Search

US-12620456-B2 - Data processing method and apparatus for quantum chemistry system

US12620456B2US 12620456 B2US12620456 B2US 12620456B2US-12620456-B2

Abstract

The present disclosure relates to a data processing method and apparatus for a quantum chemistry system. The data processing method for a quantum chemistry system comprises the following steps: acquiring a specific wave function which is constructed on the basis of a neural network and is suitable for a quantum chemistry system; performing walker processing on the basis of the specific wave function, the walker processing comprising diffusion processing of walkers; and determining related information about chemical properties of the quantum chemistry system on the basis of the specific wave function and the processed walkers.

Inventors

  • Weiluo REN
  • Chenlin CHAI
  • Weizhong FU
  • Ji Chen

Assignees

  • Beijing Youzhuju Network Technology Co., Ltd.

Dates

Publication Date
20260505
Application Date
20240614
Priority Date
20220415

Claims (20)

  1. 1 . A data processing method for a quantum chemical system, the method comprising, by at least one computing node comprising a graphics processing unit: constructing a specific wave function suitable for the quantum chemical system based on a neural network, wherein the specific wave function is a wave function that characterizes states of electrons in the quantum chemical system, wherein constructing the specific wave function suitable for the quantum chemical system based on the neural network comprises: training the neural network by using a variational Monte Carlo method with an initial electron distribution configuration as an input and with minimizing an energy state as a training goal, wherein corresponding network parameters when the energy reaches a minimum value correspond to the specific wave function; performing walker processing on walkers based on the specific wave function, wherein a walker corresponds to an electron in the quantum chemical system and comprises information about electron attributes in the quantum chemical system; and determining relevant information of chemical properties of the quantum chemical system based on the specific wave function and the processed walkers, the relevant information of the chemical properties of the quantum chemical system comprises attributes related to ground-state energy of electrons in the quantum chemical system, wherein performing the walker processing on the walkers based on the specific wave function is executed iteratively, and in each iteration, performing the walker processing further comprises: performing drift-diffusion on a walker with a preset drift amount; adjusting a weight for a diffused walker based on a local energy of the diffused walker; detecting whether the diffused walker crosses a nodal surface of the specific wave function, by judging whether a spatial position of the diffused walker crosses the nodal surface, wherein the nodal surface of the specific wave function is a plane comprised of all of points of the specific wave function having a value of zero; and performing processing related to a walker configuration on diffused walkers based on a detection result of the diffused walkers, wherein performing the processing related to the walker configuration on the diffused walkers based on the detection result of the diffused walkers, further comprises: in response to detecting that the diffused walker does not cross the nodal surface of the specific wave function, performing further processing based on the weight for the diffused walker, wherein the further processing comprises at least one of reproduction, merging, or no operation on the diffused walker, and wherein determining the relevant information of the chemical properties of the quantum chemical system based on the specific wave function and the processed walkers further comprises: calculating statistical data of the processed walkers based on the specific wave function, to determine the relevant information of the chemical properties of the quantum chemical system, wherein the statistical data comprises statistical data of energies of the walkers.
  2. 2 . The method of claim 1 , wherein performing the processing related to the walker configuration on the diffused walkers based on the detection result of the diffused walkers further comprises: resetting a configuration of a diffused walker when it is detected that the diffused walker crosses the nodal surface.
  3. 3 . The method of claim 1 , wherein performing the further processing based on the weight for the diffused walker comprises at least one of the following: in response to the weight for the diffused walker being greater than or equal to a first weight threshold, performing reproduction on the diffused walker to duplicate the diffused walker into two walkers; in response to the weight for the diffused walker being less than a second weight threshold, performing merging on the diffused walker to merge two walkers into one walker; or in response to the weight for the diffused walker being greater than or equal to the second weight threshold and less than the first weight threshold, not performing the reproduction and merging on the diffused walker.
  4. 4 . The method of claim 1 , wherein performing processing related to the walker configuration on the diffused walkers based on the detection result of the diffused walkers, further comprises: in response to detecting that the diffused walker does not cross the nodal surface of the specific wave function, performing a processing on the diffused walker that enables a number of walkers after this iteration of processing to remain constant.
  5. 5 . The method of claim 4 , wherein the processing that enables the number of walkers after this iteration of processing to remain constant comprises at least one of the following: duplicating an original walker if a weight for the original walker is greater than a specific threshold, a weight for each of two walkers obtained through the duplication being half of the weight for the original walker, and selecting two walkers with minimum weights for merging; or duplicating the original walker if the weight for the original walker is greater than a specific threshold, the weight for each of the two walkers obtained through the duplication being half of the weight for the original walker, and replacing, with one duplicated walker, a walker annihilated in the merging concurrently performed.
  6. 6 . The method of claim 1 , wherein the walker processing is performed on a plurality of computing nodes in a distributed manner, at least one of the plurality of computing nodes comprises a graphics processing unit, and a part of the walker processing is performed on each computing node, and wherein the drift-diffusion of walkers is iteratively performed, and the drift-diffusion of walkers is performed such that after each iteration of processing, a number of walkers remain constant at least on the computing node that comprises the graphics processing unit.
