prespecified value as the initial value of the set ⁇ of the plurality of parameters.
- the quantum calculation management unit 110 sets an initial value ⁇ 0 of the step size ⁇ used in parameter optimization.
- the initial value ⁇ 0 is, for example, a value specified in advance.
- the quantum calculation management unit 110 instructs the quantum computer 200 to measure the expected value.
- the quantum calculation management unit 110 transmits the generated variational quantum circuit and the value of a set of parameters ⁇ to the quantum computer 200, and instructs the quantum computer 200 to calculate the expected value of each quantum bit based on the variational quantum circuit.
- the quantum computer 200 measures the expected value of a quantum bit based on a variational quantum circuit parameterized by a plurality of parameter sets ⁇ .
- the quantum calculation management unit 110 determines whether the energy has converged. For example, the quantum calculation management unit 110 calculates the value of a cost function f( ⁇ ) representing energy, and uses the calculation result as an energy value. The quantum calculation management unit 110 determines that the energy has converged when the energy value satisfies a predetermined convergence condition. For example, the quantum calculation management unit 110 determines that the energy has converged if the energy value has reached a value known as the ground state energy value. Further, the quantum calculation management unit 110 may determine that the energy has converged when the difference between the energy value calculated this time and the energy value calculated last time is less than or equal to a predetermined threshold.
- the quantum calculation management unit 110 When the energy converges, the quantum calculation management unit 110 outputs a solution according to the state of the quantum bit at that time, and ends the VQE calculation process. Further, if the energy has not converged, the quantum calculation management unit 110 advances the process to step S105.
- the quantum calculation management unit 110 stores the set ⁇ of a plurality of parameters in the memory 102 as ⁇ old .
- the optimization calculation unit 120 performs optimization calculation of the set ⁇ of a plurality of parameters using the previously calculated step size ⁇ . For example, the optimization calculation unit 120 performs the calculation shown in equation (2). Note that in the first optimization calculation process, the optimization calculation unit 120 sets the preset initial value ⁇ 0 of the step size ⁇ as the applied step size ⁇ . The optimization calculation unit 120 updates the value of the set of parameters ⁇ to the value calculated by the optimization calculation.
- Step S107 The optimization calculation unit 120 calculates a new step size based on the average value of the differences between the updated parameter set ⁇ and ⁇ old .
- the calculated step size is used in the next optimization calculation. After that, the optimization calculation unit 120 advances the process to step S103.
- FIG. 7 is a diagram showing an example of VQE calculation of the energy of hydrogen molecules.
- graphs 31 and 32 show the results of calculating the energy of hydrogen molecules (H 2 ) using VQE when the interatomic distance is 0.7 ⁇ .
- the horizontal axis is a value indicating the number of optimization processes (the number of optimizations), and the vertical axis is the energy value.
- Graph 32 is an expanded version of the range of energy values from “-1.16" to "-1.06" in graph 31.
- black circles indicate changes in energy when the step size ⁇ is a fixed value
- open circles indicate changes in energy when the step size ⁇ is varied.
- the energy calculated in one optimization process decreases as the number of repetitions of the optimization process increases.
- a dynamically varying value is used as the step size ⁇
- the energy decreases faster than when a fixed value is used as the step size ⁇ .
- the number of optimizations when the energy converges is “93”.
- the number of optimizations when the energy converges is “42”.
- FIG. 8 is a diagram showing an example of how the step size changes.
- the horizontal axis is the number of optimizations
- the vertical axis is the step size.
- black circles indicate changes in the step size when the step size ⁇ is a fixed value
- white circles indicate changes in the step size when the step size ⁇ is varied.
- the initial values of the step size ⁇ are both “0.05”.
- the step size ⁇ is a fixed value, the step size remains “0.05” throughout the VQE calculation.
- step size ⁇ When dynamically varying the step size ⁇ , each time the optimization count increases by 1, the step size ⁇ alternates between a larger value and a smaller value than the initial value. Furthermore, as the number of optimizations increases, the difference in step size ⁇ from the initial value becomes larger. That is, the amplitude of the variation in step size ⁇ increases as the number of optimizations increases.
