CN117198417A - Stable crystal structure prediction method and system based on machine learning and target optimization - Google Patents
Stable crystal structure prediction method and system based on machine learning and target optimization Download PDFInfo
- Publication number
- CN117198417A CN117198417A CN202311343525.0A CN202311343525A CN117198417A CN 117198417 A CN117198417 A CN 117198417A CN 202311343525 A CN202311343525 A CN 202311343525A CN 117198417 A CN117198417 A CN 117198417A
- Authority
- CN
- China
- Prior art keywords
- crystal
- crystal structure
- primary
- target optimization
- stable
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 239000013078 crystal Substances 0.000 title claims abstract description 227
- 238000005457 optimization Methods 0.000 title claims abstract description 54
- 238000000034 method Methods 0.000 title claims abstract description 35
- 238000010801 machine learning Methods 0.000 title claims abstract description 19
- 230000006870 function Effects 0.000 claims abstract description 49
- 230000015572 biosynthetic process Effects 0.000 claims abstract description 39
- 239000000126 substance Substances 0.000 claims abstract description 27
- 238000010586 diagram Methods 0.000 claims abstract description 23
- 238000005381 potential energy Methods 0.000 claims abstract description 20
- 238000012216 screening Methods 0.000 claims abstract description 18
- 238000003062 neural network model Methods 0.000 claims description 17
- 238000000547 structure data Methods 0.000 claims description 17
- 238000012549 training Methods 0.000 claims description 16
- 238000004510 Lennard-Jones potential Methods 0.000 claims description 8
- 230000008569 process Effects 0.000 claims description 5
- 238000010200 validation analysis Methods 0.000 claims description 3
- 238000006664 bond formation reaction Methods 0.000 abstract 1
- 238000003775 Density Functional Theory Methods 0.000 description 8
- 238000004422 calculation algorithm Methods 0.000 description 8
- 239000000463 material Substances 0.000 description 8
- 238000004364 calculation method Methods 0.000 description 5
- 238000013135 deep learning Methods 0.000 description 5
- 238000005516 engineering process Methods 0.000 description 4
- 241000894007 species Species 0.000 description 3
- 239000013598 vector Substances 0.000 description 3
- 238000013528 artificial neural network Methods 0.000 description 2
- 238000004590 computer program Methods 0.000 description 2
- 230000003247 decreasing effect Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- 238000012795 verification Methods 0.000 description 2
- 241000270730 Alligator mississippiensis Species 0.000 description 1
- 241000238097 Callinectes sapidus Species 0.000 description 1
- 241001313846 Calypso Species 0.000 description 1
- 230000009471 action Effects 0.000 description 1
- 230000004075 alteration Effects 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 150000001875 compounds Chemical class 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000005398 intermolecular potential Methods 0.000 description 1
- 239000007777 multifunctional material Substances 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Abstract
The invention discloses a stable crystal structure prediction method based on machine learning and target optimization, and relates to the technical field of crystal structures. Comprising the following steps: obtaining chemical components of primary crystals; acquiring atomic coordinates of the primary crystal; acquiring a data set from a crystal database according to the atomic coordinates of the primary crystal; obtaining the formation energy of primary crystals; potential energy of the primary crystal is obtained by adopting a potential function; obtaining a contact diagram of the optimized crystal structure; and screening out a stable crystal structure according to the contact diagram of the optimized crystal structure and the bonding of atoms in the crystal structure. According to the invention, the energy prediction, structure search and atomic bond formation screening of the crystal are realized by running codes, so that a new stable structure can be obtained on the basis.
Description
Technical Field
The invention relates to the technical field of crystal structures, in particular to a stable crystal structure prediction method and system based on machine learning and target optimization.
