S. Engblom's free software

Overview

• License
• Platforms
• Research-related codes:
• SimInf.
• URDME.
• FMM3D
• FMM2D
• BISDE
• FLOW
• FIBR
• Packages for daily use: stenglib

License

Platforms

Research-related codes

• SimInf is an R-package with the ability to simulate arbitrary models of epidemics using temporal contact data. The package has its own page.
  
  
  
  The SimInf manual is available via arXiv. The modeling and the simulation method is described in P. Bauer, S. Engblom, S. Widgren: Fast event-based epidemiological simulations on national scales in Int. J. High Perf. Comput. Appl., 2016: (doi).
  
  A sample output result is animated here.
  
  
• The URDME-package is an efficient implementation of the next subvolume method with the ability to sample the reaction-diffusion master equation in general unstructured geometries. The package has its own page.
  
  New! Version 1.4 is now released and can be downloaded from www.urdme.org. The manual is available via arXiv.
  
  A sample output result is animated here.
  
  The numerical modeling is described in S. Engblom, L. Ferm, A. Hellander, P. Lötstedt: Simulation of Stochastic Reaction-Diffusion Processes on Unstructured Meshes, in SIAM J. Sci. Comput. 31(3):1774--1797, 2009: (doi).
  
  There is also a software paper about URDME: B. Drawert, S. Engblom, and A. Hellander: URDME: a modular framework for stochastic simulation of reaction-transport processes in complex geometries, in BMC Syst. Biol. 6(76):1--17, 2012: (doi).
  
  
  
  Old! Version 1.0 is still available. This release is designed for use with Matlab 7.0 and above and Comsol Multiphysics 3.4. A manual in .pdf-format is included. Download the original stable version here: urdme.tar.gz.
  
  
• FMM3D, New! an adaptive distributed fast multipole method in 3D with a Matlab interface. Download here: daFMM3D.tar.
• FMM2D, an adaptive quite general fast multipole method in 2 dimensions with a nice Matlab-interface. Download here: FMM2D.tar. A relevant paper is S. Engblom: On well-separated sets and fast multipole methods, in Appl. Numer. Math. 61(10):1096--1102, 2011: (doi).
  
  There is also a GPU-version available (joint work with Anders Goude). It both provides support for NVIDIA's CUDA and a CPU-only version. It is highly configurable when it comes to compilation options and additionally, has support for (CPU only) high order panel potentials. Download here: FMM2Dgpu_small.tar.
  
  Previous version, for reproducibility: Download here: FMM2Dgpu_small_prev.tar. A larger .tar-file containing not only the source-code but also the data from our computational experiments is also available: FMM2Dgpu.tar (this file is large). We describe the GPU-implementation and perform various performance experiments in: A. Goude and S. Engblom: Adaptive fast multipole methods on the GPU, in J. Supercomput., 63(3):897--918, 2013: (doi).
  
  The GPU-code has bee ported into a stand-alone C++ interace by Artur Palha. This library is available via Bitbucket: libfmm.
  
  An experimental concurrent hybrid CPU/GPU-version using a task-based approach on the CPU is also available. Download here: FMM2Db.tar. This code has extended control of certain algorithmic parameters and can therefore be tuned for performance very easily. We describe the design of this code and test the feasability for autotuning in a paper: M. Holm, S. Engblom, A. Goude, S. Holmgren: Dynamic autotuning of adaptive fast multipole methods on hybrid multicore CPU & GPU systems, in SIAM J. Sci. Comput. 36(4):C376--C399 (2014): (doi).
  
  
  
  Old! There is also an older, no longer maintained uniform (non-adaptive) version available: FMM2Du.tar.
  
  Acknowledgment: The development of FMM2D and FMM3D has been partially supported by UPMARC, Uppsala Programming for Multicore Architectures Research Center.


• The BISDE-package is a small research code in Matlab and R for reproducing the computational experiments in the paper Bayesian inference in Epidemics: linear noise analysis by Samuel Bronstein, Stefan Engblom, and Robin Marin. Download the paper here: (arXiv) and the BISDE software here: BISDE.tar.


• The FLOW-package is a research code for accurately simulating surfactant laden interface flows. The method employed is a spectral Galerkin-Legendre discretization. The distributed experiments all use the time-stepper ode1s from Scicomp below. Download here: FLOW.tar.
  
  The relevant paper is S. Engblom, M. Do-Quang, G. Amberg, and A-K. Tornberg: On diffuse interface modeling and simulation of surfactants in two-phase fluid flow, in Commun. Comput. Phys. 14(4):879--915, 2013: (doi).


• The FIBR-package is a research code for simulating fibers in Stokes flow. Download the FIBR.tar-file into a fresh directory (eg. 'mkdir FIBR'), cd into it and untar (eg. 'untar' at the Matlab prompt, or 'tar -xf' at the Unix prompt). Run 'startup' to put some suitable directories on your path, then try the tests 'checkin' and 'spin'. Type 'help fibrsetup' and run the example there (it's the simplest possible). Then you might want to try the animation found by typing 'help advect'.
  
  A sample output result is animated here.

Packages for daily use ("stenglib")