Here's a PDF with the slides for my ICOPS 2017 talk entitled Validation and benchmarking of two particle-in-cell codes for a glow discharge (presentation "TH 1.1-3" in the conference program).
A follow-up comment to the talk was that "a code with known flaws is much better than a code with unknown flaws". I think that summarizes the topic of validation nicely. The purpose of validation isn't necessarily to pass with flying colors, but to learn for what cases to trust a code, and for which cases NOT to trust a code. If you find issues that are easy to fix, great! But even if you find issues that are not possible to fix immediately (or ever), validation can still be worthwhile.
The full paper can be downloaded from here.
No comments:
Post a Comment