Background
Crystal structure prediction is an important technology for finding new materials and understanding the nature of materials, and with the continuous improvement and perfection of theoretical methods and the rapid development of computer technology, it has become possible to find new materials using crystal structure prediction. The most mainstream method for predicting the crystal structure is to find the crystal structure meeting the requirements by taking the energy calculated by DFT as the stability measure according to the chemical components of the primary substances and combining with a search optimization algorithm. By this approach, a number of structure prediction software have been developed, such as CALYPSO, USPEX, AIRSS, GAtor, etc., which predicts stable or metastable crystal structures under the chemical composition of a given compound, and can be used to predict or determine clusters, two-dimensional layer structures, two-dimensional surfaces and three-dimensional crystal structures and design multifunctional materials. However, this method has some problems in the structure searching process: (1) The DFT calculation time cost is high, so that the time consumption for searching the crystal structure is high; (2) The stability of the crystal structure is affected by a number of factors, and the selection of a single energy as a measure of stability does not guarantee that a stable crystal structure is being searched for.
Disclosure of Invention
In order to solve the above-mentioned drawbacks in the background art, mainly, the stability of the crystal structure is affected by a plurality of factors, and selecting a single energy as a stability measure cannot guarantee that a stable crystal structure is searched. The invention provides a stable crystal structure prediction method and a system based on machine learning and target optimization.
To achieve the above object, a first aspect of the present invention provides a stable crystal structure prediction method based on machine learning and target optimization, including:
obtaining chemical components of primary crystals;
setting the space characteristics of the crystal according to the chemical components of the primary crystal, and acquiring the atomic coordinates of the primary crystal;
acquiring a data set from a crystal database according to the atomic coordinates of the primary crystal;
training the graph neural network model according to the acquired data set to acquire the formation energy of the primary crystal;
potential energy of the primary crystal is obtained by adopting a potential function;
setting a target optimization function according to the formation energy of the primary crystal and the potential energy of the primary crystal, and optimizing the primary crystal structure by using the target optimization function to obtain a contact diagram of the optimized crystal structure;
and screening out a stable crystal structure according to the contact diagram of the optimized crystal structure and the bonding of atoms in the crystal structure.
Preferably, the chemical composition of the primary crystal includes the atomic species and atomic number within the primary crystal.
Preferably, the data set comprises stable structure data and metastable structure data of crystal formation energy.
Preferably, the training set and the test set and the verification set are divided according to the proportion of 5:1 to train the graph neural network model.
Preferably, during the training of the graph neural network model, stable structure data and metastable structure data are taken as inputs, and the formation energy and band gap of the crystal are taken as outputs.
Preferably, the potential function is a Lennard-Jones potential function.
Preferably, the objective optimization function is as follows:
f＝ΔH+λ*|E|
wherein f is a target optimization function, delta H is formation energy, lambda is a weight coefficient, and the value range is 0-1; e| is the absolute value of the potential energy between two molecules.
Preferably, the spatial characteristics of the crystal include space group, symmetry, wyckoff position.
A second aspect of the present invention provides a stable crystal structure prediction system based on machine learning and target optimization, comprising:
the data acquisition module is used for acquiring chemical components of the primary crystal; setting the space characteristics of the crystal according to the chemical components of the primary crystal, and acquiring the atomic coordinates of the primary crystal; acquiring a data set from a crystal database according to the atomic coordinates of the primary crystal;
the crystal energy acquisition module is used for training the graph neural network model according to the acquired data set to acquire the formation energy of the primary crystal; potential energy of the primary crystal is obtained by adopting a potential function; setting a target optimization function according to the formation energy of the primary crystal and the potential energy of the primary crystal, and optimizing the primary crystal structure by using the target optimization function to obtain a contact diagram of the optimized crystal structure;
and the atomic bond screening module is used for screening out a stable crystal structure according to the optimized contact diagram of the crystal structure and the bond of atoms in the crystal structure.
Compared with the prior art, the invention has the beneficial effects that:
the invention provides a stable crystal structure prediction method and a system based on machine learning and target optimization, wherein the method can obtain a new stable structure on the basis of energy prediction, structure search and atomic bond screening of crystals by operating codes only by providing chemical composition and space group parameters of the predicted crystals. Compared with the traditional crystal structure prediction framework, the method uses the graph neural network to replace DFT calculation, improves the objective function of the optimization algorithm, and considers thermodynamic stability and kinetic stability. In addition, an atomic bonding screening module based on a contact diagram is added on the basis. By the method, the calculation time of DFT can be effectively reduced, and a new stable crystal structure can be obtained. The method is improved according to the mainstream crystal structure prediction flow, and the efficiency of structure calculation and screening is improved through deep learning and target optimization technology.
