Initial conditions / png color scheme are scary

usnistgov / hiperc

High Performance Computing Strategies for Boundary Value Problems

https://pages.nist.gov/hiperc/en/latest/index.html

39 stars 7 forks source link

Initial conditions / png color scheme are scary #59

Closed amjokisaari closed 7 years ago

amjokisaari commented 7 years ago

It looks like the initial condition is two high-concentration lines on the edge of the computational domain, and concentration diffuses toward the center (see attached). This IC combined with the black-and-white color scheme made the first output image look like the code had errored. Perhaps choose a different color scheme (viridis?) and maybe consider a different initial condition. Will it serve your purposes to put a blob of mass in the center of the domain and let it diffuse outward? dat_ic

tkphd commented 7 years ago

Black-and-white is the One True Color Scheme for single-variable output. It's painful to extract quantitative values from Viridis.
A circular blob would work, but seems too simplistic. This half-wall fixed-value BC has been useful for testing corner cases (pun intended!) with tiled array access.
I can definitely do a better job explaining why I made these choices and what the user should expect from the code.

amjokisaari commented 7 years ago

What kind of quantitative data are you expecting people to extract from the image? Can they not do it with the csv instead?

edit: the IC makes sense, I figured it was something like that. Also, yes, the "what to expect" thing would definitely be nice... kind of like various people's complaints about the newest benchmark III problems.

amjokisaari commented 7 years ago

1/4 circle on the lower-left corner of the computational domain? ...this would allow examination of diffusion along non-axis/grid directions.

tkphd commented 7 years ago

I'm not sure what kind of data people would want to extract. The point is that they could, given grayscale data.
Since the CSV is plaintext, it uses a bit of disk space when the domain size increases. I only intend to write the final condition. Intervening checkpoints are images (compressed), only.
A quarter-circle in one corner, or a blob that disappears, won't stress the hardware enough. It has to be an "interesting" solution.

amjokisaari commented 7 years ago

ok, chalk my thoughts up to more "what to expect / justification" type doco i guess!

tkphd commented 7 years ago

OK. Thanks for questioning the initial conditions! I clearly have some 'splaining to do.