c3k-lib

Code here is for generating low-resolution spectra files for (python-)fsps, using the c3k library.

Resolution

Resolution is specified in a yml file as tuples of (lambda_low, lambda_hi, R, usefft)

Note the c3k_v1.3 default resolutions are

0.009micron - 0.0912micron; opacity binning with wavelength spacing lambda/Delta_lambda=4500
0.09121micron - 2.5micron; opacity binning with wavelength spacing lambda/Delta_lambda=100000
2.5micron - 40micron; opacity binning with wavelength spacing lambda/Delta_lambda=4500

Beyond 40 microns we stitch a Rayleigh-Jeans tail onto the spectra.

Procedure

Install. You should have FSPS already installed and $SPS_HOME environment variable set.

cd $SCRATCH/conroy_lab/$USER
git clone [email protected]:bd-j/c3k-fsps-lib.git
cd c3k-fsps-lib
module purge
module load Anaconda3/2020.11
conda env create -f environment.yml
source activate c3k
cd ..

Check that all the relevant flux files are there

fdir=/n/holystore01/LABS/conroy_lab/Lab/bdjohnson/data/kurucz/c3k_v1.3/fullres
for f in $fdir/*full*h5; do ls ${f/.full./.flux.}; done

If they are not, then you need to make them using make_flux.py, driven by ody_flux.sh

Make the low resolution SED H5 files, and the H5 files of the SEDs interpolated to the FSPS logt-logg gridpoints (based on BaSeL or, in new versions, c3k). Also make binary format versions for FSPS itself, and metadata files (wavelength, resolution, zlegend)
```
cd jobs/
libname=nirspec # name of the segments_<libname>.yml file
sbatch --export=ALL,libname=${libname} --array=0-64 ody_resample.sh
```
Note that you may wish to change the resampling here, the metallicities to consider, or the output filename and directory. The output names are given by the supplied libname and the metallicities to consider are specified in c3k_resample.py. The number of jobs in the array should be equal to the number of feh-afe pairs (usually 13 * N_afe)

There are now several sets of binary files and ancillary files in the output/${libname}/for_fsps/ directory (by default) that can be moved to the $SPS_HOME/SPECTRA/C3K directory. By altering sps_vars.f90 you can choose to use these spectra.
If you're me the output/${libname} directory should be copied to /n/holystore01/LABS/conroy_lab/Lab/bdjohnson/data/kurucz/${ck_vers}/fsps-lib/
```
cd jobs/
libname=nirspec # name of the segments_<libname>.yml file
sbatch --export=ALL,libname=${libname} ody_copy.sh
```
Implement in FSPS

This depends a bit on what version of FSPS you have. Instructions her are for the 'older' FSPS. Assuming you have downloaded FSPS, the first thing to do is copy the relevant files to the FSPS repo. Note this changes version tracked files! The prefix also changes.
```
cp <path/to/output/libname>/for_fsps/<prefix>*bin $SPS_HOME/SPECTRA/C3K/
cp <path/to/output/libname>/for_fsps/<prefix>*dat $SPS_HOME/SPECTRA/C3K/
cp <path/to/output/libname>/for_fsps/<prefix>_zlegend.dat $SPS_HOME/SPECTRA/C3K/zlegend.dat
cp <path/to/output/libname>/for_fsps/<prefix>.wave $SPS_HOME/SPECTRA/C3K/<prefix>.lambda
```
Then, you need to change the sps_vars.f90 code. You'll want to specify the the C3K library using the pre-compiler directives, and you'll need to change environment variables in the elif (C3K) block; the values for nzinit and nspec can be obtained by wc on the *.lambda and *_zlegend.dat files, and the spec_type variable (and its length!) could be changed to e.g. c3k_ns. Here's what the diff of sps_vars.f90 looks like with nz=11, nspec=13749 and a prefix of c3k_ns:
```
@@ 7
-#define MILES 1
+#define MILES 0

@@ 16
-#define C3K 0
+#define C3K 1

   @@ 239
   #elif (C3K)
      REAL(SP), PARAMETER :: zsol_spec = 0.0134
-        CHARACTER(11), PARAMETER :: spec_type = 'c3k_afe+0.0'
-        INTEGER, PARAMETER :: nzinit=11
-        INTEGER, PARAMETER :: nspec=11149  !46666 !47378 !, 26500
+        CHARACTER(6), PARAMETER :: spec_type = 'c3k_ns'
+        INTEGER, PARAMETER :: nzinit=11
+        INTEGER, PARAMETER :: nspec=13749
```
You can now recompile fsps (cd $SPS_HOME/src; make clean; make all) and it should use the new C3K spectral library.

Implement in python-fsps

This requires a development install of python-fsps. You need to copy the files to $SPS_HOME as above. Then, after cloning the repo, you'll need to edit fsps/src/fsps/libfsps/src/sps_vars.f90 in the same way as described above. Finally, install with the C3K library selected. This looks like:

cp <path/to/output/libname>/for_fsps/c3k_ns* $SPS_HOME/SPECTRA/C3K/
cp <path/to/output/libname>/for_fsps/c3k_ns_zlegend.dat $SPS_HOME/SPECTRA/C3K/zlegend.dat

git clone --recursive https://github.com/dfm/python-fsps.git
<change python-fsps/fsps/src/fsps/libfsps/src/sps_vars.f90 as in step 5>

cd python-fsps
python -m pip uninstall fsps
FFLAGS="-DMILES=0 -DC3K=1" python -m pip install .

Test python-fsps implementation

You can test that this worked by running a script in the test/ directory and looking at the output. (Make sure to do this the environment where you installed python-fsps above) . This may take some time, as the SSPs are regenerated several times.
```
cd c3k-fsps/lib/tests
python fsps_feature_demo.py
open features.pdf
```
You can also plot the grid for every metallicity and the corresponding isochrones with make_hrd.py

It might also be instructive to compare the SSP spectra to another library (e.g. MILES)