installing relxill models with sherpa

[JohannesBuchner]

Is it currently possible to install these models https://www.sternwarte.uni-erlangen.de/~dauser/research/relxill/index.html and fit them with sherpa?

[anetasie]

@JohannesBuchner thanks for your question. We are currently investigating issues with compiling the most recent version of the model. @nplee will give you more detailed descriptions of the current status.

[JohannesBuchner]

I see. What is the latest model version known to work?

I guess the instructions to follow are https://cxc.harvard.edu/sherpa4.8/threads/xspec_user_model/

[nplee]

@JohannesBuchner for a bit more information, are you presently using Sherpa with one of the CIAO distributions or as standalone?

Coincidentally, I've been looking into this the past week, and v1.4.3 of the model can be used with the Sherpa build packaged with the CIAO 4.13 binaries and Conda distribution.

With CIAO 4.14, I've finally been able to get v1.4.3 and v2.1 to build with the binaries. Unfortunately, my work system was migrated to a new OS earlier this week and it's taken a bit of time to get my Anaconda setup working again so that I can test against the 4.14 Conda distribution.

[JohannesBuchner]

oh excellent. I usually use the conda install. I think relxill combined with BXA could be a very good combination to fit effectively.

[anetasie]

Yes, using BXA and relxill should be the way to go. Hopefully, Nick will resolve issues with building the model.

[nplee]

In your CIAO environment, please replace $ASCDS_CONTRIB/bin/convert_xspec_user_model with: convert_xspec_user_model.aug31-relxill.

Assuming you're using the Conda-compilers, then set:

(t)csh:

setenv CC ${CONDA_PREFIX}/bin/x86_64-conda-linux-gnu-gcc
setenv CXX ${CONDA_PREFIX}/bin/x86_64-conda-linux-gnu-g++
setenv CFLAGS "-Wno-unused-variable -Wno-array-bounds"

bash/zsh:

export CC="${CONDA_PREFIX}/bin/x86_64-conda-linux-gnu-gcc"
export CXX="${CONDA_PREFIX}/bin/x86_64-conda-linux-gnu-g++"
export CFLAGS="-Wno-unused-variable -Wno-array-bounds"

If installing with the CIAO 4.14 binaries and default system compilers instead, it may be necessary to set the LDFLAGS variable to -Wl,-rpath,${ASCDS_INSTALL}/ots/lib and on my clean RHEL8 system I also had to also explicitly point to the system's libfftw, "l:libfftw.so.3 -Wl,-rpath,${ASCDS_INSTALL}/ots/lib".

For relxill 2.1, in the unpacked directory with the source code you will need to change the line:

#include "fftw/fftw3.h"

to

#include "fftw3.h"

in the common.h and Relbase.cpp files.

The script can then be run from the source directory:

unix% convert_xspec_user_model relxill_v2 lmodel_relxill.dat --clobber

which should compile the model, and then import into Sherpa:

sherpa> import relxill_v2.ui

so you'll be able to then set a source model like set_source(xsumrelline.rll1). An important step is to make sure that the models are available to access the data tables the models are dependent upon. Assuming you've unpacked them in /path/to/directory/with/relxill/tables/files, then:

sherpa> os.environ["RELXILL_TABLE_MODEL"] = "/path/to/directory/with/relxill/tables/files"

will set the location from the models to look in. If you would like to avoid declaring the os.environ["RELXILL_TABLE_PATH"] each time you import the model package, you can hard code it on line 119 of common.h, changing:

#define RELXILL_TABLE_PATH "./"

to

#define RELXILL_TABLE_PATH "/path/to/directory/with/relxill/tables/files"

[DougBurke]

I'm going to close this as there's not much more we can do (the heavy lifting is done by convert_xspec_user_model, which requires CIAO, although is uses code in Sherpa), but please let me know if we need to re-open it.

This model is "annoying" as it uses fftw headers which are in a different location to those in XSPEC, meaning the code needs to be updated, as mentioned above.