tdk111/baryonmodel

Baryon feedback model on the ratio of the observed matter power spectrum to the dark-matter-only power spectrum. Used in Kitching, Verde, Heavens, Jimenez (2016). This is a PCA component based model that can reproduce the range of feedback in both the OWLS and Illustris simulations.

Baryon Model used in Kitching, Verde, Heavens, Jimenez (2016)

The file PCA.dat contains the first two eigenvectors and the mean for the ratio P_baryon/P_DM

the columns are 
z,k,eigen1(k,z),eigen2(k,z), mean(k,z)

so any P(k,z) would be P(k,z)=Pkdm*(mean+coeff1*eigen1+coeff2*eigen2) where coeff1 and coeff2 are free parameters; their priors should go from -1 to 1 if we one requires corrections not larger than what is in the suite of the input sims from OWLS. To also allow the flexibility to fit the Illustris simulations the functional form used in the paper was 

P(k,z)=Pkdm*[1.+(mean-1)*coeff1+eigen1*coeff2]

where coeff1 and coeff2 are free parameters.

If one wants to use all the components then the file is PCAall.dat

PCAall.dat was written like this(same structure as above but now with 9 PCA's). In Fortran 90 the code to do the writing is below.

count=0
eigen=fltarr(nz,nkk,9)
meancorr=fltarr(nz,nkk)
openw,1,'PCAall.dat'
for j=0,nz-1 do begin ;loop over z
   for i=0,nkk-1 do begin
      eigen(j,i,*)=evec(count,*)
      meancorr(j,i)=mean(count)
      printf,1,zz(j),k(i),evec(count,0),evec(count,1),evec(count,2),evec(count,3), evec(count,4),evec(count,5),evec(count,6), evec(count,7),evec(count,8), mean(count)
     count=count+1 
   endfor
endfor
close,1

The Matlab code model.m reads in the PCA data and produces output power spectrum ratios for specified coefficient values.