search

SciML / JumpProcesses.jl

Build and simulate jump equations like Gillespie simulations and jump diffusions with constant and state-dependent rates and mix with differential equations and scientific machine learning (SciML)
https://docs.sciml.ai/JumpProcesses/stable/
Other
136 stars 35 forks source link

Eliminate small allocations in RSSA/RSSACR #317

Closed isaacsas closed 1 year ago

isaacsas commented 1 year ago

I noticed after the recent PRs that RSSA/RSSACR still show number of time steps dependent allocations. After some profiling I believe this is because they call rate functions with views from a rate bound matrix for the upper and lower bounds on u, when the function wrappers are concretely typed based on the type of u. So we need to switch to storing the upper and lower bounds in elements of the same type as u in these steppers.

isaacsas commented 1 year ago

Coevolve needs to be checked too.

More generally the allocations test should just include every aggregator.