For a system in a given state, Gillespie's direct algorithm asks two questions:
Which reaction occurs next?
When does it occur?
Both of these questions must be answered probabilistically.
The probability that a certain reaction μ will take place in the next instant of time dt is given by:
(for example)
So we can specify the probability density P( μ ,τ) that the next reaction is μ and it occurs at time τ.
It can be shown that
![]()
This equation leads directly to the answers of the two aforementioned questions.
First, what is the probability distribution for reactions? Integrating P( μ ,τ) over all τ from 0 to ∞ results in:
*
Second, what is the probability distribution for times? Summing P( μ ,τ) over all μ results in
**
These two distributions lead to Gillespie's direct algorithm:
Exact Stochastic Simulation - Gillespie's Direct Method:
1. Initialize (i.e., set initial numbers of molecules, set t = 0, set stopping criteria e.g. maximum time or minimum reactants numbers ) .
2. Calculate the propensity function, ai, for all i.
3. Choose μ according to the distribution in eq *.
4. Choose t according to an exponential with parameter
(as in eq **).
5. Change the number of molecules to reflect execution of reaction μ . Set t = t + τ .
6. If stopping criteria are not met, go to Step 2.
'Research > Etc_Research' 카테고리의 다른 글
| Monte Carlo Method (0) | 2007.01.08 |
|---|---|
| Gradient (0) | 2007.01.08 |
| Partial differential equations (0) | 2007.01.08 |
| Matlab Integral 명령어 (1) | 2007.01.08 |
| 산화환원 반응 (0) | 2007.01.07 |