The system includes a material property prediction presenting unit, a cross-task compatible feature value generating unit, and a material property predicting unit. The material property prediction presenting unit accepts a specification of first task data that includes a record in which a material property is unknown and is to be a target of material property prediction through a first predictive model. The cross-task compatible feature value generating unit predicts feature values from material compositions in the first task data by using a second predictive model. The material property predicting unit generates the first predictive model by using the material compositions, experimental condition, feature values, and the known material property in the first task data. Also, the material property predicting unit inputs the material composition, experimental condition, and feature value in a record in which the material property is unknown in the first task data and predicts the unknown material property.

In some embodiments, a method includes receiving a set of target attributes associated with a chemical product formulation and a set of priority values of the plurality of target attributes. The method includes determining, based on (1) a first neural network, (2) the set of target attributes and (3) the set of priority values, a set of sample formulations. The method includes determining a set of scores based on the set of sample formulations. The method includes selecting, based on the set of scores and the set of target attributes, a sample formulation from the set of sample formulations having a score greater than remaining scores from the set of scores. The method includes determining an origin associated with the sample formulation. When the origin is included in a pre-determined group, the method includes generating a report including the sample formulation as the chemical product formulation.

Methods, systems, and apparatus, including computer programs encoded on computer storage media, for simulating a concrete mixture. One of the methods includes obtaining an optical characterization of physical particles, generating a multispherical approximation of the physical particles, the multispherical approximation having reduced dimensionality compared to the optical characterization, simulating an aggregate mixture by applying the multispherical approximation of the particles to a physics simulator to obtain a predicted performance of the proposed aggregate mixture, selectively altering the aggregate mixture based on a comparison with performance metrics and simulating the altered aggregate mixture until the predicted performance satisfies the performance metrics to obtain a final aggregate mixture, and outputting the final aggregate mixture

Methods, systems, and apparatus, including computer programs encoded on computer storage media, for simulating a concrete mixture. One of the methods includes obtaining an optical characterization of physical particles, generating a multispherical approximation of the physical particles, the multispherical approximation having reduced dimensionality compared to the optical characterization, simulating an aggregate mixture by applying the multispherical approximation of the particles to a physics simulator to obtain a predicted performance of the proposed aggregate mixture, selectively altering the aggregate mixture based on a comparison with performance metrics and simulating the altered aggregate mixture until the predicted performance satisfies the performance metrics to obtain a final aggregate mixture, and outputting the final aggregate mixture

A computational method for training a neural network force field (NNFF) configured to simulate molecular and/or atomic motion within a material system. The method includes the step of receiving molecular structure data of a molecule in the material system. The method also includes optimizing a geometry of the molecule using the molecular structure data and a density functional theory (DFT) simulation to obtain DFT optimized geometry data. The method further includes optimizing the geometry of the molecule using the molecular structure data and a classical force field (FF) simulation to obtain FF optimized geometry data. The method also includes outputting NNFF training data comprised of the DFT optimized geometry data and the FF optimized geometry data. The NNFF training data is configured to train an NNFF for simulating molecular and/or atomic molecular and/or atomic motion within the material system.

A computational method for training a neural network force field (NNFF) configured to simulate molecular and/or atomic motion within a material system. The method includes the step of receiving molecular structure data of a molecule in the material system. The method also includes optimizing a geometry of the molecule using the molecular structure data and a density functional theory (DFT) simulation to obtain DFT optimized geometry data. The method further includes optimizing the geometry of the molecule using the molecular structure data and a classical force field (FF) simulation to obtain FF optimized geometry data. The method also includes outputting NNFF training data comprised of the DFT optimized geometry data and the FF optimized geometry data. The NNFF training data is configured to train an NNFF for simulating molecular and/or atomic molecular and/or atomic motion within the material system.

Disclosed are a method and apparatus of predicting properties of feed and products in a reformer. The method of predicting properties of feed and products in a reformer includes training a first predictive model for predicting the properties of feed in the reformer and a second predictive model for predicting the properties of products in the reformer; predicting the properties of feed being currently supplied to the reactor in real time by allowing a first prediction unit including the trained first prediction model to receive a current operating condition of the reactor in the reformer; and predicting the properties of products being produced in the reactor in real time by allowing a second prediction unit including the trained second prediction model to receive the current operating condition and the predicted properties of feed.

An embodiment includes a materials development support apparatus including an input data acquisition device configured to acquire input data including a material of a base forming a thin film and a function of the thin film, a candidate data generator configured to provide a preset verification target material as an input to a first learning, output a plurality of candidates for a function provided by the verification target material, an inverse analyzer configured to select a material that provides the function of the thin film included in the input data from the plurality of candidates for the function included in the candidate data, provide the material of the base included in the input data and the selected material as inputs to a second learning model, output a candidate for structure of the thin film, and a presenter configured to present the candidate for the structure of the thin film output.

An embodiment includes a materials development support apparatus including an input data acquisition device configured to acquire input data including a material of a base forming a thin film and a function of the thin film, a candidate data generator configured to provide a preset verification target material as an input to a first learning, output a plurality of candidates for a function provided by the verification target material, an inverse analyzer configured to select a material that provides the function of the thin film included in the input data from the plurality of candidates for the function included in the candidate data, provide the material of the base included in the input data and the selected material as inputs to a second learning model, output a candidate for structure of the thin film, and a presenter configured to present the candidate for the structure of the thin film output.

Provided are a device and method for predicting product properties of a naphtha splitting unit (NSU). The method includes training a prediction model for predicting the product properties of the NSU, inputting an input variable to the trained prediction model to acquire a prediction value for each output variable, and outputting the acquired prediction values for the output variables.