MOOSE (software)

Last updated
MOOSE
Original author(s) Derek Gaston
Developer(s) Idaho National Laboratory and contributors
Initial releaseJune 10, 2008;15 years ago (2008-06-10)
Operating system Linux, Mac OS X, Unix
Available in C++
Type Finite element analysis
License LGPL
Website mooseframework.inl.gov

MOOSE (Multiphysics Object Oriented Simulation Environment) is an object-oriented C++ finite element framework for the development of tightly coupled multiphysics solvers from Idaho National Laboratory. [1] MOOSE makes use of the PETSc non-linear solver package and libmesh to provide the finite element discretization.

Contents

A key design aspect of MOOSE is the decomposition of weak form residual equations into separate terms that are each represented by compute kernels. The combination of these kernels into complete residuals describing the problem to be solved is performed at run time. This allows modifications such as toggling of mechanisms and the addition of new physics without recompilation. MOOSE comes with an extensive library of kernels providing residual terms for solid mechanics, Navier–Stokes equations, phase-field models and more.

MOOSE uses VTK for visualisation.

Background

The development of MOOSE at Idaho National Laboratory (INL) since May 2008, has resulted in a unique approach to computational engineering that combines computer science with a strong underlying mathematical description in a unique way that allows scientists and engineers to develop engineering simulation tools in a fraction of the time previously required. [2] The heart of MOOSE is the Kernel. A Kernel is a "piece" of physics. To add new physics to an application built using MOOSE, all that is required is to supply a new Kernel that describes the discrete form of the equation. It's usually convenient to think of a Kernel as a mathematical operator, such as a Laplacian or a convection term in a partial differential equation (PDE). Kernels may be swapped or coupled together to achieve different application goals. These Kernels, which now number in the hundreds, allow a scientist or engineer to develop an application rapidly.

For a new application, existing Kernels are selected as-is, or modified as necessary, and "plugged" in. An advection-diffusion-reaction equation is of the same mathematical form no matter what application it is being used for. Typically, only the form of the coefficients or the dependencies on other physics need to be defined; rarely must complete Kernels be constructed from scratch. With MOOSE, only the Kernel development is required from the scientist or engineer (application developer). MOOSE is designed to do everything else for the application developer, such as finite element discretization of the PDEs, the nonlinear solver, and the parallel high performance computing.

The idea of naming MOOSE-based applications after Idaho indigenous animal species is loosely based on Los Alamos National Laboratory's 1970s – 1990s approach to naming their codes after Native American tribes and artifacts, such as APACHE, CONCHAS, and the KIVA series of codes. There are now over twenty MOOSE-based application animals in various stages of development, ranging from recently obtaining preliminary results to being nationally recognized as state-of-the-art efforts (such as BISON and MARMOT for fuels performance modeling & simulation activities).

Description

MOOSE / BISON simulation: A piece of a fuel pellet has chipped away (center left) due to a manufacturing defect or damage incurred while it was in transit. The damaged pellet surface induces a high-stress state in the adjacent cladding. As a result, the pellets warm up and densify before swelling back out due to fission products building up inside of them, further stressing the surrounding fuel cladding. Multiphysics Object-Oriented Simulation Environment (MOOSE).jpg
MOOSE / BISON simulation: A piece of a fuel pellet has chipped away (center left) due to a manufacturing defect or damage incurred while it was in transit. The damaged pellet surface induces a high-stress state in the adjacent cladding. As a result, the pellets warm up and densify before swelling back out due to fission products building up inside of them, further stressing the surrounding fuel cladding.

MOOSE is a development and run-time environment for the solution of multi-physics systems that involve multiple physical models or multiple simultaneous physical phenomena. The systems are generally represented (modeled) as a system of fully coupled nonlinear partial differential equation systems (an example of a multi-physics system is the thermal feedback effect upon neutronics cross-sections where the cross-sections are a function of the heat transfer). Inside MOOSE, the Jacobian-Free Newton Krylov (JFNK) method is implemented as a parallel nonlinear solver that naturally supports effective coupling between physics equation systems (or Kernels). [3] The physics Kernels are designed to contribute to the nonlinear residual, which is then minimized inside of MOOSE. MOOSE provides a comprehensive set of finite element support capabilities (libMesh) and provides for mesh adaptation and parallel execution. The framework heavily leverages software libraries from the Department of Energy (DOE) and the National Nuclear Security Administration (NNSA), such as the nonlinear solver capabilities in either the Portable, Extensible Toolkit for Scientific Computation (PETSc) project or the Trilinos project.

