# BAOFIT Output Files

All output files use a common prefix that is set using the --output-prefix command-line option. This is denoted <PREFIX> in the examples below.

The BaoFitTools.m module in the public mathpkg provides some Mathematica utilities for processing, analyzing, and plotting BAOFIT output files.

## <PREFIX>fit.config

This file gives the best-fit results in the same format used by the --model-config command-line option. Numerical values are given in full machine precision, e.g.

...
value[(1+beta)*bias]=-0.4368240439658288; error[(1+beta)*bias]=0.053396638860647487;
...

This allows you to easily use the results of one fit to configure a new fit, e.g. (note that value has been replaced by fix in this example, but that is not necessary):

baofit ... --model-config "fix[(1+beta)*bias]=-0.4368240439658288;"

You can also insert (possibly edited) lines from this file into the file passed to the program via the -i option (or the equivalent --input option). The config/ subdirectory provides sample input files.

## <PREFIX>fit.chisq

This file gives the minimum chi-square of the best fit (first column), together with the number of data bins after cuts (second column), number of floating parameters (third column) and the chi-square probability (fourth column). Numerical values are given in full machine precision, e.g.

1499.0486481245553 1515 13 0.51663808887756257

## <PREFIX>save.pars

This file gives the best-fit results using three columns of numbers, e.g.

0 1.4106099313779259 0.37692331585396388
1 -0.4368240439658288 0
2 3.0635839287517128 0.88451264991993672
3 0 0
4 1 0
...

The first column is the parameter integer index, starting from zero and including fixed parameters. The order of parameters matches the order used in fit results printed by the program, e.g.

                       beta =     1.411155 +/-     0.399842 $1.4 \pm 0.4$ gauss prior @ (1,1.8;1)
(1+beta)*bias =    -0.436824 +/-     0.053397 $-0.44 \pm 0.05$
gamma-bias =     3.062579 +/-     0.887827 $3.1 \pm 0.9$ gauss prior @ (2.8,4.8;1)
gamma-beta =     0.000000
BAO amplitude =     1.000000
...

The second column gives the best-fit parameter value and the third column gives the estimated symmetric parameter error (sqrt of the diagonal covariance matrix element). Numerical values are printed to full machine precision (unlike the program output). Fixed parameters have zero error.

## <PREFIX>save.pcov

This file gives the estimated covariance matrix of the fit parameters in the same format used for input matrices with .cov or .icov extensions described here. The global indexing used in this file matches the index used in the <PREFIX>save.pars file.

## <PREFIX>residuals.dat

This file summarizes the best fit model using 10 columns of numbers:

 Column Description 0 bin index, matching the indexing used in the input .data or .wdata file 1 axis-0 bin center 2 axis-1 bin center 3 axis-2 bin center 4 radial coordinate of bin center in Mpc/h 5 mu value of bin center (or ell value for multipole input data) 6 redshift of bin center 7 best fit model evaluated at bin center 8 input data for this bin 9 diagonal error for this bin (sqrt of input covariance diagonal matrix element)

The values in columns 1-3 depend on the --data-format option as specified in this table. The value in column 5 will be an integer multipole number when using --data-format comoving-multiple. There will often be duplication between columns 1-3 and 4-6 but both are included in the file to support different applications.

You can calculate the residual pull vector as ([8]-[7])/[9] but be aware that this ignores off-diagonal correlations between input data bins.

## <PREFIX>fit.dat

This is an optional output file that only be generated when the value passed via the --ndump command-line option is greater than zero. The file provides multipoles for the best-fit model calculated at the redshift specified by the --zdump command-line option (or the the redshift of the first input data bin if no value is specified).

The file contains four columns of numbers. The first column is a vector of increasing radial coordinate values in Mpc/h covering [rmin,rmax] with ndump steps (the radial limits are specified using the --rmin and --rmax command-line options). The last three columns give the corresponding calculated values of the ell = 0,2,4 correlation function multipoles.