Molecular Force Field Multi-Objective Fitting Algorithm Library

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Current molecular force field parameter development tools are complex, often requiring multiple programming languages and software packages, with limited fitting targets and lack of modularity, making them difficult to use and develop, especially for multi-objective fitting functions.

Innovation Solution

A molecular force field multi-objective fitting algorithm library that includes modules for input/output, energy calculation, and thermodynamic property estimation, supporting user-defined multi-objective fitting and compatible with mainstream force field formats, enabling GPU operations and modular design for ease of use and scalability.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If multiple programming languages and software packages are used for force field parameter development, then the functionality and fitting targets are enhanced, but the device complexity and ease of operation deteriorate

Engineering Contradiction:

Improvefitting targetsVSAvoiddependencies

Core Design Contradiction:

Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The patent merges multiple programming languages (Python, C++, Fortran) and software packages (OpenMM, ParmEd, L-BFGS-B) into a unified force field optimization framework. The FFOptIterator class integrates these diverse components, allowing users to access complex functionality through a single Python interface without managing multiple dependencies manually.

Inventive Principle:

Principle #5Merging (Combining)

Solution Approach 2:

The patent introduces an intermediary layer (the FFOptIterator and EnergyCalculator classes) that mediates between the user and the complex underlying systems. This intermediary handles the complexity of multiple software packages and programming languages, presenting a simplified interface while maintaining access to diverse fitting targets and optimization algorithms.

Inventive Principle:

Principle #24Intermediary (Mediator)

2Adaptability or versatility

If command line tools with multiple software packages are used, then the fitting functionality is enhanced, but the ease of operation and installation deteriorate

Engineering Contradiction:

Improvefitting functionsVSAvoidinstallation and compilation

Core Design Contradiction:

Adaptability or versatilityVSEase of operation

Solution Approach 1:

The patent replaces the mechanical command-line interface system with a Python-based object-oriented framework. Instead of requiring users to manually construct command-line arguments and manage file formats across different software packages, the system provides Python classes (FFOptIterator, EnergyCalculator) that handle these operations programmatically and intuitively.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The patent creates a universal Python-based framework that can perform multiple fitting functions (energy minimization, molecular dynamics simulations, property calculations) through a single unified interface. This universal approach eliminates the need for separate command-line tools for each fitting task, simplifying both installation and operation.

Inventive Principle:

Principle #6Universality (Multi-functionality)

3Reliability

If detailed implementations are deeply bound to the process, then the implementation completeness is enhanced, but the modularity and ease of development deteriorate

Engineering Contradiction:

Improveimplementation completenessVSAvoidmodularity

Core Design Contradiction:

ReliabilityVSDevice complexity

Solution Approach 1:

The patent segments the force field optimization process into distinct modular components: FFOptIterator for controlling the optimization workflow, EnergyCalculator for computing energy and derivatives, and PropertyEstimator for calculating thermodynamic properties. Each module is independently implementable and can be combined flexibly to achieve complete implementation while maintaining code reusability and ease of development.

Inventive Principle:

Principle #1Segmentation

Data Source

PatentUS20220189586A1Molecular force field multi-objective fitting algorithm library and workflow

Publication Date: 2022.06.16 SHENZHEN JINGTAI TECH CO LTD

AI summary

The present invention provides a molecular force field multi-objective fitting algorithm library and the workflow, including: FFOptlterator as the main module for input and output and force field parameter training iteration; EnergyCalculator module used for MM energy and energy derivatives calculation for the required value of each iteration in the optimization algorithm; the PropertyEstimator module used for the thermodynamic property calculation based on the MD simulation. Wherein, when initializing the FFOptlterator and EnergyCalculator objects, the user specifies the training force field parameters, adjustable parameter ranges, system setting arguments, and MD simulation parameters. The invention is suitable for the related applications of molecular force field training and verification, and implement the framework for different training targets, the prediction of different molecular physical properties, the compatibility and conversion of force field parameters in different formats, integration of multi-objective optimization, result analysis and graphing, etc.