ELK (Extended Library of Kernels)

ELK is a library for common Kernels, boundary conditions and material base classes. [4]

YAK (Yet Another Kernel)

YAK is a library of common actions, Kernels, boundary conditions, and material base classes for radiation transport applications. YAK is currently linked with RattleSnake (multi-group radiation Sn transport), Pronghorn (multi-group diffusion), and Critter (multi-length scale temperature feedback).

Applications

BISON

BISON was the first MOOSE-based application "animal," and is a finite element-based nuclear fuel performance code applicable to a variety of fuel forms including light water reactor fuel rods, TRISO fuel particles, and metallic rod and plate fuel. [5] [6] [7] It solves the fully coupled equations of thermomechanics and species diffusion and includes important fuel physics such as fission gas release and material property degradation with burnup. BISON is based on the MOOSE framework [8] and can therefore efficiently solve problems on either two-dimensional axisymmetric or three-dimensional geometries using standard workstations or large high performance computers. Plasticity, irradiation growth, and thermal and irradiation creep models are implemented for clad materials. Models are also available to simulate gap heat transfer, mechanical contact, and the evolution of the gap/plenum pressure with plenum volume, gas temperature, and fission gas addition. BISON is also coupled to a MOOSE-based mesoscale phase field material property simulation capability. [9] [10]

MARMOT

MARMOT is a finite element-based phase field code for modeling irradiation-induced microstructure evolution. MARMOT predicts the effect of radiation damage on microstructure evolution, including void nucleation and growth, bubble growth, grain boundary migration, and gas diffusion and segregation. The phase field equations can be coupled with heat conduction and solid mechanics from ELK to consider the effect of temperature and stress gradients on the evolution. In addition, MARMOT calculates the effect of the microstructure evolution on various bulk material properties, including thermal conductivity and porosity. Once the bulk properties have been calculated, they can be passed to BISON for a fuel performance simulation. This coupling between MARMOT and BISON has been achieved in the hybrid code BARMOT. [11] [12] [13]

FALCON

FALCON is being developed to enable simulation of the tightly coupled fluid-rock behavior in hydrothermal and engineered geothermal system (EGS) reservoirs, targeting the dynamics of fracture stimulation, fluid flow, rock deformation, and heat transport in a single integrated code, with the ultimate goal of providing a tool that can be used to test the viability of EGS in the United States and worldwide. [14] [15] [16] Reliable reservoir performance predictions of EGS systems require accurate and robust modeling for the coupled thermal-hydrological-mechanical processes. Conventionally, these types of problems are solved using operator splitting methods, usually by coupling a subsurface flow and heat transport simulator with a solid mechanics simulator via input files. [17] [18] [19] [20] FALCON eliminates the need for using operator splitting methods to simulate these systems, and the scalability of MOOSE supported applications allows for simulating these tightly coupled processes at the reservoir scale, allowing for examination of the system as a whole (something the operator splitting methodologies generally cannot do). [21] [22]

RAT

ReActive Transport (RAT) has been developed to solve reactive transport problems in subsurface porous media that involves highly nonlinearly coupled physical processes of fluid flow, solute transport, biogeochemical reactions and media-solution interactions. These problems are common in various subsurface-engineered systems, such as engineered environmental remediation, enhanced geothermal systems and carbon dioxide geological sequestration. Currently, the physics that could be coupled in RAT include: single-phase fluid flow in porous media, advection, dispersion and diffusion transport, aqueous kinetic reaction, aqueous equilibrium reaction, kinetic mineral precipitation/dissolution reaction, and Carmen-Kozeny porosity-permeability relationship. [23] [24]

This software is not to be confused with the Reactor Analysis Tool [25] (RAT) which is a toolkit based on ROOT [26] and GEANT4 [27] for microphysical simulations of scintillation detectors used in neutrino and dark matter experiments including Braidwood, SNO+, and DEAP-3600.

