meeting 2024 10 18 n315 - JacobPilawa/TriaxSchwarzschild_wiki_5 GitHub Wiki

Two quick updates here before the week is over:

• Grid C + D Reminimization:

  • We had these libraries already computed, so we wanted to minimize our kinematics to add these to the mix since they're totally legit models. The results are quite consistent with what we've seen in the past, with Grids C and D performing substantially better in terms of the chi2.
  • I think the interesting result here is that the low Tminor and high Tminor minima seem about equally favored now, whereas in the old versions of the N315 models, it seemed like the low Tminor part of the space was slightly more favored than the high portion.
    • With this said, it'll be interesting to see if addings scalings changes things around at all. I am curious if the addition of scalings will end up pulling us back toward the low Tmin part of the space, but we'll have to see.

• Grid Bprime d3 Results:

  • I re-ran the ~300 lowest chi2 models from our dither1 experiement with dither3, and the results are not too surprising, I think. It seems like Grid B prime confirms that the low and high Tminor portions are approximately equally well-fit, with Tmin=[0.1,0.4] being ruled out by the new Grid Bprime models.
  • With that said, I think this was still a useful set of models to run to confirm the general shape of the landscape and to assess the relative constraints of the low and high Tmin parts of the space.

Plan Moving Ahead:

• It's not obvious to me that we need to run any additional Grid Bprime models beyond the ~300 we added to the low Tminor part of the space. It seems like the Grid Bprime results that we do have rule out this part of the space quite convincingly, leaving instead the bimodal minima at high and low Tmin.
• I think the most obvious step forward is to run our usual scalings for Grid Bprime, Grid C, and Grid D. This will improve our coverage of the minimum quite cheaply, and then we can come back and assess where things are. I can also try to run our GPR and dynesty after getting the best-scaled versions of all the models.

• Note that the left and right columns are identical aside from the y-limits.
• Here's a link to the old Github page which has the old Grid C + Grid D results for quick comparison

• Note -- I hadn't noticed this when I scraped the data, but it looks like only ~2/3 of the models ended up minimizing last night, with the other ~1/3 getting cancelled due to Expanse filesystem issues (but looks like they never started so we didn't lose hours or anything). Those are now minimizing and will be added shortly!
• With that said, I think that this looks reasonable for well-fitting models. There is a bit of scatter in the resulting trend line (roughly a width of ~100 on either side), but still roughly consistent with what we've been seeing. I also think this is somewhat expected since we're only covering a range of ~200 in chi2 with ~200 model points, so I expected the trend to be a bit noisier.