gillespied 0.3.0
Physical time reaction propagation library in D. Based on the Gillespie algorithm of 1977 it contains various implementations and enhancements for fast reaction index and timing sampling.
To use this package, run the following command in your project's root directory:
Manual usage
Put the following dependency into your project's dependences section:
README
This repository contains a physical time reaction propagation library in D. Based on the Gillespie algorithm of 1977 it contains various implementations and enhancements for fast reaction index and timing sampling.
The algorithm models the time and reaction evolution as stated by Gillespie, see [1] and [2]. After [3], all properly or rather "related to the intended master equation" formulated algorithms, modelling physical time propagation algorithms are equivalent. This concerns algorithms with rejection as well as the ones which are rejectionfree, as the current algorithm. The rejectionfree algorithms form a subset of the algorithms with rejection.
The default Gillespie algorithm was enhanced by two features.
If the reaction propensities are known, i. e. they do not have to be estimated, one of the random numbers needed by the original algorithm can be saved. As consequence the longer logarithmic operation is cancelled. See [4].
The search of next reaction is done over the cumulative sum of provided propensities. This search can be enhanced by using memory space. In this case, any of available search algorithms can be applied to the cumulative sum range, which is naturally ordered. In [5] the binary search algorithm was applied, whereas in the present case, the search policy is managed by the standard library.
[1] D. T. Gillespie, J. Comput. Phys. 434, 403 (1976). [2] D. T. Gillespie, 93555, 2340 (1977). [3] S. A. Serebrinsky, Phys. Rev. E  Stat. Nonlinear, Soft Matter Phys. 83, 2010 (2011). [4] W. Sandmann, Comput. Biol. Chem. J. 32, 292 (2008). [5] H. Li and L. R. Petzold, Tech. Rep. 1 (2006). (logarithmic direct method)
Example usage:
import gillespied;
import std.random : uniform, uniform01, rndGen;
import std.math : log, isNaN;
import std.range;
void main()
{
import std.stdio;
real[] inputPropensities = new real[uniform(1, ubyte.max)];
foreach(ref el; inputPropensities)
el =  uniform01!real.log;
auto algorithm = gillespieAlgorithm;
put(algorithm, inputPropensities);
assert(!algorithm.tau.isNaN);
assert(algorithm.tau != real.infinity);
assert(algorithm.index != inputPropensities.length);
}
Copyright: Copyright (c) 2019 Alexander Orlov. All rights reserved.
License: https://opensource.org/licenses/BSL1.0, BSL License
Author: Alexander Orlov, sascha.orlov@gmail.com
 Registered by Alexander Orlov
 0.3.0 released 2 years ago
 Sandman83/gillespied
 BSL1.0
 Copyright © 2019, Alexander Orlov
 Authors:
 Dependencies:
 mirrandom
 Versions:

0.3.3 2019Feb08 0.3.2 2019Feb08 0.3.1 2019Feb05 0.3.0 2019Feb05 0.2.0 2019Jan27  Download Stats:


0 downloads today

0 downloads this week

0 downloads this month

7 downloads total

 Score:
 0.8
 Short URL:
 gillespied.dub.pm