Drawings
FIG. 1 is a flow chart of a stable crystal structure prediction method based on deep learning and target optimization.
FIG. 2 is a BPO 4 Representation of crystals.
FIG. 3 is a graph showing the results of crystal formation prediction.
FIG. 4 is a result diagram of a crystal structure prediction method based on deep learning and target optimization.
Detailed Description
In order that those skilled in the art will better understand the technical solution of the present invention, the present invention will be further described with reference to the specific examples and the accompanying drawings, but the examples are not intended to be limiting.
The invention provides a stable crystal structure prediction method based on machine learning and target optimization, which is shown in fig. 1 and comprises the following steps:
s1, obtaining chemical components of primary crystals;
the chemical composition of the primary crystal includes the atomic species and atomic number within the primary crystal.
In this example, first, a primary chemical component is introduced, and the primary chemical component is the atomic species and the atomic number in a primary crystal.
S2, setting the spatial characteristics of the crystal according to the chemical components of the primary crystal, and acquiring the atomic coordinates of the primary crystal;
the spatial characteristics of the crystal include space group, symmetry, wyckoff position.
In this embodiment, the representation of the crystal structure features determines the space group, symmetry, wyckoff position of the crystal according to the chemical composition of the crystal, thereby obtaining the atomic coordinates of the random primary crystal;
the range of the space group is set according to the chemical composition of the introduced crystal, and the proper symmetry and Wyckoff position are selected to obtain the coordinates of atoms in the crystal structure. Referring to FIG. 2, it is shown as crystalline BPO 4 The crystal structure diagram, the crystal parameter information and the crystal contact diagram of the crystal structure diagram, wherein the crystal parameter information mainly comprises crystal composition, crystal lattice, atomic position and symmetry.
S3, acquiring a data set from a crystal database according to the atomic coordinates of the primary crystal;
the data set includes stable structure data and metastable structure data for crystal formation energy.
In this embodiment, the construction of the crystal dataset downloads stable and metastable structure data containing crystal formation energy from the crystal database, and the training set and the validation set are divided into the dataset according to a ratio of 5:1 to train the graph neural network model.
In the process of training the graph neural network model, stable structure data and metastable structure data are taken as input, and the formation energy and band gap of the crystal are taken as output. Specifically, the stable and metastable crystal structure data in 60000 pieces Materials Project are taken as the input of a model, and the data set mainly comprises crystal energy and band gap calculated by DFT; of these, 48000 data were used for training of the model, and the remaining data were used for testing and validation. Taking the formation energy and band gap of the crystal as the output of the model; the formation energy and band gap of the crystal are predicted by the trained graph neural network model.
And S4, training a graph neural network model according to the acquired data set so as to predict formation energy.
Training a graph neural network model according to the acquired data set to predict crystal formation energy, wherein fig. 3 shows the prediction performance of the graph neural network model on the formation energy, blue points are prediction values, a purple solid line is a fitting function of the prediction values, a red dotted line is a fitting function formed by using DFT calculation, the purple solid line and the red dotted line basically coincide, and a prediction result is close to a calculation result;
s5, potential energy of the primary crystal is obtained by adopting a potential function;
it should be noted that, prediction of material properties, the formation energy of primary crystals was predicted by training a graph neural network model through the constructed data set. And calculating potential functions, namely calculating potential energy corresponding to the selected crystal structure by using a Lennard-Jones potential function formula.
Adopting a potential function as a Lennard-Jones potential function, calculating a specific potential function, and calculating potential energy corresponding to the selected crystal structure by using a Lennard-Jones potential function formula;
where E is the potential energy between two molecules, ε represents the strength of the attraction, σ represents the effective diameter between the molecules, and r is the distance between the two molecules. The first term in the potential function (sigma/r ij ) 12 Represents the long-range attraction, gradually decreasing with increasing distance r between two molecules, while the second term (sigma/r ij ) 6 The repulsive force action of the short range is indicated to increase rapidly with decreasing distance r.
