You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Is your feature request related to a problem? Please describe.
At present, no validation of output values is done prior to writing to file. If a simulation runs into numerical issues that produce NaN or Infiinity values, these are written to the output file without getting caught.
Describe the solution you'd like
Before writing to file, step through all output values and check that they are finite values. This can be done using the CheckRealVar subroutine in the library to check that the value is not NaN and is finite.
Describe alternatives you've considered
We've simply ignored any possible NaN values in output files. Catching these at file writing will lead to more robust code as programmers are forced to ensure valid outputs, and will help catch model issues early.
This can also help catch uninitialized variables if testing similar to the procedure in #711 is done.
Additional context
This idea was prompted by an issue in the regression testing not catching NaN values (#711).
The text was updated successfully, but these errors were encountered:
Is your feature request related to a problem? Please describe.
At present, no validation of output values is done prior to writing to file. If a simulation runs into numerical issues that produce
NaN
orInfiinity
values, these are written to the output file without getting caught.Describe the solution you'd like
Before writing to file, step through all output values and check that they are finite values. This can be done using the
CheckRealVar
subroutine in the library to check that the value is notNaN
and is finite.Describe alternatives you've considered
We've simply ignored any possible NaN values in output files. Catching these at file writing will lead to more robust code as programmers are forced to ensure valid outputs, and will help catch model issues early.
This can also help catch uninitialized variables if testing similar to the procedure in #711 is done.
Additional context
This idea was prompted by an issue in the regression testing not catching
NaN
values (#711).The text was updated successfully, but these errors were encountered: