search

joellebegin / Power-Spectra

0 stars 0 forks source link

ps to field scale problem #17

Open joellebegin opened 4 years ago

joellebegin commented 4 years ago

New pspec to field works very nicely.

When I give the generated field to my power spectrum code, it returns the shape and peak location perfectly, but the power scale is wildly off (like, many many orders of magnitude).

Have no idea where the issue is. Old pspec to field code did this as well, I just didn't care much because it was qualitative but now we actually need this to work properly.

acliu commented 4 years ago

Volume factors, maybe?

joellebegin commented 4 years ago

Yeahh that's what I've been thinking. I figure I have to multiply by the volume somewhere in there (since in computing the power spectrum you divide by the volume) so I'm just multiplying by volume wherever it seems reasonable to see if changes things.

No luck yet.

acliu commented 4 years ago

I call that "debugging by Monte Carlo". Very dangerous. Not recommended.

If P(k) ~ < |T-tilde (k)|^2 > / V, then the standard deviation is sqrt(V P(k)).

joellebegin commented 4 years ago

Yeah, that's I what I have! still doesn't give the right scale though

acliu commented 4 years ago

Fascinating.

joellebegin commented 4 years ago

Yeah :( I'm sure it's just some units not being converted properly or something; will keep looking!