- the amount of change in the set of parameters ⁇ in the k-th optimization process is given by the average value D k of equation (4).
- the average value D k is the denominator in the parentheses of the (D 0 /D k ) m part, so the larger the average value D k
- the step size ⁇ k becomes smaller. That is, the larger the amount of change in the set of parameters ⁇ , the smaller the step size ⁇ k of the next optimization process.
- the value of m in equation (3) is "0.6".
- FIG. 9 is a diagram showing an example of the relationship between m and the number of optimizations.
- a graph 34 shown in FIG. 9 shows the number of optimizations at the time of energy convergence depending on the value of m.
- the horizontal axis of the graph 34 is the value of m, and the vertical axis is the number of optimizations at the time of energy convergence.
- the step size ⁇ always remains at the initial value ⁇ 0 . Therefore, when the value of m is "0", the number of optimizations during energy convergence is "93" as shown in FIG. Further, when the value of m is "0.6", the number of optimizations at the time of energy convergence is "42" as shown in FIG. That is, the effect of reducing the number of optimizations at the time of energy convergence by dynamically varying the step size ⁇ is about 55% at maximum.
- the step size ⁇ is calculated for each optimization process, but the step size for each optimization process is set in advance so that the step size varies as shown in FIG. You can stay there.
- the step size ⁇ alternately repeats values larger and smaller than the initial value for each optimization process, but for example, it is possible to make the oscillation period of the step size ⁇ longer.
- one cycle consists of four optimization processes, and after executing two optimization processes with a step size larger than the initial value, the optimization process is executed twice with a step size smaller than the initial value. It is also possible to do so.
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Computing Systems (AREA)
- Software Systems (AREA)
- Mathematical Physics (AREA)
- General Engineering & Computer Science (AREA)
- Condensed Matter Physics & Semiconductors (AREA)
- Mathematical Optimization (AREA)
- Pure & Applied Mathematics (AREA)
- Mathematical Analysis (AREA)
- Evolutionary Computation (AREA)
- Data Mining & Analysis (AREA)
- Computational Mathematics (AREA)
- Artificial Intelligence (AREA)
- Complex Calculations (AREA)
Abstract
Description
コンピュータは、パラメータの値の更新処理で使用する係数の更新処理ごとの値を、更新処理が何回目なのかを示す更新回数の増加に伴って所定の基準値より高い値と低い値とに周期的に変化する値に決定する。そしてコンピュータは、VQE計算で複数回実行する更新処理において、実行する更新処理の更新回数に対して決定された係数の値に応じた変化量だけ、更新前のパラメータの値から変化させた値へ、パラメータの値を更新する。