RELAP-7

RELAP-7 is the next-generation tool in the RELAP safety/systems analysis application series and is based upon the MOOSE development and runtime environment framework. [28] [29] RELAP-7 will retain and improve the basic analysis capability of RELAP5. The four major improvements are 1) A well-posed seven-equation two-phase flow model (liquid, gas, and interface pressures) versus the obsolete ill-posed six-equation flow model (non-physical mixture sound speed) found in RELAP5; 2) Improved numerical approximations resulting in second-order accuracy in both space and time versus the first order approximations in RELAP5; 3) Implicit tightly coupled time integration for long duration transients, such as providing plant behavior for full life fuel cycle evaluations; and 4) the ability to easily couple to multi-dimensional core simulators being developed in other programs (NEAMS, CASL, ATR LEP). [30] [31] [32]

Pronghorn

Pronghorn was originally developed for simulation of the gas-cooled pebble-bed VHTR concept. The current capabilities of Pronghorn include transient and steady coupled porous fluid flow and solid-state heat conduction with a standard multi-group diffusion model (i.e., fixed-source, criticality, and time-dependent). [33] Recently added capabilities include a nonlinear acceleration scheme for criticality problems and a simple thermal-fluid model for the prismatic reactor concept. Future capabilities will include a more advanced multi-phase flow type of model (to study thermal boundary layer effects) and a radiation transport model. The physics can be solved in three-dimensional Cartesian (x, y, z) or cylindrical (r, q, z) space, with precursor and adiabatic thermal feedback models. This code has been validated against the PBMR400 benchmark problem. Using Pronghorn, rod ejection simulations have been conducted for thermal-fluids/neutronics for both pebble-bed and prismatic gas-cooled reactors and simple thermal-neutronics coupled LWR benchmark problems. [34] [35]

