Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Chunksize manual benchmarking notebook for equilibrium spectra added #12

Open
wants to merge 2 commits into
base: master
Choose a base branch
from

Conversation

sagarchotalia
Copy link

Added a notebook for benchmarking of chunksize vs. non-chunksize computations, plotting the results.

Changes Made

  • Computed and compared an NO equilibrium spectrum, with ~53k lines.
  • Calculated and showed residual, plot_diff
  • Computed the same spectra multiple times for statistical fluctuations, comparing it in the end using the no-chunksize spectrum as a baseline.

Future Objectives

  • Add the comparison with chunksize = "auto" feature, once it is implemented (see the Chunksize PR)
  • Add non-equilibrium spectra calculations in the same benchmark.

@erwanp
Copy link
Member

erwanp commented Aug 1, 2022

Saw your notebook, great! https://github.com/sagarchotalia/radis-benchmark/blob/master/manual_benchmarks/chunksize_benchmark.ipynb

Can you somehow output the number of chunks ?
I see there are only ~50,000 lines in the NO example, which is a small number, so it's possible that beyond chunksize=1e5 we're having all lines in a single chunk.

@sagarchotalia
Copy link
Author

Yes, so I included the "N" calculation parameter, here are the results:

  • For the case of the molecule NO, the temporary chunk dataframe dg has a size of 1 until chunksize = 1e6, since the N parameter gets larger than df1.
  • Starting at chunksize = 1e6, the sizes of the chunks in memory are:
    • 8, 83, 828, and 7572, corresponding to 1e6, 1e7, 1e8 and 1e9 respectively. Starting 1e10, the value of N becomes 1, hence the entire dataframe is considered at once.

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.

2 participants