本発明の上記および他の目的、特徴および利点は本発明の例として好ましい実施の形態を表す添付の図面と関連した以下の説明により明らかになるであろう。
〔第1の実施の形態〕
第1の実施の形態は、VQE計算においてエネルギーの収束を早めることで最適化の繰り返し回数を削減し、計算時間を短縮する情報処理方法である。
処理部12は、VQEの計算過程で繰り返し実行する更新処理において、実行する更新処理の更新回数に対して決定された係数の値に基づいてパラメータの値の変化量を決定する。例えば処理部12は、係数の値が大きいほどパラメータの値の変化量を大きくする。そして処理部12は、決定した変化量だけ更新前のパラメータの値から変化させた値へ、パラメータの値を更新する。
第2の実施の形態は、量子コンピュータを用いたVQE計算において、エネルギーの収束を早めることで、VQE計算時間を短縮するものである。なお、第2の実施の形態では、量子多体系のエネルギーを小さくするように複数のパラメータの組θの値を更新する処理を最適化処理と呼ぶ。VQE計算過程での最適化処理の実行回数を最適化回数と呼ぶ。また最適化処理において複数のパラメータの組θの変化量を調整する係数をステップサイズと呼ぶ。
このような変分量子回路30を用いてVQE計算を行う場合、例えば複数のパラメータの組θの最適化において勾配が用いられる。ここでp番目(pは自然数)のパラメータθpのk回目(kは自然数)の最適化処理における値をθp,kとする。例えば、最適化のk+1回目の最適化処理におけるθp,k+1は、θp,kを用いて次の式(1)で計算できる。
図5は、VQE計算のための古典コンピュータの機能の一例を示すブロック図である。古典コンピュータ100は、量子計算管理部110と最適化計算部120とを有する。
次にVQE計算処理の手順について詳細に説明する。
[ステップS101]量子計算管理部110は、複数のパラメータの組θ={θ1,θ2,・・・}によりパラメタライズされた変分量子回路を生成する。量子計算管理部110は、複数のパラメータの組θの初期値としては、例えば予め指定された値を用いる。
[ステップS103]量子計算管理部110は、量子コンピュータ200へ、期待値の測定を指示する。例えば量子計算管理部110は、生成した変分量子回路と複数のパラメータの組θの値とを量子コンピュータ200に送信し、変分量子回路に基づく各量子ビットの期待値の計算を指示する。量子コンピュータ200は、複数のパラメータの組θでパラメタライズされた変分量子回路に基づいて、量子ビットの期待値を測定する。
[ステップS106]最適化計算部120は、前回計算したステップサイズηを用いて、複数のパラメータの組θの最適化計算を行う。例えば最適化計算部120は、式(2)に示す計算を行う。なお最適化計算部120は、最初の最適化計算処理では、予め設定されたステップサイズηの初期値η0を、適用するステップサイズηとする。最適化計算部120は、最適化計算によって算出された値に、複数のパラメータの組θの値を更新する。
図7は、水素分子のエネルギーのVQE計算の一例を示す図である。図7では、原子間距離が0.7Åの場合の水素分子(H2)のエネルギーをVQEで計算した結果が、グラフ31,32に示されている。
第2の実施の形態では、ステップサイズηを最適化処理ごとに計算しているが、図8に示す通りにステップサイズが変動するように、最適化処理ごとのステップサイズを予め設定しておいてもよい。
2 量子コンピュータ
10 情報処理装置
11 記憶部
12 処理部
Claims (7)
- VQE(Variational Quantum Eigensolver)計算に使用する変分量子回路に適用するパラメータの値の更新処理を複数回コンピュータに実行させるプログラムであって、
前記パラメータの値の前記更新処理で使用する係数の前記更新処理ごとの値を、前記更新処理が何回目なのかを示す更新回数の増加に伴って所定の基準値より高い値と低い値とに周期的に変化する値に決定し、
前記VQE計算で複数回実行する前記更新処理において、実行する前記更新処理の前記更新回数に対して決定された前記係数の値に応じた変化量だけ、更新前の前記パラメータの値から変化させた値へ、前記パラメータの値を更新する、
処理をコンピュータに実行させる情報処理プログラム。 - 前記パラメータの値を更新する処理では、前記係数の値が大きいほど前記変化量を大きくする、
請求項1記載の情報処理プログラム。 - 前記係数の前記更新処理ごとの値を決定する処理では、前記VQE計算の過程で前記パラメータの値のk回目(kは自然数)の第1の更新処理が行われると、前記第1の更新処理における前記パラメータの値の第1の変化量に基づいて、k+1回目の第2の更新処理で使用する前記係数の値を算出する、
請求項1または2に記載の情報処理プログラム。 - 前記係数の前記更新処理ごとの値を決定する処理では、複数の前記パラメータそれぞれの値の前記第1の変化量の平均値に基づいて、前記第2の更新処理における複数の前記パラメータの更新後の値の決定に共通で使用する前記係数の値を計算する、
請求項3記載の情報処理プログラム。 - 前記係数の前記更新処理ごとの値を決定する処理では、前記更新回数が1増えるごとに前記基準値より高い値と低い値とを交互に繰り返す値に、前記更新回数ごとの前記係数の値を決定する、
請求項1記載の情報処理プログラム。 - VQE計算に使用する変分量子回路に適用するパラメータの値の更新処理を複数回コンピュータに実行させるプログラムであって、
前記パラメータの値の前記更新処理で使用する係数の前記更新処理ごとの値を、前記更新処理が何回目なのかを示す更新回数の増加に伴って所定の基準値より高い値と低い値とに周期的に変化する値に決定し、
前記VQE計算で複数回実行する前記更新処理において、実行する前記更新処理の前記更新回数に対して決定された前記係数の値に応じた変化量だけ、更新前の前記パラメータの値から変化させた値へ、前記パラメータの値を更新する、
処理をコンピュータが実行する情報処理方法。 - VQE計算に使用する変分量子回路に適用するパラメータの値の更新処理を複数回実行する情報処理装置であって、
