Analyze Crashes #99
Analyze Crashes #99
Conversation
mjohnson541
force-pushed
the
analyzecrash
branch
from
d442ffc to
6089e44
Compare
mjohnson541
force-pushed
the
analyzecrash
branch
from
c94b6fc to
39577c0
Compare
Codecov Report
@@ Coverage Diff @@
## master #99 +/- ##
==========================================
- Coverage 55.45% 53.90% -1.55%
==========================================
Files 25 26 +1
Lines 4660 4794 +134
==========================================
Hits 2584 2584
- Misses 2076 2210 +134
Continue to review full report at Codecov.
|
mjohnson541
force-pushed
the
analyzecrash
branch
from
39577c0 to
bf19fc7
Compare
|
@hwpang can you review this? It functions for single domain simulations, it's quite useful and other people have asked about getting access to it. I would've liked to clean up the collision limit analysis and add tests, but I don't really know when I might have time to devote to that...with the anniversary of the PR coming up I figure its better if we just get in what we have before this tool gets forgotten and never makes it in. It's certainly easier to upgrade this in the future than start from scratch. |
Specify StochasticDiffEq version to fix builds for now
these contain information related to the fluxes at the crash point and the model parameters associated with these objects
mjohnson541
force-pushed
the
analyzecrash
branch
from
bf19fc7 to
be01846
Compare
this function analyzes the rates and dydt to identify reactions and thermochemistry that seems to be contributing to the failure either through contributing NaNs or high fluxes relative to the rest of the simulation
Ideally this would produce comparable output to analyzecrash But for now it just prints out its own report
mjohnson541
force-pushed
the
analyzecrash
branch
from
be01846 to
b91b6a3
Compare
|
replaced by #174 |
This provides functionality for analyzing why a mechanism simulation crashed in terms of reactions, kinetics and thermochemistry. Also adds some functionality for analyzing collision limit violations, but only relative to H+H->H2 since sigma and epsilon often aren't available.