The ARL Hierarchical MultiScale Framework (ARL-HMS) is a software library for development of multiscale models on heterogeneous high-performance computing systems. Multiscale models are created in ARL-HMS from an assembly of individual at-scale model components. A runtime system adaptively evaluates at-scale model components in parallel and performs data extraction necessary for scale-bridging. Adaptive machine learning methods are also available in ARL-HMS to significantly reduce computational expense.
The ARL-HMS software is implemented in C++ and includes bindings to Python. It can incorporate individual at-scale model components written in any programming language and can also incorporate closed-source or proprietary software allowing users to leverage existing complex at-scale models for multiscale model development.
A User's Guide to ARL-HMS is included in this software repository
(ARL-TR-9820.pdf). The User's Guide documents the steps necessary to
develop multiscale models with ARL-HMS. It includes an overview of the
mathematical approach to multiscale modeling the framework adopts, the
steps necessary to incorporate at-scale models, use of the ARL-HMS
Broker to perform at-scale model evaluation and data extraction, and
use of the adaptive surrogate modeling capabilities implemented in
ARL-HMS. A simple example ARL-HMS application, implemented in both C++
and Python, is included that performs Monte Carlo estimation of
ARL-HMS requires a compiled Boost C++ installation for data serialization and shared pointer capabilities.
Optional dependencies include a Message Passing Interface (MPI) library to enable MPI communication and fan-in data communication between ARL-HMS components. A Python installation is required to build the ARL-HMS Python bindings. The Adaptive Sampling Framework (ASF) library from LLNL (https://github.com/exmatex/aspa) is also required to use the Kriging database interpolation method.
ARL-HMS uses a standard autotools installation procedure (configure, make, make install) documented in Section 5 of the ARL-HMS User's Guide.
We request that if you publish results that make use of the ARL-HMS software, please cite the following paper and ARL-HMS User's Guide:
@article{Knap2016,
author = {Knap, J. and Spear, C. and Leiter, K. and Becker, R. and Powell, D.},
title = {A computational framework for scale-bridging in multi-scale simulations},
journal = {International Journal for Numerical Methods in Engineering},
volume = {108},
number = {13},
pages = {1649-1666},
doi = {https://doi.org/10.1002/nme.5270},
year = {2016}
}
@techreport{Leiter2023,
institution = {{DEVCOM} Army Research Laboratory},
title={User's Guide for the {Hierarchical MultiScale Framework} ({HMS})},
author={Leiter, Kenneth W and Crone, Joshua C and Knap, Jaroslaw},
year={2023},
month={Oct},
number={ARL-TR-9820}
}
ARL Hierarchical MultiScale Framework (ARL-HMS) is licensed under the Creative Commons Zero 1.0 Universal (CC0 1.0) license. Please see LICENSE for details.