前記パラメータの値の前記更新処理で使用する係数の前記更新処理ごとの値を、前記更新処理が何回目なのかを示す更新回数の増加に伴って所定の基準値より高い値と低い値とに周期的に変化する値に決定し、前記VQE計算で複数回実行する前記更新処理において、実行する前記更新処理の前記更新回数に対して決定された前記係数の値に応じた変化量だけ、更新前の前記パラメータの値から変化させた値へ、前記パラメータの値を更新する処理部、
を有する情報処理装置。
Priority Applications (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
JP2024528001A JPWO2023243011A1 (ja) | 2022-06-15 | 2022-06-15 | |
PCT/JP2022/024000 WO2023243011A1 (ja) | 2022-06-15 | 2022-06-15 | 情報処理プログラム、情報処理方法、および情報処理装置 |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
PCT/JP2022/024000 WO2023243011A1 (ja) | 2022-06-15 | 2022-06-15 | 情報処理プログラム、情報処理方法、および情報処理装置 |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2023243011A1 true WO2023243011A1 (ja) | 2023-12-21 |
Family
ID=89192456
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/JP2022/024000 WO2023243011A1 (ja) | 2022-06-15 | 2022-06-15 | 情報処理プログラム、情報処理方法、および情報処理装置 |
Country Status (2)
Country | Link |
---|---|
JP (1) | JPWO2023243011A1 (ja) |
WO (1) | WO2023243011A1 (ja) |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20190384597A1 (en) | 2018-06-13 | 2019-12-19 | International Business Machines Corporation | Short circuit depth variational quantum computation of monte carlo minimization and nth order moments |
US20200057957A1 (en) | 2018-08-17 | 2020-02-20 | Zapata Computing, Inc. | Quantum Computer with Improved Quantum Optimization by Exploiting Marginal Data |
JP2020534607A (ja) | 2017-09-22 | 2020-11-26 | インターナショナル・ビジネス・マシーンズ・コーポレーションInternational Business Machines Corporation | 量子コンピューティング・マシンのための高ハードウェア効率変分量子固有値ソルバを実現するためのシステム、方法、量子コンピューティング・デバイスおよびコンピュータ・プログラム |
JP2021026370A (ja) | 2019-08-01 | 2021-02-22 | 日本電信電話株式会社 | 量子演算装置及び方法 |
-
2022
- 2022-06-15 WO PCT/JP2022/024000 patent/WO2023243011A1/ja active Application Filing
- 2022-06-15 JP JP2024528001A patent/JPWO2023243011A1/ja active Pending
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2020534607A (ja) | 2017-09-22 | 2020-11-26 | インターナショナル・ビジネス・マシーンズ・コーポレーションInternational Business Machines Corporation | 量子コンピューティング・マシンのための高ハードウェア効率変分量子固有値ソルバを実現するためのシステム、方法、量子コンピューティング・デバイスおよびコンピュータ・プログラム |
US20190384597A1 (en) | 2018-06-13 | 2019-12-19 | International Business Machines Corporation | Short circuit depth variational quantum computation of monte carlo minimization and nth order moments |
US20200057957A1 (en) | 2018-08-17 | 2020-02-20 | Zapata Computing, Inc. | Quantum Computer with Improved Quantum Optimization by Exploiting Marginal Data |
JP2021026370A (ja) | 2019-08-01 | 2021-02-22 | 日本電信電話株式会社 | 量子演算装置及び方法 |
Non-Patent Citations (2)
Title |
---|
HUANG RUI, TAN XIAOQING, XU QINGSHAN: "Learning to Learn Variational Quantum Algorithm", IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS, IEEE, USA, vol. 34, no. 11, 1 November 2023 (2023-11-01), USA, pages 8430 - 8440, XP093117830, ISSN: 2162-237X, DOI: 10.1109/TNNLS.2022.3151127 * |