Features

Notes

  1. ↑ "MOOSE Framework - Open Source Multiphysics". Idaho National Laboratory. Retrieved 2013-03-28.
  2. ↑ [ dead link ]
  3. ↑ "Archived copy" (PDF). www.global-sci.com. Archived from the original (PDF) on 19 March 2013. Retrieved 17 January 2022.{{cite web}}: CS1 maint: archived copy as title (link)
  4. ↑ "MOOSE instructions - EVOWiki". Icme.hpc.msstate.edu. 2012-04-17. Archived from the original on 2016-03-05. Retrieved 2012-12-11.
  5. ↑ Williamson, R.L.; Hales, J.D.; Novascone, S.R.; Tonks, M.R.; Gaston, D.R.; Permann, C.J.; Andrs, D.; Martineau, R.C. (2012-01-19). "Journal of Nuclear Materials - Multidimensional multiphysics simulation of nuclear fuel behavior". Journal of Nuclear Materials. 423 (1–3): 149–163. Bibcode:2012JNuM..423..149W. doi:10.1016/j.jnucmat.2012.01.012.
  6. ↑ [ dead link ]
  7. ↑ Newman, Chris; Hansen, Glen; Gaston, Derek (2009). "Journal of Nuclear Materials - Three dimensional coupled simulation of thermomechanics, heat, and oxygen diffusion in UO2 nuclear fuel rods". Journal of Nuclear Materials. 392: 6–15. doi:10.1016/j.jnucmat.2009.03.035.
  8. ↑ Gaston, Derek; Newman, Chris; Hansen, Glen; Lebrun-Grandié, Damien (2009-07-19). "Nuclear Engineering and Design - MOOSE: A parallel computational framework for coupled systems of nonlinear equations" (PDF). Nuclear Engineering and Design. 239 (10): 1768–1778. doi:10.1016/j.nucengdes.2009.05.021.
  9. ↑ Tonks, Michael R.; Gaston, Derek; Millett, Paul C.; Andrs, David; Talbot, Paul (2011-08-28). "Computational Materials Science - An object-oriented finite element framework for multiphysics phase field simulations". Computational Materials Science. 51: 20–29. doi: 10.1016/j.commatsci.2011.07.028 .
  10. ↑ Tonks, Michael; Gaston, Derek; Permann, Cody; Millett, Paul; Hansen, Glen; Wolf, Dieter (2010-07-03). "Nuclear Engineering and Design - A coupling methodology for mesoscale-informed nuclear fuel performance codes". Nuclear Engineering and Design. 240 (10): 2877–2883. doi:10.1016/j.nucengdes.2010.06.005.
  11. ↑ Tonks, MR; Gaston, D; Millett, PC; Andrs, D; Talbot, P (2012). "An object-oriented finite element framework for multiphysics phase field simulations". Computational Materials Science. 51: 20–29. doi: 10.1016/j.commatsci.2011.07.028 .
  12. ↑ K Chockalingam, MR Tonks, JD Hales, DR Gaston, PC Millett, L Zhang. 2012. Crystal plasticity with Jacobian-Free Newton–Krylov. Computational Mechanics 0178-7675:1–10.
  13. ↑ Zhang, L; Tonks, MR; Millett, PC; Zhang, Y; Chockalingam, K; Biner, B (2012). "Phase-field modeling of temperature gradient driven pore migration coupling with thermal conduction". Computational Materials Science. 56: 161–165. doi:10.1016/j.commatsci.2012.01.002.
  14. ↑ "ESTSC - Find the latest in U.S. Department of Energy Software". Osti.gov. Archived from the original on 2015-10-22. Retrieved 2012-12-11.
  15. ↑ Podgorney, R.K, H. Huang, and D. Gaston, 2010, Massively Parallel Fully Coupled Modeling of Coupled Thermal-Hydro-Mechanical Processes for Enhanced Geothermal System Reservoirs, 35th Stanford Geothermal Workshop, Stanford, CA, February 2010
  16. ↑ Gaston, D., L. Guo, H. Huang, R. Johnson, H. Park, R. Podgorney, M. Tonks, and R, Williamson. 2010. Parallel algorithms and software for nuclear, energy, and environmental applications Part I: Multiphysics Algorithms, Communications in Computational Physics
  17. ↑ Gaston, D., L. Guo, H. Huang, R. Johnson, H. Park, R. Podgorney, M. Tonks, and R, Williamson. 2010. Parallel algorithms and software for nuclear, energy, and environmental applications Part II: Multiphysics Software, Communications in Computational Physics
  18. ↑ Podgorney, R.K, G. Gunnarsson, and H. Huang, 2011, Numerical Simulation of temperature dependent fluid reinjection behavior, Hellisheidi geothermal field, Southwest Iceland, Geothermal Resources Council 35th Annual Meeting, San Diego, CA, October 23–26, 2011
  19. ↑ Huang, H., P. Meakin, Podgorney, R.K, S. Deng, and C. Lu. 2011, Physics-based modeling of fracturing and permeability evolution in engineered geothermal systems, Geothermal Resources Council 35th Annual Meeting, San Diego, CA, October 23–26, 2011
  20. ↑ Podgorney, R.K, C. Lu, and H. Huang. 2012, Thermo-hydro-mechncial modeling of working fluid injection and thermal energy extraction in EGS fractures and rock matrix, 37th Stanford Geothermal Workshop, Stanford, CA, February 2012.
  21. ↑ Podgorney, R.K, H. Huang, and D. Gaston, 2010, FALCON: A Hybrid Finite Element-Discrete Element Physics Based Model for Simultaneously Solving Fully Coupled Multiphase Fluid Flow, Heat Transport, Rock Deformation, and Fracturing, Geothermal Resources Council 34th Annual Meeting, Sacramento, CA, October 24–27, 2010
  22. ↑ Podgorney, R.K, H. Huang, and D. Gaston, 2010, Physics-Based Modeling for Coupled Thermo-Hydro-Mechanical Problems in Geologic Media: An Application for Enhanced Geothermal System, Geological Society of America Annual Meeting, Denver CO, Oct 31-Nov. 3, 2010
  23. ↑ [ dead link ]
  24. ↑ Guo, L.; Huang, H.; Gaston, D.; Redden, G. D.; Fox, D. T.; Fujita, Y. (2010). "Reactive Transport Modeling of Induced Calcite Precipitation Reaction Fronts in". AGU Fall Meeting Abstracts. 2010. Adsabs.harvard.edu. Bibcode:2010AGUFM.H51C0911G.
  25. ↑ "Overview — RAT 1.0 documentation". rat.readthedocs.io.
  26. ↑ "ROOT a Data analysis Framework | ROOT a Data analysis Framework". root.cern.ch.
  27. ↑ "Overview | geant4.web.cern.ch". geant4.web.cern.ch.
  28. ↑ "NEAMS Update" (PDF). Ipd.anl.gov. April 2012. Archived from the original (PDF) on 2016-12-24. Retrieved 2016-09-27.
  29. ↑ [ dead link ]
  30. ↑ "NEAMS Update" (PDF). Ipd.anl.gov. January 2012. Archived from the original (PDF) on 2016-12-25. Retrieved 2016-09-27.
  31. ↑ [ dead link ]
  32. ↑ "Microsoft Word - Demonstration of a Steady State PWR Problem with RELAP7 r3.docx" (PDF). Inl.gov. Retrieved 2016-09-27.
  33. ↑ [ dead link ]
  34. ↑ H. Park, D. Gaston, S. Kadiouglu, D. Knoll, D. Lebrun-Grandie, R. Martineau and W. Taitano", "Tightly Coupled Multiphysics Simulation for Pebble Bed Reactors," American Nuclear Society 2009 International Conference on Advances in Mathematics, Computational Methods, and Reactor Physics, Saratoga Springs, NY, May 3–7, 2009.
  35. ↑ [ dead link ]