S6, setting a target optimization function according to the formation energy of the primary crystal and the potential energy of the primary crystal;
in this embodiment, search optimization is performed by setting an objective function, which includes forming an objective function capable of and a Lennard-Jones potential function. Modification of an objective function in an optimization algorithm, and optimized parameter setting.
The objective optimization function is as follows:
f＝ΔH+λ*|E| (2)
wherein f is a target optimization function, delta H is formation energy, lambda is a weight coefficient, and the value range is 0-1; i E is the absolute value of the potential energy between two molecules.
S7, performing iterative optimization on the primary crystal structure according to a set target optimization function;
according to formula (2), a Bayes optimization algorithm is used for multiple iterations to obtain a stable crystal structure which can be formed to be minimum and has intermolecular potential energy of 0. The primary crystal structure is iterated continuously by using a Bayes optimization algorithm, so that more and more stable crystal structures can be predicted.
S8, screening out a stable crystal structure according to the contact diagram of the optimized crystal structure and the bonding of atoms in the crystal structure. Referring to FIG. 4, cu is shown 2 As 3 Crystal structure diagram of (2)And a structure optimization process thereof. The red circles as in fig. 4 (a) represent the crystals searched during the 5000-step iterative optimization process, and the labels in fig. 4 (a) represent that the crystal formation energy obtained at 3702-th step is-2.13596, where the crystal formation energy is the lowest. Cu corresponding to 1245668 in the Materials Project material library of FIG. 4 (B) 2 As 3 Crystal structure diagram, FIG. 4 (B) shows Cu predicted by the method of the present invention 2 As 3 The crystal structure diagram is that the formation energy of the corresponding crystal is the lowest.
In this embodiment, the contact diagram of the optimized crystal structure is analyzed to determine the atom bonding condition of the searched crystal, and a more stable crystal structure is selected.
The method provided by the invention can obtain a new stable structure on the basis of energy prediction, structure search and atomic bond screening of the crystal by only providing the chemical composition and space group parameters of the predicted crystal by an operator and operating codes. Compared with the traditional crystal structure prediction framework, the scheme uses the graph neural network to replace DFT calculation, improves the objective function of the optimization algorithm, and considers thermodynamic stability and kinetic stability. In addition, an atomic bonding screening module based on a contact diagram is added on the basis. By the scheme, the calculation time of DFT can be effectively reduced, and a new stable crystal structure can be obtained.
The method is improved according to the mainstream crystal structure prediction flow, and the efficiency of structure calculation and screening is improved through deep learning and target optimization technology. The following is a flowchart of a stable crystal structure prediction method based on deep learning and target optimization with reference to fig. 1, which is described in detail:
(1) Firstly, a crystal formation prediction data set is constructed, the prediction accuracy of the crystal formation can be greatly affected by the data set, stable and metastable crystal structure data in 60000 Material projects are selected to train and verify a graph neural network model, 48000 crystal structure data are used for model training, and 12000 crystal structure data are used for verification. The dataset contains chemical composition parameters, lattice parameters and formation energies of the crystal structure in order to train a predictive model of the formation energies of the crystals.
(2) The crystal is built to form a predictive model that essentially includes a meget layer to update the matrices vi and ek, and two set2set layers learn a representation vector from the matrices vi and ek, respectively. These vectors are then combined using the concatate layer, and then the fully connected layer consisting of multiple dense layers is traversed to obtain the formation energy.
(3) The chemical composition of the predicted crystal is input, one symmetry is selected from 230 space groups by means of random screening, and then the lattice parameters are randomly generated according to the selected symmetry. In the same way, given the number of atoms and symmetry, the appropriate wyckoff position is selected. On this basis, the corresponding atomic coordinates are obtained by the lattice parameter and the wyckoff position.
(4) From the crystal parameters obtained in (3), the relative crystal structure using vectors to represent the crystal structure can be passed to the quantity v i And e k Respectively representing an atomic attribute and an atomic bond attribute, wherein i epsilon (1, the..N), k epsilon (1, the..M), N is the total number of atoms, and M is the total number of atom pairs.
(5) According to the input crystal parameters, a primary crystal structure can be obtained, the formation energy of the primary crystal structure is calculated through a crystal formation energy prediction model, and the potential energy of the primary crystal is calculated according to a Lennard-Jones potential function formula and is used as an objective function of an optimization algorithm.
