Skip to content

Conversation

@vayner-s
Copy link

@vayner-s vayner-s commented Apr 2, 2021

Set simulated ignition delay to nan instead of 0.0 when no ignition is found. This was done so that the datapoint is ignored when calculating the error function for the dataset, rather than having it become inf.

…nd. Changed corresponding warning message.
@kyleniemeyer
Copy link
Member

Thanks @vayner-s, for some reason it looks like the tests are failing with Python 3.6, due to a different average error value. Could the NaN behavior be affecting this somehow?

@kyleniemeyer
Copy link
Member

For reference, this closes #18

@kyleniemeyer
Copy link
Member

and closes #1

@vayner-s
Copy link
Author

vayner-s commented Apr 6, 2021

Yeah, I'm not sure because I tried running the example model locally with Python 3.6.13 and it seems like I'm getting the 58.882 value with and without the NaN change, so doesn't seem like it causes this.

rwest added a commit that referenced this pull request Apr 9, 2021
This unit test is failing on Travis only on Python 3.6 but not Python 3.5 on PR #20,
and we can't make sense of why. This is mostly a "no change" pull request to trigger 
the CI tests, but I added a print(output) statement to help debug.
@rwest
Copy link
Member

rwest commented Apr 9, 2021

I made Pull request #21 to try to see if (and why) the unit tests fail without this nan change.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

No ignition is causing infinite error Prevent infinite error function for simulated ignition delay of zero

3 participants