  7. 7 . The method of claim 1 , wherein when the walker processing is performed on a plurality of computing nodes in a distributed manner, a detection point is generated synchronously among the plurality of computing nodes periodically or as required, and wherein the drift-diffusion of walkers is iteratively performed, and the drift-diffusion of walkers is performed such that after each iteration of processing, a number of walkers remain constant at least on the computing node that comprises the graphics processing unit.
  8. 8 . The method of claim 1 , wherein determining the relevant information of the chemical properties of the quantum chemical system based on the specific wave function and the processed walkers comprises: calculating local energy of each walker based on the specific wave function; performing weighted averaging on the local energy of each walker to obtain total energy reflected by all walkers; and determining, based on the total energy reflected by all the walkers, information reflecting the ground-state energy of the quantum chemical system.
  9. 9 . The method of claim 8 , wherein the drift-diffusion of walkers is iteratively performed, and the iteration of the drift-diffusion of walkers is terminated under at least one of the following conditions: a number of iterations exceeds a threshold number; or in a specific number of consecutive iterations, a change in the total energy reflected by all the walkers is less than a specific change threshold.
  10. 10 . An electronic device, comprising: a memory; and a processor coupled to the memory, wherein the memory stores executable instructions that, when executed by the processor, cause the electronic device to perform: constructing a specific wave function suitable for a quantum chemical system based on a neural network, wherein the specific wave function is a wave function that characterizes states of electrons in the quantum chemical system, wherein constructing the specific wave function suitable for the quantum chemical system based on the neural network comprises: training the neural network by using a variational Monte Carlo method with an initial electron distribution configuration as an input and with minimizing an energy state as a training goal, wherein corresponding network parameters when the energy reaches a minimum value correspond to the specific wave function; performing walker processing on walkers based on the specific wave function, wherein a walker corresponds to an electron in the quantum chemical system and comprises information about electron attributes in the quantum chemical system; and determining relevant information of chemical properties of the quantum chemical system based on the specific wave function and the processed walkers, the relevant information of the chemical properties of the quantum chemical system comprises attributes related to ground-state energy of electrons in the quantum chemical system, wherein performing the walker processing on the walkers based on the specific wave function is executed iteratively, and in each iteration, performing the walker processing further comprises: performing drift-diffusion on a walker with a preset drift amount; adjusting a weight for a diffused walker based on a local energy of the diffused walker; detecting whether the diffused walker crosses a nodal surface of the specific wave function, by judging whether a spatial position of the diffused walker crosses the nodal surface, wherein the nodal surface of the specific wave function is a plane comprised of all of points of the specific wave function having a value of zero; and performing processing related to a walker configuration on diffused walkers based on a detection result of the diffused walkers, wherein performing the processing related to the walker configuration on the diffused walkers based on the detection result of the diffused walkers, further comprises: in response to detecting that the diffused walker does not cross the nodal surface of the specific wave function, performing further processing based on the weight for the diffused walker, wherein the further processing comprises at least one of reproduction, merging, or no operation on the diffused walker, and wherein determining the relevant information of the chemical properties of the quantum chemical system based on the specific wave function and the processed walkers further comprises: calculating statistical data of the processed diffused walkers based on the specific wave function, to determine the relevant information of the chemical properties of the quantum chemical system, wherein the statistical data comprises statistical data of energies of the walkers.
  11. 11 . The electronic device of claim 10 , wherein performing the further processing based on the weight for the diffused walker comprises at least one of the following: in response to the weight for the diffused walker being greater than or equal to a first weight threshold, performing reproduction on the diffused walker to duplicate the diffused walker into two walkers; in response to the weight for the diffused walker being less than a second weight threshold, performing merging on the diffused walker to merge two walkers into one walker; or in response to the weight for the diffused walker being greater than or equal to the second weight threshold and less than the first weight threshold, not performing the reproduction and merging on the diffused walker.
  12. 12 . The electronic device of claim 10 , wherein performing the processing related to the walker configuration on the diffused walkers based on the detection result of the diffused walkers further comprises: resetting a configuration of a diffused walker when it is detected that the diffused walker crosses the nodal surface.
  13. 13 . The electronic device of claim 10 , wherein performing the processing related to the walker configuration on the diffused walkers based on the detection result of the diffused walkers, further comprises: in response detecting that the diffused walker does not cross the nodal surface of the specific wave function, performing a processing on the diffused walker that enables a number of walkers after this iteration of processing to remain constant.