(6) And continuously iterating by adopting an optimization algorithm to obtain a new crystal structure so as to form a function and a Lennard-Jones potential function to construct an objective function as a search measure, thereby obtaining the crystal structure meeting the requirement.
(7) Comparing the searched contact patterns of the crystal structures, analyzing the bonding condition of atoms in the crystal structures, removing the crystals which do not meet the requirement (namely, atoms in the crystals are not bonded), reserving the crystal structures with atoms bonded, updating the primary structure by optimizing iteration, and continuously searching for more stable crystals.
The invention provides a stable crystal structure prediction system based on machine learning and target optimization, which comprises the following components:
the data acquisition module is used for acquiring chemical components of the primary crystal; setting the space characteristics of the crystal according to the chemical components of the primary crystal, and acquiring the atomic coordinates of the primary crystal; acquiring a data set from a crystal database according to the atomic coordinates of the primary crystal;
the crystal energy acquisition module is used for training the graph neural network model according to the acquired data set to acquire the formation energy of the primary crystal; potential energy of the primary crystal is obtained by adopting a potential function; setting a target optimization function according to the formation energy of the primary crystal and the potential energy of the primary crystal, and optimizing the primary crystal structure by using the target optimization function to obtain a contact diagram of the optimized crystal structure;
and the atomic bond screening module is used for screening out a stable crystal structure according to the optimized contact diagram of the crystal structure and the bond of atoms in the crystal structure.
It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as a method, system, or computer program product. Thus, the present invention may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein. Although embodiments of the present invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made therein without departing from the principles and spirit of the invention, the scope of which is defined in the appended claims and their equivalents.
Claims (9)
1. A stable crystal structure prediction method based on machine learning and target optimization, comprising:
obtaining chemical components of primary crystals;
setting the space characteristics of the crystal according to the chemical components of the primary crystal, and acquiring the atomic coordinates of the primary crystal;
acquiring a data set from a crystal database according to the atomic coordinates of the primary crystal;
training the graph neural network model according to the acquired data set to acquire the formation energy of the primary crystal;
potential energy of the primary crystal is obtained by adopting a potential function;
setting a target optimization function according to the formation energy of the primary crystal and the potential energy of the primary crystal, and optimizing the primary crystal structure by using the target optimization function to obtain a contact diagram of the optimized crystal structure;
and screening out a stable crystal structure according to the contact diagram of the optimized crystal structure and the bonding of atoms in the crystal structure.
2. The machine learning and target optimization based stable crystal structure prediction method of claim 1 wherein the chemical composition of the primary crystal includes atomic species and atomic numbers within the primary crystal.
3. The machine learning and target optimization based stable crystal structure prediction method of claim 1 wherein the dataset includes stable structure data and metastable structure data for crystal formation energy.
4. A machine learning and target optimization based stable crystal structure prediction method according to claim 3 wherein the training set and the validation set are partitioned by a ratio of 5:1 into the dataset to train the graph neural network model.
5. The machine learning and target optimization based stable crystal structure prediction method according to claim 1, wherein stable structure data and metastable structure data are taken as input and crystal formation energy and band gap are taken as output in the process of training a graph neural network model.
6. The method for predicting stable crystal structure based on machine learning and target optimization of claim 1, wherein the potential function is a Lennard-Jones potential function.
7. The machine learning and target optimization based stable crystal structure prediction method of claim 1, wherein the target optimization function is as follows:
f＝ΔH+λ*|E|
wherein f n As the target optimization function, delta H is the formation energy, lambda is the weight coefficient, and the value range is 0-1]；
I E is the absolute value of the potential energy between two molecules.
8. The machine learning and target optimization based stable crystal structure prediction method of claim 1 wherein the spatial features of the crystal include spatial group, symmetry, wyckoff location.
