# US 2021/0241176 A1 · Method and device for searching new material > Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt - Number: US 2021/0241176 A1 - Document ID: US20210241176A1 - Status: Published application - Kind: A1 - Application number: 17/238,460 - Filed: 2021-04-23 - Published: 2021-08-05 - Earliest claimed: 2015-11-04 - Assignees: Samsung Electronics Co., Ltd. - Applicants: Samsung Electronics Co., Ltd. - Inventors: Jiho YOO; Youngchun KWON; Kyungdoc KIM; Jaikwang SHIN; Hyosug LEE; Younsuk CHOI - Web page: https://patentgrep.com/patent/US20210241176A1 ## CPC - G06N 20/00: Machine learning - G06F 16/245: Query processing - G06F 30/20: Design optimisation, verification or simulation (optimisation, verification or simulation of circuit designs G06F30/30) - G06N 7/01: Probabilistic graphical models, e.g. probabilistic networks - G16C 20/30: Prediction of properties of chemical compounds, compositions or mixtures - G16C 60/00: Computational materials science, i.e. ICT specially adapted for investigating the physical or chemical properties of materials or phenomena associated with their design, synthesis, processing, characterisation or utilisation - G16C 20/70: Machine learning, data mining or chemometrics - G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS - G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N) - G16C: COMPUTATIONAL CHEMISTRY; CHEMOINFORMATICS; COMPUTATIONAL MATERIALS SCIENCE ## Abstract A method for searching a new material includes: performing a learning on a material model, which is modeled based on a known material; determining a candidate material by inputting a targeted physical property to a result of the learning; and determining the new material from the candidate material. ## Claims ### Claim 1 A method for searching a material having a targeted physical property using a high-throughput screening, the method comprising: performing a learning on a material model which is modeled based on a known material; determining a candidate material by inputting the targeted physical property to a result of the learning; and determining the material having the targeted physical property from the candidate material, wherein the performing the learning on the material model comprises: modeling the material model based upon first structure information, which is information on a structure of the known material, first physical property information, which is information on a physical property of the known material, information on a potential factor, and a learning parameter, wherein the learning parameter comprises at least one of a default value parameter on the structure of the known material, a default value parameter on the physical property of the known material, a default value parameter of the potential factor, a relationship parameter between the structure of the known material and the potential factor, and a relationship parameter between the physical property of the known material and the potential factor. ### Claim 2 (depends on 1) The method of claim 1, wherein the performing the learning on the material model further comprises: updating the learning parameter; and determining whether to finish the learning based on an update result of the learning parameter. ### Claim 3 (depends on 2) The method of claim 2, wherein the first structure information comprises information on a partial structure in a molecular structure of the known material. ### Claim 4 (depends on 2) The method of claim 2, wherein the first structure information comprises information on atoms in the molecular structure of the known material and information on a combination of the atoms. ### Claim 5 (depends on 2) The method of claim 2, wherein the first structure information comprises two-dimensional image information on the molecular structure of the known material. ### Claim 6 (depends on 2) The method of claim 2, wherein the performing the learning on the material model further includes resetting the learning parameter with a random value before the modeling. ### Claim 7 (depends on 2) The method of claim 2, wherein the modeling the material model comprises: determining a probability function of the material model from an energy function based on the first structure information, the first physical property information, and the learning parameter; and generating second structure information, which is information on a structure of the material having the targeted physical property, and second physical property information, which is information on a physical property of the material having the targeted physical property, based on the probability function of the material model. ### Claim 8 (depends on 7) The method of claim 7, wherein the updating the learning parameter comprises: updating the learning parameter based on a free energy function of the material model deduced based on the probability function and the energy function. ### Claim 9 (depends on 8) The method of claim 8, wherein the updating the learning parameter based on the probability function and the free energy function comprises: updating the learning parameter by using a difference between the learning parameter and an objective function expressed as the free energy function. ### Claim 10 (depends on 9) The method of claim 9, wherein the objective function represents a function deduced from a differential expression of the probability function. ### Claim 11 (depends on 1) The method of claim 1, wherein the determining the material having the targeted physical property among the candidate material comprises: verifying at least one of physical validity, and chemical stability the candidate material; and determining the verified candidate material from the candidate material to be the advanced material having the targeted physical property. ### Claim 12 A device for searching a material as a device for searching an advance material having a targeted physical property using a high-throughput screening, the device comprising a processor, wherein the processor includes: a learning performer which performs a learning on a material model that is modeled based on a known material; and a material determiner which determines a candidate material by inputting a targeted physical property to a result of the learning, and determines a material having the targeted physical property from the candidate material, wherein the learning performer comprises: a material modeler which models the material model based on first structure information, which is information on a structure of the known material, first physical property information, which is information on a physical property of the known material, information on a potential factor, and a learning parameter, wherein the learning parameter comprises at least one of a default value parameter on the structure of the known material, a default value parameter on a physical property of the known material, a default value parameter of the potential factor, a relationship parameter between the structure of the known material and the potential factor, and a relationship parameter between the physical property of the known material and the potential factor. ### Claim 13 (depends on 12) The device of claim 12, wherein the learning performer further comprises: a parameter updater which updates the learning parameter; and a learning performance finisher which determines whether to finish the learning based on an update result of the learning parameter. ### Claim 14 (depends on 13) The device of claim 13, wherein the first structure information comprises information on a partial structure included in a molecular structure of the known material. ### Claim 15 (depends on 13) The device of claim 13, wherein the first structure information comprises information on atoms in the molecular structure of the known material and information on a combination of the atoms. ### Claim 16 (depends on 13) The device of claim 13, wherein the first structure information comprises two-dimensional image information on the molecular structure of the known material. ### Claim 17 (depends on 13) The device of claim 13, wherein the material modeler resets the learning parameter with a random value before the modeling. ### Claim 18 (depends on 13) The device of claim 13, wherein the material modeler determines a probability function of the material model from an energy function based on the first structure information, the first physical property information, and the learning parameter, and generates second structure information, which is information on a structure of the material having the targeted physical property, and second physical property information, which is information on a physical property of the material having the targeted physical property, based on the probability function of the material model. ### Claim 19 (depends on 18) The device of claim 18, wherein the parameter updater updates the learning parameter based on a free energy function of the material model deduced based on the probability function and the energy function. ### Claim 20 (depends on 19) The device of claim 19, wherein the parameter updater updates the learning parameter by using a difference between the learning parameter and an objective function expressed as the free energy function. ### Claim 21 (depends on 20) The device of claim 20, wherein the objective function represents a function deduced from a differential expression of the probability function. ### Claim 22 (depends on 12) The device of claim 12, wherein the material determiner verifies at least one of physical validity, and chemical stability of the candidate material, and determines the verified candidate material from the candidate material to be the material having the targeted physical property.