MTNG 3Gpc/h inference

[maho3]

This issue is solely to keep notes for analyzing 3 Gpc/h inference using MTNG 3 Gpc/h ground-truths.

Notebook: https://github.com/maho3/ltu-gobig/blob/run_experiments/experiments/mtnglike/inference_mtng.ipynb

[maho3]

Configuration

• Training suite: ltu-cmass, BORGPM + CHARM, SS=1, B=1
• Test suite: MTNG galaxy-occupied halos

Coverage

Notes

This is really bad likely because of many things:

• We are taking all galaxy-occupied halos, at all masses. We should cut on mass
• We have some issues with CHARM inputs (See https://github.com/maho3/ltu-gobig/issues/22)
• We are using B=1, SS=1, which results in low-biased DM densities
• MTNG galaxies were populated on a lightcone, but I don't know which direction the lightcone was taken at

[maho3]

The reason for this ^ was that the MTNG simulation data was already masked in a conical lightcone. Specifically, it covered one octant of the sky.

When we measured P(k) without the mask correction, this messed with the large-scale modes of the galaxy P(k), and removed the classic 'turnover' we expect of large-scale surveys.

Instead, I have taken this raw MTNG lightcone and rotated it to overlap with the NGC footprint.

And then trimmed things down in the observational sample to only include the overlap.

After redshift reweighting, this matches well:

Now, the monopole P(k) looks closer to the observational sample

Though, the MTNG cosmology + galaxy configuration here may not be ideal.