Related Research Articles

<span class="mw-page-title-main">Computational physics</span> Numerical simulations of physical problems via computers

Computational physics is the study and implementation of numerical analysis to solve problems in physics. Historically, computational physics was the first application of modern computers in science, and is now a subset of computational science. It is sometimes regarded as a subdiscipline of theoretical physics, but others consider it an intermediate branch between theoretical and experimental physics — an area of study which supplements both theory and experiment.

<span class="mw-page-title-main">Computational fluid dynamics</span> Analysis and solving of problems that involve fluid flows

Computational fluid dynamics (CFD) is a branch of fluid mechanics that uses numerical analysis and data structures to analyze and solve problems that involve fluid flows. Computers are used to perform the calculations required to simulate the free-stream flow of the fluid, and the interaction of the fluid with surfaces defined by boundary conditions. With high-speed supercomputers, better solutions can be achieved, and are often required to solve the largest and most complex problems. Ongoing research yields software that improves the accuracy and speed of complex simulation scenarios such as transonic or turbulent flows. Initial validation of such software is typically performed using experimental apparatus such as wind tunnels. In addition, previously performed analytical or empirical analysis of a particular problem can be used for comparison. A final validation is often performed using full-scale testing, such as flight tests.

<span class="mw-page-title-main">Finite-difference time-domain method</span> Numerical analysis technique

Finite-difference time-domain (FDTD) or Yee's method is a numerical analysis technique used for modeling computational electrodynamics. Since it is a time-domain method, FDTD solutions can cover a wide frequency range with a single simulation run, and treat nonlinear material properties in a natural way.

<span class="mw-page-title-main">Multiphysics simulation</span>

In computational modelling, multiphysics simulation is defined as the simultaneous simulation of different aspects of a physical system or systems and the interactions among them. For example, simultaneous simulation of the physical stress on an object, the temperature distribution of the object and the thermal expansion which leads to the variation of the stress and temperature distributions would be considered a multiphysics simulation. Multiphysics simulation is related to multiscale simulation, which is the simultaneous simulation of a single process on either multiple time or distance scales.

<span class="mw-page-title-main">COMSOL Multiphysics</span> Physics and engineering software package

COMSOL Multiphysics is a finite element analyzer, solver, and simulation software package for various physics and engineering applications, especially coupled phenomena and multiphysics. The software facilitates conventional physics-based user interfaces and coupled systems of partial differential equations (PDEs). COMSOL provides an IDE and unified workflow for electrical, mechanical, fluid, acoustics, and chemical applications.

Dassault Systèmes Simulia Corp. is a computer-aided engineering (CAE) vendor. Formerly known as Abaqus Inc. and previously Hibbitt, Karlsson & Sorensen, Inc., (HKS), the company was founded in 1978 by David Hibbitt, Bengt Karlsson and Paul Sorensen, and has its headquarters in Providence, Rhode Island.

<span class="mw-page-title-main">Computational engineering</span>

Computational Engineering is an emerging discipline that deals with the development and application of computational models for engineering, known as Computational Engineering Models or CEM. Computational engineering uses computers to solve engineering design problems important to a variety of industries. At this time, various different approaches are summarized under the term Computational Engineering, including using computational geometry and virtual design for engineering tasks, often coupled with a simulation-driven approach In Computational Engineering, algorithms solve mathematical and logical models that describe engineering challenges, sometimes coupled with some aspect of AI, specifically Reinforcement Learning.

<span class="mw-page-title-main">Abaqus</span> Software for finite element analysis

Abaqus FEA is a software suite for finite element analysis and computer-aided engineering, originally released in 1978. The name and logo of this software are based on the abacus calculation tool. The Abaqus product suite consists of five core software products:

  1. Abaqus/CAE, or "Complete Abaqus Environment". It is a software application used for both the modeling and analysis of mechanical components and assemblies (pre-processing) and visualizing the finite element analysis result. A subset of Abaqus/CAE including only the post-processing module can be launched independently in the Abaqus/Viewer product.
  2. Abaqus/Standard, a general-purpose Finite-Element analyzer that employs implicit integration scheme (traditional).
  3. Abaqus/Explicit, a special-purpose Finite-Element analyzer that employs explicit integration scheme to solve highly nonlinear systems with many complex contacts under transient loads.
  4. Abaqus/CFD, a Computational Fluid Dynamics software application which provides advanced computational fluid dynamics capabilities with extensive support for preprocessing and postprocessing provided in Abaqus/CAE - discontinued in Abaqus 2017 and further releases.
  5. Abaqus/Electromagnetic, a Computational electromagnetics software application which solves advanced computational electromagnetic problems.

