C++ Implementation of the Tau Hybrid solver #557
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
- Creates platformutils.py, for putting platform-specific wrapper functions - Adds Popen wrapper (`open_simulation`) to "inject" platform-specific kwargs
- Extracts the `sim_kill` lambda into a separate helper function
- Adds new class in platformutils: `SimulationReader` - Handles reading the output of simulation in background thread - Optionally handles timeout
- Define signal handler info as macros - Use `INTERRUPT_*` macros as type definitions - Use `SET_INTERRUPT_HANDLER` as a wrapper around signal installer
gillespy2/solvers/cpp/c_base/tau_hybrid_cpp_solver/TauHybridSimulation.cpp
Show resolved
Hide resolved
gillespy2/solvers/cpp/c_base/tau_hybrid_cpp_solver/TauHybridSolver.h
Outdated
Show resolved
Hide resolved
gillespy2/solvers/cpp/c_base/tau_hybrid_cpp_solver/TauHybridSolver.cpp
Outdated
Show resolved
Hide resolved
ethangreen-dev
requested changes
Jul 2, 2021
ethangreen-dev
approved these changes
Jul 5, 2021
BryanRumsey
requested changes
Jul 5, 2021
There was a problem hiding this comment.
Simulations with multiple trajectories are generating constant/near constant results after the first trajectory.
Error occurs with the following models:
- Michaelis Menten
- Opiod
- Tyson 2 State Oscillator
NumPy Tau Hybrid Results for Michaelis Menten
Tau Hybrid C Results for Michaelis Menten
Toogle Switch model is returning strictly deterministic results when they should be stochastic
NumPy Tau Hybrid Results
Tau Hybrid C Solver Results
Specifying >1 for `number_of_trajectories` was resulting in constant output at best, integration errors at worst. - Supply simulation seed to prevent deterministic behavior - Back up `y0` and reset integrator completely on each trajectory - Explicitly initialize partition modes for each species
Species would often get "stuck" on deterministic output. This was due to propensities not being initialized on the first run. - Add propensity computation to beginning of solver loop
- Implement `validate` function to set integrator status based on CVODE return code - Stop trajectory if the timestep is too small
BryanRumsey
approved these changes
Jul 7, 2021
mdip226
approved these changes
Jul 7, 2021
- Fixes issue where fractional initial populations have unexpected results (i.e. Lotka-Volterra model)
|
Included update to ODE solver to use |
Merged
- Revert `SimulationState` enum properties to not use bit packing - Fix incorrect computation of standard deviation, which caused unexpected output - Fix CVODE error caused by uninitialized time value
- Limits the number of sequential invalid states from exceeding 1000 - Prevents "occasionally invalid" output from terminating the simulation - Allow `tau_tol` to be properly accepted as argument
seanebum
approved these changes
Jul 15, 2021
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Initial Implementation
This PR only includes the base C++ solver. None of the additional SBML features (rate rules, events, etc.) have been implemented, making this an MVP as it is not yet up to feature parity with the Python implementation. Each additional SBML feature will likely be included as its own PR.
TauHybridCSolverclass to Pythontau_hybrid_cpp_solverdirectory to C++ solver basetemplate_opts.himplementation to support solver-specific argumentsTauHybridCSolverto C++ solver tests, both variable and non-variableHybrid Data Structures
Extended data structures specific to the Tau Hybrid solver were added to C++. This is to prevent requiring hybrid-specific runtime state from being applied directly to the base
Species,Reaction, andModelclasses.Compile Times
As with the ODE C++ solver, the compile step is slow and may give the appearance of poor performance compared to its Python equivalent. The majority of the compile time is due to the Sundials compile dependency, and may be mitigated in the future by caching.
Species Mode Support
Assigning a discrete/stochastic mode value to a species is properly reflected by the C++ solver. This is achieved with the
template_opts.h"optional" header described above.