  14. 14 . The electronic device of claim 13 , wherein the processing that enables the number of walkers after this iteration of processing to remain constant comprises at least one of the following: duplicating an original walker if a weight for the original walker is greater than a specific threshold, a weight for each of two walkers obtained through the duplication being half of the weight for the original walker, and selecting two walkers with minimum weights for merging; or duplicating the original walker if the weight for the original walker is greater than a specific threshold, the weight for each of the two walkers obtained through the duplication being half of the weight for the original walker, and replacing, with one duplicated walker, a walker annihilated in the merging concurrently performed.
  15. 15 . The electronic device of claim 10 , wherein determining the relevant information of the chemical properties of the quantum chemical system based on the specific wave function and the processed walkers comprises: calculating local energy of each walker based on the specific wave function; performing weighted averaging on the local energy of each walker to obtain total energy reflected by all walkers; and determining, based on the total energy reflected by all the walkers, information reflecting the ground-state energy of the quantum chemical system.
  16. 16 . A non-transitory computer-readable storage medium having executable instructions stored thereon, wherein the instructions, when executed by a processor, cause implementation of: constructing a specific wave function suitable for a quantum chemical system based on a neural network, wherein the specific wave function is a wave function that characterizes states of electrons in the quantum chemical system, wherein constructing the specific wave function suitable for the quantum chemical system based on the neural network comprises: training the neural network by using a variational Monte Carlo method with an initial electron distribution configuration as an input and with minimizing an energy state as a training goal, wherein corresponding network parameters when the energy reaches a minimum value correspond to the specific wave function; performing walker processing on walkers based on the specific wave function, wherein a walker corresponds to an electron in the quantum chemical system and comprises information about electron attributes in the quantum chemical system; and determining relevant information of chemical properties of the quantum chemical system based on the specific wave function and the processed walkers, the relevant information of the chemical properties of the quantum chemical system comprises attributes related to ground-state energy of electrons in the quantum chemical system, wherein performing the walker processing on the walkers based on the specific wave function is executed iteratively, and in each iteration, performing the walker processing further comprises: performing drift-diffusion on a walker with a preset drift amount; adjusting a weight for a diffused walker based on a local energy of the diffused walker; detecting whether the diffused walker crosses a nodal surface of the specific wave function, by judging whether a spatial position of the diffused walker crosses the nodal surface, wherein the nodal surface of the specific wave function is a plane comprised of all of points of the specific wave function having a value of zero; and performing processing related to a walker configuration on diffused walkers based on a detection result of the diffused walkers, wherein performing the processing related to the walker configuration on the diffused walkers based on the detection result of the diffused walkers, further comprises: in response to detecting that the diffused walker does not cross the nodal surface of the specific wave function, performing further processing based on the weight for the diffused walker, wherein the further processing comprises at least one of reproduction, merging, or no operation on the diffused walker, and wherein determining the relevant information of the chemical properties of the quantum chemical system based on the specific wave function and the processed walkers further comprises: calculating statistical data of the processed diffused walkers based on the specific wave function, to determine the relevant information of the chemical properties of the quantum chemical system, wherein the statistical data comprises statistical data of energies of the walkers.
  17. 17 . The non-transitory computer-readable storage medium of claim 16 , wherein performing the further processing performed based on the weight for the diffused walker comprises at least one of the following: in response to the weight for the diffused walker being greater than or equal to a first weight threshold, performing reproduction on the diffused walker to duplicate the diffused walker into two walkers; in response to the weight for the diffused walker being less than a second weight threshold, performing merging on the diffused walker to merge two walkers into one walker; or in response to the weight for the diffused walker being greater than or equal to the second weight threshold and less than the first weight threshold, not performing the reproduction and merging on the diffused walker.
  18. 18 . The non-transitory computer-readable storage medium of claim 16 , wherein performing the processing related to the walker configuration on the diffused walkers based on the detection result of the diffused walkers further comprises: resetting a configuration of a diffused walker when it is detected that the diffused walker crosses the nodal surface.
  19. 19 . The non-transitory computer-readable storage medium of claim 16 , wherein performing the processing related to the walker configuration on the diffused walkers based on the detection result of the diffused walkers, further comprises: in response to detecting that the diffused walker does not cross the nodal surface of the specific wave function, performing a processing on the diffused walker that enables a number of walkers after this iteration of processing to remain constant.
  20. 20 . The non-transitory computer-readable storage medium of claim 19 , wherein the processing that enables the number of walkers after this iteration of processing to remain constant comprises at least one of the following: duplicating an original walker if a weight for the original walker is greater than a specific threshold, a weight for each of two walkers obtained through the duplication being half of the weight for the original walker, and selecting two walkers with minimum weights for merging; or duplicating the original walker if the weight for the original walker is greater than a specific threshold, the weight for each of the two walkers obtained through the duplication being half of the weight for the original walker, and replacing, with one duplicated walker, a walker annihilated in the merging concurrently performed.

