fix: need max_step in solve_ivp to get accurate dry density profile like in PySDM example

[yoctoyotta1024]

see: https://github.com/open-atmos/PySDM/blob/e22f01ea1f4a5272965b0867e7104f1fc5dd101b/examples/PySDM_examples/Shipway_and_Hill_2012/settings.py#L142

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 98.39%. Comparing base (3bafb8b) to head (678cc30).

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #649      +/-   ##
==========================================
+ Coverage   92.93%   98.39%   +5.46%     
==========================================
  Files          40        9      -31     
  Lines        1754      187    -1567     
==========================================
- Hits         1630      184    -1446     
+ Misses        124        3     -121     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[slayoo]

thanks @yoctoyotta1024 !
how does the problem manifests? Would be great to add a test that fails without this change.

[yoctoyotta1024]

thanks @yoctoyotta1024 ! how does the problem manifests? Would be great to add a test that fails without this change.

Hard to say what the test could be, but the issue is that without max_step the dry density profile is slightly different compared to the PySDM Shipway_and_Hill_2012 example (see plot here). This matters for the pressure and density and therefore condensation in the example (because they're so sensitive to the density profile).

(note mislabeled units should read kg/m^3)

[slayoo]

Thanks! The code generating this plot is a perfect test! Please add it into this PR, and we can work to shape it into a unit test.

[yoctoyotta1024]

Thanks! The code generating this plot is a perfect test! Please add it into this PR, and we can work to shape it into a unit test.

Oh the code I ran to make the plot is very dirty and actually an extract from my superdrops-in-action repository (i.e. not PyMPDATA directly). But all I did was add a flag to the Shipway_and_Hill_2012 settings in PyMPDATA which turns on/off the use of a max_step in solve_ivp for the dry density.

Let me just do that more neatly and get back to you.

[yoctoyotta1024]

I can't reproduce he plot with PyMPDATA because using apprx_drhod_dz=False seems to raise an error? But I've added two more scripts which show basically what I did in my code, hopefully that helps!