Open dilpath opened 1 year ago
I agree that this is not the desired behavior. In case of errors, x and sx (and others) should be NaN. It seems there is a call to ReturnData::invalidate missing. Since rdata.status
is AMICI_ERROR
, it wouldn't consider this high priority, and would rather have the root cause fixed via #2101. (Volunteers?)
What did you expect to happen? Sensitivities to be computed
What has happened instead?
rdata.x
andrdata.sx
are zero matricesTo Reproduce Compile a model with algebraic expressions, e.g. SBML semantic test case 00039, simulate with and without computation of forward sensitivities.
AMICI version and system environment
develop
branch