| Name | Modified | Size | Downloads / Week |
|---|---|---|---|
| Parent folder | |||
| README.md | 2026-10-26 | 2.0 kB | |
| v17.0.0 source code.tar.gz | 2026-10-26 | 4.0 MB | |
| v17.0.0 source code.zip | 2026-10-26 | 4.1 MB | |
| Totals: 3 Items | 8.1 MB | 0 | |
Catalyst v17.0.0
Breaking changes
Catalyst 17.0 updates to JumpProcesses 10, and requires ModelingToolkitBase 1.78 or later, the first release that supports it.
- The random number generator moves from
JumpProblemtosolve/init. PassingrngtoJumpProblem, or to aHybridProblemwith jumps, now throws anArgumentError. Writesolve(jprob, SSAStepper(); rng = StableRNG(1234))instead.solveandinitalso accept aseed. - Jumps draw their random numbers from the generator passed to
solve. Creating a new generator with the same seed for everysolvetherefore repeats one trajectory. To draw independent, reproducible samples, create one generator and pass it to everysolve. - A
JumpProblemsupports one solve at a time. Solves reuse the problem's jump state instead of copying it, and problems created withremakeshare it. For parallel simulations, use anEnsembleProblemor adeepcopyof the problem per thread or task. - Mass-action rate constants are recomputed from the current parameters at every
solve/initand everyreset_aggregated_jumps!. Callbacks written directly with DiffEqCallbacks or SciMLBase that change parameters must still callreset_aggregated_jumps!(integrator), whoseupdate_jump_paramskeyword is removed.jprob.massaction_jump.scaled_ratesno longer holds the rate constants.
See HISTORY.md (github.com) for details and examples, and the JumpProcesses [HISTORY.md](https://github.com/SciML/JumpProcesses.jl/blob/master/HISTORY.md#100 for the complete list of JumpProcesses 10 changes.
Merged pull requests:
- Support JumpProcesses 10 in tests, benchmarks, and docs (#1572) (@isaacsas)