A CFD-DEM model is suitable for the modeling or simulation of fluid-solids or fluid-particles systems. In a typical CFD-DEM model, the phase motion of discrete solids or particles is obtained by the Discrete Element Method (DEM) which applies Newton's laws of motion to every particle and the flow of continuum fluid is described by the local averaged Navier–Stokes equations that can be solved by the traditional Computational Fluid Dynamics (CFD). The model is first proposed by Tsuji et al. The interactions between the fluid phase and solids phase is better modeled according to Newton's third law.

<span class="mw-page-title-main">Coolfluid</span> Scientific computing environment

COOLFluiD is a component based scientific computing environment that handles high-performance computing problems with focus on complex computational fluid dynamics (CFD) involving multiphysics phenomena.

OOFEM is a free and open-source multi-physics finite element code with object oriented architecture. The aim of this project is to provide efficient and robust tool for FEM computations as well as to offer highly modular and extensible environment for development.

ADINA is a commercial engineering simulation software program that is developed and distributed worldwide by ADINA R & D, Inc. The company was founded in 1986 by Dr. Klaus-Jürgen Bathe, and is headquartered in Watertown, Massachusetts, United States. On April 7, 2022, Bentley Systems acquired ADINA R&D, Inc.

Model order reduction (MOR) is a technique for reducing the computational complexity of mathematical models in numerical simulations. As such it is closely related to the concept of metamodeling, with applications in all areas of mathematical modelling.

Goma is an open-source, parallel, and scalable multiphysics software package for modeling and simulation of real-life physical processes, with a basis in computational fluid dynamics for problems with evolving geometry. It solves problems in all branches of mechanics, including fluids, solids, and thermal analysis. Goma uses advanced numerical methods, focusing on the low-speed flow regime with coupled phenomena for manufacturing and performance applications. It also provides a flexible software development environment for specialty physics.

Nektar++ is a spectral/hp element framework designed to support the construction of efficient high-performance scalable solvers for a wide range of partial differential equations (PDE). The code is released as open-source under the MIT license. Although primarily driven by application-based research, it has been designed as a platform to support the development of novel numerical techniques in the area of high-order finite element methods.

<span class="mw-page-title-main">FEATool Multiphysics</span>

FEATool Multiphysics is a physics, finite element analysis (FEA), and partial differential equation (PDE) simulation toolbox. FEATool Multiphysics features the ability to model fully coupled heat transfer, fluid dynamics, chemical engineering, structural mechanics, fluid-structure interaction (FSI), electromagnetics, as well as user-defined and custom PDE problems in 1D, 2D (axisymmetry), or 3D, all within a graphical user interface (GUI) or optionally as script files. FEATool has been employed and used in academic research, teaching, and industrial engineering simulation contexts.

<span class="mw-page-title-main">Simcenter STAR-CCM+</span>

Simcenter STAR-CCM+ is a commercial Computational Fluid Dynamics (CFD) based simulation software developed by Siemens Digital Industries Software. Simcenter STAR-CCM+ allows the modeling and analysis of a range of engineering problems involving fluid flow, heat transfer, stress, particulate flow, electromagnetics and related phenomena.

<span class="mw-page-title-main">Physics-informed neural networks</span> Technique to solve partial differential equations

Physics-informed neural networks (PINNs) are a type of universal function approximators that can embed the knowledge of any physical laws that govern a given data-set in the learning process, and can be described by partial differential equations (PDEs). They overcome the low data availability of some biological and engineering systems that makes most state-of-the-art machine learning techniques lack robustness, rendering them ineffective in these scenarios. The prior knowledge of general physical laws acts in the training of neural networks (NNs) as a regularization agent that limits the space of admissible solutions, increasing the correctness of the function approximation. This way, embedding this prior information into a neural network results in enhancing the information content of the available data, facilitating the learning algorithm to capture the right solution and to generalize well even with a low amount of training examples.

References

This page is based on this Wikipedia article
Text is available under the CC BY-SA 4.0 license; additional terms may apply.
Images, videos and audio are available under their respective licenses.