Previous Page Parent Page Next Page TOC
User Manual | Methods | Time Course Calculation | Hybrid Simulation

Hybrid Simulation

Hybrid (RK-45)

Unlike the other hybrid methods, which use a dynamic partitioning strategy based on particle numbers, Hybrid (RK-45) relies on a user-defined partitioning strategy.

It is also the only hybrid method, with support for discrete events.

Options for Hybrid (RK-45)


Hybrid (Runge-Kutta)

The Hybrid (Runge-Kutta) simulation method combines deterministic numerical integration of ordinary differential equations (ODEs) with a stochastic simulation algorithm. The entire biochemical network is internally partitioned into a deterministic subnet and a stochastic subnet.

Users can specify which particle numbers are considered “low” or “high” using the Lower Limit and Upper Limit parameters. Species with particle numbers between these limits do not change their status, resulting in a hysteresis-like behavior. This avoids unnecessary switching when particle numbers fluctuate in the middle range.

Partitioning can be recalculated dynamically during the simulation, at intervals specified by the Partitioning Interval parameter. During each run, the deterministic and stochastic subnets are simulated in parallel. The deterministic part is integrated using a 4th-order Runge-Kutta method, while the stochastic part uses the Gibson and Bruck algorithm (Gibson00). Reaction probabilities in the stochastic subnet are approximated as constant during each stochastic step, even though they may actually change due to deterministic effects.

Note: The current Hybrid Runge-Kutta implementation is inefficient for models with assignment rules, causing significantly longer calculation times.

Options for Hybrid (Runge-Kutta)


Hybrid (LSODA)

The Hybrid (LSODA) method is similar to the Hybrid (Runge-Kutta) method but uses the LSODA algorithm for integration instead of 4th-order Runge-Kutta.

Options for Hybrid (LSODA)