>> designgui
Program Name | SDE Numerical Integrator | Pros | Cons |
---|---|---|---|
SSA | None -- Next Reaction variant of the Gillespie algorithm | Essentially Exact | Extremely slow for "large" systems with many molecules and fast reactions |
HyJCMSS-EM | Fixed time step Euler method | Much faster for "large" systems. Fastest SDE numerical integrator for non-stiff systems. | For stiff systems, species populations may go negative. Finding an accurate time step that produces a valid solution can be tedious. Error is proportional to the square root of the time step. |
HyJCMSS-Mil | Fixed time step Milstein method | Increased accuracy. Error is proportional to the time step, allowing one to use a larger time step. | Evaluation of 2D Ito integrals decreases the speed of the simulation. |
HyJCMSS-EM-Adaptive | Adaptive time step Euler method | Automatically chooses a time step based on the amount of error generated. | Does not always converge to the correct solution! Usage is not advisable, unless for educational purposes. |
HyJCMSS-Mil-Adaptive | Adaptive time step Milstein method | Automatically chooses a time step based on the amount of error generated. Increased simulation efficiency when transient stiffness exists. With a reasonable tolerance, convergence to the correct solution is gauranteed. | Slower than fixed methods for systems with constant timescales, due to the computational overhead in the adaptive code. |
Command Line Parameter | Description | In Which Methods | Default Value |
---|---|---|---|
Filename | NetCDF file name | All | None |
Random Seed | The random number generator's seed value/TD> | All | Randomized (system dependent, however!) |
Epsilon | Minimum number of molecules both reactant and product species required for approximation as a continuous Markov process | All HyJCMSS methods | 100 molecules |
Lambda | Minimum rate of reaction required for approximation to a continuous Markov process | All HyJCMSS methods | 10 molecules/second |
Multiple Slow Reaction (MSR) Tolerance | Maximum allowed effect of executing multiple slow reactions per numerical integration of the SDEs | All HyJCMSS methods | 1 / Epsilon |
SDE Tolerance | Maximum allowed value of the drift and diffusion errors | Adaptive Methods | 1e-4 |
SDE Time Step | The maximum time step of the SDE numerical integrator | Fixed Methods (but may also be set for Adaptive methods to decrease memory requirements | 0.10 seconds or the Save Time |
Maintained
by Howard Salis Kaznessis Research Group Dept. of Chemical Engineering & Materials Science University of Minnesota lastname@cems.umn.edu
(spam is evil)Site last modified at Wednesday, 25-Jan-2006 21:40:21 UTC |