Description

CROSS-REFERENCE TO RELATED APPLICATIONS The present application is a continuation application of PCT/CN2023/088084 filed Apr. 13, 2023, which is based on and claims priority to Chinese Application No. 202210395445.9, filed on Apr. 15, 2022, both the above-mentioned applications are incorporated herein by reference in their entireties. TECHNICAL FIELD The present disclosure relates to the field of quantum chemistry, and in particular, to data processing for a chemical system. BACKGROUND ART Quantum chemistry is an important branch of theoretical chemistry. Its research scope may include structures and properties of stable and unstable molecules, and relationships between the structures and properties; interaction between molecules; and mutual collisions and reactions between molecules, and other issues. The main theoretical basis in the quantum chemistry research is quantum mechanics, which describes the laws that govern operations in the microscopic world. SUMMARY OF THE INVENTION The Summary is provided to give a brief overview of concepts, which will be described in detail later in the section Detailed Description of Embodiments. The Summary is neither intended to identify key or necessary features of the claimed technical solutions, nor is it intended to be used to limit the scope of the claimed technical solutions. In a first aspect of the present disclosure, there is provided a data processing method for a quantum chemical system, the method may include the following steps: acquiring a specific wave function suitable for the quantum chemical system that is constructed based on a neural network; performing walker processing based on the specific wave function, the walker processing including diffusion of walkers; and determining related information about chemical properties of the quantum chemical system based on the specific wave function and the processed walkers. In a second aspect of the present disclosure, there is provided a data processing apparatus for a quantum chemical system. The apparatus may include: an acquisition unit configured to acquire a specific wave function suitable for the quantum chemical system that is constructed based on a neural network; a walker processing unit configured to perform walker processing based on the specific wave function, the walker processing including diffusion of walkers; and an information generation unit configured to determine related information about chemical properties of the quantum chemical system based on the specific wave function and the processed walkers. In a third aspect of the present disclosure, there is provided a data processing method for a quantum chemical system, the method may include the following steps: acquiring a specific wave function suitable for the quantum chemical system; performing walker processing based on the specific wave function, the walker processing including diffusion of walkers, and the diffusion of the walkers being performed such that the number of walkers remains constant during the processing; and determining related information about chemical properties of the quantum chemical system based on the specific wave function and the processed walkers. In a fourth aspect of the present disclosure, there is provided a data processing apparatus for a quantum chemical system, the apparatus may include an acquisition unit configured to acquire a specific wave function suitable for the quantum chemical system; a walker processing unit configured to performs walker processing based on the specific wave function, the walker processing including diffusion of walkers, and the diffusion of the walkers being performed such that the number of walkers remains constant during the processing; and an information generation unit that determines related information about chemical properties of the quantum chemical system based on the specific wave function and the processed walkers. In a fifth aspect of the present disclosure, there is provided an electronic device. The electronic device includes: a memory storing instructions; and a processor, wherein the instructions, when executed by the processor, cause the method according to any one of the embodiments of the present disclosure to be implemented. In a sixth aspect of the present disclosure, there is provided a computer-readable storage medium having a computer program stored thereon, wherein the program, when executed by a processor, causes the method according to any one of the embodiments of the present disclosure to be implemented. In a seventh aspect of the present disclosure, there is provided a computer program product including instructions that, when executed by a processor, cause the method according to any one of the embodiments of the present disclosure to be implemented. In an eighth aspect of the present disclosure, there is provided a computer program including program codes that, when executed by a processor, cause the method according to any one of the embodiments