SMITH LESLIE N: "Cyclical Learning Rates for Training Neural Networks", ARXIV (CORNELL UNIVERSITY), CORNELL UNIVERSITY LIBRARY, ARXIV.ORG, ITHACA, 4 April 2017 (2017-04-04), Ithaca, XP093117833, [retrieved on 20240110], DOI: 10.48550/arxiv.1506.01186v6 * |
Also Published As
Publication number | Publication date |
---|---|
JPWO2023243011A1 (ja) | 2023-12-21 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Bataineh et al. | Neural network for regression problems with reduced training sets | |
US11288589B1 (en) | Quantum circuit modeling | |
CN114897173B (zh) | 基于变分量子线路确定PageRank的方法及装置 | |
JP7015405B1 (ja) | 学習モデルの生成方法、プログラム、情報処理装置及び学習用データの生成方法 | |
US20220012291A1 (en) | Information processing system, information processing method, and non-transitory computer-readable storage medium for storing program | |
Patel et al. | Curriculum reinforcement learning for quantum architecture search under hardware errors | |
JP2021086371A (ja) | 学習プログラム、学習方法および学習装置 | |
JP2024508076A (ja) | 量子回路シミュレーション方法、装置、コンピュータ機器及びプログラム | |
JP7511230B2 (ja) | 量子回路生成装置、量子回路生成方法及び量子回路生成プログラム | |
WO2023243011A1 (ja) | 情報処理プログラム、情報処理方法、および情報処理装置 | |
EP3614314B1 (en) | Method and apparatus for generating chemical structure using neural network | |
EP4273761A1 (en) | Solution search program, solution search method, and information processing device | |
JP7624420B2 (ja) | 情報処理システム、情報処理方法、及び情報処理プログラム | |
WO2024134823A1 (ja) | 情報処理プログラム、情報処理方法、および情報処理装置 | |
JP7428932B2 (ja) | 量子計算制御プログラム、量子計算制御方法及び情報処理装置 | |
CN116010754A (zh) | 存储程序的计算机可读记录介质、数据处理方法和设备 | |
JP7436830B2 (ja) | 学習プログラム、学習方法、および学習装置 | |
JP7323519B2 (ja) | データ管理システム、データ管理装置、データ管理方法及びデータ管理プログラム | |
WO2023144884A1 (ja) | パラメータ最適化プログラム、パラメータ最適化方法、および情報処理装置 | |
WO2024004072A1 (ja) | 量子回路設計プログラム、量子回路設計方法および量子回路設計装置 | |
EP4475041A1 (en) | Information processing program, information processing method, and information processing device | |
JP7489876B2 (ja) | モデル解析装置、モデル解析方法、及びプログラム | |
JP7553873B2 (ja) | 量子回路設計装置、量子回路設計プログラムおよび量子回路設計方法 | |
WO2024029540A1 (ja) | 計算方法、計算システム、及びプログラム | |
WO2024089862A1 (ja) | 分子シミュレーションプログラム、分子シミュレーション方法および情報処理装置 |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 22946829 Country of ref document: EP Kind code of ref document: A1 |
|
ENP | Entry into the national phase |
Ref document number: 2024528001 Country of ref document: JP Kind code of ref document: A |
|
WWE | Wipo information: entry into national phase |
Ref document number: 2022946829 Country of ref document: EP |
|
ENP | Entry into the national phase |
Ref document number: 2022946829 Country of ref document: EP Effective date: 20250115 |