9. A stable crystal structure prediction system based on machine learning and target optimization, comprising:
the data acquisition module is used for acquiring chemical components of the primary crystal; setting the space characteristics of the crystal according to the chemical components of the primary crystal, and acquiring the atomic coordinates of the primary crystal; acquiring a data set from a crystal database according to the atomic coordinates of the primary crystal;
the crystal energy acquisition module is used for training the graph neural network model according to the acquired data set to acquire the formation energy of the primary crystal; potential energy of the primary crystal is obtained by adopting a potential function; setting a target optimization function according to the formation energy of the primary crystal and the potential energy of the primary crystal, and optimizing the primary crystal structure by using the target optimization function to obtain a contact diagram of the optimized crystal structure;
and the atomic bond screening module is used for screening out a stable crystal structure according to the optimized contact diagram of the crystal structure and the bond of atoms in the crystal structure.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311343525.0A CN117198417A (en) | 2023-10-17 | 2023-10-17 | Stable crystal structure prediction method and system based on machine learning and target optimization |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311343525.0A CN117198417A (en) | 2023-10-17 | 2023-10-17 | Stable crystal structure prediction method and system based on machine learning and target optimization |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117198417A true CN117198417A (en) | 2023-12-08 |
Family
ID=88998111
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311343525.0A Pending CN117198417A (en) | 2023-10-17 | 2023-10-17 | Stable crystal structure prediction method and system based on machine learning and target optimization |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117198417A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117423396A (en) * | 2023-12-18 | 2024-01-19 | 烟台国工智能科技有限公司 | Crystal structure generation method and device based on diffusion model |
-
2023
- 2023-10-17 CN CN202311343525.0A patent/CN117198417A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117423396A (en) * | 2023-12-18 | 2024-01-19 | 烟台国工智能科技有限公司 | Crystal structure generation method and device based on diffusion model |
| CN117423396B (en) * | 2023-12-18 | 2024-03-08 | 烟台国工智能科技有限公司 | Crystal structure generation method and device based on diffusion model |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108985335B (en) | Integrated learning prediction method for irradiation swelling of nuclear reactor cladding material | |
| Nguyen et al. | Filter based backward elimination in wrapper based PSO for feature selection in classification | |
| Shah et al. | Learning DTW-shapelets for time-series classification | |
| US8301390B2 (en) | Quantum chemistry simulations using optimization methods | |
| CN117198417A (en) | Stable crystal structure prediction method and system based on machine learning and target optimization | |
| JP7411977B2 (en) | Machine learning support method and machine learning support device | |
| Meldgaard et al. | Generating stable molecules using imitation and reinforcement learning | |
| Masters et al. | Gps++: An optimised hybrid mpnn/transformer for molecular property prediction | |
| JP4887909B2 (en) | Simulation parameter determination method | |
| Coello | Evolutionary multiobjective optimization: current and future challenges | |
| Anstine et al. | Aimnet2: A neural network potential to meet your neutral, charged, organic, and elemental-organic needs | |
| Kumar et al. | Application of genetic algorithms for the prioritization of association rules | |
| CN109086568B (en) | Computer antibody combined mutation evolution system and method and information data processing terminal | |
| Park et al. | Inverse design of metal–organic frameworks for direct air capture of CO 2 via deep reinforcement learning | |
| Mendiburu et al. | Parallel EDAs to create multivariate calibration models for quantitative chemical applications | |
| CN112002380B (en) | Self-adaptive design method of high-generation heat energetic material based on machine learning | |
| KR101586386B1 (en) | A comprehensive molecular orbitals library having exclusive molecular orbital distributions, the method for estimating the molecular orbitals distributions region using the same and the system using the same | |
| Kim et al. | Optimal machine learning feature selection for assessing the mechanical properties of a zeolite framework | |
| Bonchev | My life–long journey in mathematical chemistry | |
| CN114997366A (en) | Protein structure model quality evaluation method based on graph neural network | |
| Erhart et al. | Constructing Local Bases for a Deep Variational Quantum Eigensolver for Molecular Systems | |
| Eremin et al. | Choice of the supercell with the optimum atomic configuration in simulation of disordered solid solutions | |
| CN114328221A (en) | Cross-project software defect prediction method and system based on feature and instance migration | |
| CN113689919A (en) | Method for predicting sub-state energy of organic chemical molecules based on BP artificial neural network | |
| CN113722853B (en) | Group intelligent evolutionary engineering design constraint optimization method for intelligent computation |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |