:orphan: .. status publishable .. product Forcefield Optimizer .. sectionauthor Clive .. SME Clive .. PR Erich .. TW Ken .. date 2018 .. _ffoDTS: |medea| |ffo| ---------------------------- .. admonition:: **At-a-Glance** The |medea|\ :sup:`®`\ [#TM]_ *Forcefield Optimizer* enables you to determine optimum forcefield parameters for energy minimization (EM), molecular dynamics (MD), and Monte Carlo (MC) simulations. The |medea| |ffo| adjusts forcefield parameters to reproduce *ab initio* quantum mechanical data. With the |medea| |ffo|, enable forcefield based simulation techniques to compute properties that require extensive sampling of configurational space. Then, carry out simulations that employ accurate forcefields with LAMMPS and GIBBS to compute the properties and behavior of systems, even those comprised of large numbers of atoms, for many configurations. This capability significantly expands the range of properties that can be obtained using simulation, to large scale cooperative atomic motions and diffusive properties, for example. **Key Benefits** Productivity * Extends *ab initio* results to larger length and time scales * Automates forcefield fitting * Searches parameter space using evolutionary methods Accuracy * Validates parameter quality using standard least squares methods * Provides access to all calculation details and information * Analyses fit quality automatically with generated reports and plots Forcefields employ simple mathematical expressions and carefully optimized numerical parameters to describe the energetic behavior of atomic and molecular systems. The |medea| |ffo| allows you to compute forcefield parameters by accurately reproducing supplied target first-principles data for systems of interest. The |medea| |ffo| employs first-principles derived information as the desired target behavior which is to be reproduced, using the selected forcefield. Configuration dependent energy, force, and stress data can be employed in the fitting process. The |medea| |ffo| has been developed in active research and development projects over a number of years and has been thoroughly validated. :myquote:`...force fields will improve as their application in chemistry and molecular biology will advance, and that Darwinian natural selection and survival of the fittest will lead to better force fields which will be gradually accepted according to agreed-upon standards.` :myquotee:`Shneior Lifson, Forcefield Pioneer, in 1981` The desired target data for a given system are collected using a |medea| structure list. This compendium of information can contain configurations obtained from high temperature *ab initio* molecular dynamics simulations. These configurations sample the energy surface of the desired system, and the |medea| |ffo| adjusts all or selected forcefield parameters to optimally reproduce the supplied target information. Supply weights for the supplied data types. Specify whether energies are handled in a relative or absolute manner (which is useful for some forcefield forms such as the embedded atom method). An interactive user interface allows you to select the parameters included in the fitted procedure and to set up any desired bound limits for parameters. An evolutionary algorithm can be employed to search the forcefield parameter space thoroughly for any given system. Control the parameters of search using a variety of options, including the population size and the number of generations employed by the genetic algorithm. Similarly, least squares optimization algorithm parameters, such as the maximum number of least squares cycles, can be adjusted for specific system needs. The |medea| |ffo| provides considerable analytical output, including automated graphical analysis of the degree of fit of the optimized forcefield and supplied target information. Forcefield optimization is often an iterative process, as the appropriate compromises are determined to efficiently account for a broad range of desired targets. In order to support the entire process of forcefield optimization the |medea| |ffo| supports the use of validation and training set input, and retains all input and output (including forcefield files which are entirely open and accessible), using the |medea| |jobserver| system. .. figure:: /Datasheets/images/FFO1.png :align: center Comparison of target and forcefield derived forces. A range of comparison plots like this are automatically generated by the |medea| |ffo| to allow the rapid assessment of optimized forcefield parameter sets. Technical Features ^^^^^^^^^^^^^^^^^^ User Interface ************** * Selection of training data * Selection of validation data * Specification of terms for optimization * Handling of relative and absolute energies * Selection of evolutionary and least squares parameters * Report and plot creation for analysis Supported Target Data ********************* * Energies * Forces * Stress tensors Supported Forcefield Types ************************** * PCFF+ (class II forcefields) * Buckingham [#Asahi01]_ * Morse with Coulomb terms * EAM (embedded atom method) [#Christensen01]_\ :sup:`,`\ [#Christensen02]_ * Tersoff (three body + two body) [#Tersoff]_ * REBO [#Humbird]_ * Stillinger-Weber [#Stillinger]_ * COMB3 (Charge-Optimized Many-Body) [#FranceLanord01]_ Key Features ^^^^^^^^^^^^ * Uses VASP-derived DFT results * Interactive selection and control * Automated results analysis * Efficient handling of optimization Required Modules ^^^^^^^^^^^^^^^^ * |medea| |menvironment| * |medea| |ffo| * |medea| |mlammps| |mstdenv| * |medea| |mvasp| Find Out More ^^^^^^^^^^^^^ Learn more about the |medea| |ffo| by watching the webinar |webinarclassforcefield|. .. [#TM] |regTMinfo| .. [#Asahi01] R. Asahi *et al.*, *Modelling Simul. Mater. Sci. Eng.* **22**, 075009 (2014) (`DOI `__) .. [#Christensen01] M. Christensen *et al.*, *J. Nucl. Mater.* **445**, 241 (2014) (`DOI `__) .. [#Tersoff] J. Tersoff, *Phys. Rev. B* **37**, 6991 (1988) (`DOI `__) .. full author list: D. Humbird and D. B. Graves .. [#Humbird] D. Humbird *et al.*, *J. Chem. Phys.* **120**, 2405 (2004) (`DOI `__) .. full author list: F. H. Stillinger and T. A. Weber .. [#Stillinger] F. H. Stillinger *et al.*, *Phys. Rev. B* **31**, 5262 (1985) (`DOI `__) .. [#FranceLanord01] A. France-Lanord *et al.*, *J. Chem. Phys.* **144**, 104705 (2016) (`DOI `__) .. [#Christensen02] M. Christensen *et al.*, *Integr. Mater. Manuf. Innov.* **6**, 92 (2017) (`DOI `__) .. only:: html :download: :download:`pdf `