This is an experiment in providing support for the XSPEC local models in Sherpa. It relies on you having used the conda installation method to install CIAO 4.12.
At present the only supported models are:
agnslimthcompc(although this is known to be problematic, so don't try it)zkerrbb
and there has been essentially no testing to check this works
correctly (although development of this package did find
a bug in the FORTRAN code of the zkerrbb model).
At the moment I have only got this working with the conda release of CIAO 4.12, using Linux/Python 3.7 and macOS/Python 3.7. I have not tested the other conda Python versions, and it does not seem to work with the ciao-install version of CIAO (this version was build using an old version of gcc and it looks like it creates a different ABI for the compiled code, and this is more than I ever wanted to know about linkers so I am concentrating on the conda version).
It could be made to work with CIAO 4.11 but I do not have time
to do so at this time, and CIAO 4.12 is now available. If you did
want to try, the include files in the xspec/ directory
would have to be replaced with those for XSPEC 12.10.0, the
library versions for the XSPEC libraries changed in setup.py,
and the link issues mentioned above worked out (since CIAO 4.11
is only available via ciao-install).
The following models are in XSPEC 12.10.1, and so are included in CIAO 4.12 (there will be no attempt to include these models in this module):
- agnsed/:
agnsed,qsosed - cluscool/:
cph,vcph - ky/:
kyrline(kyconvis a convolution model which is not directly-supported by Sherpa)
The models are taken from the XSPEC GitHub repository at the following revision: December 11 2019 - 5ef6af6ed9991746cbaef42c1fac3c054f6b955c.
This code is released into the public domain, using the CC0 1.0 Universal (CC0 1.0) Public Domain Dedication, which can also be found in the LICENSE file in the distribution.
This was written by Doug Burke - dburke.gw@gmail.com - and support is on a best-effort basis. It is not an official product of my employer.
Start up a conda-installed version of CIAO 4.12 and then try
% git clone https://github.com/DougBurke/xspeclmodels
% cd xspeclmodels
% pip install -e .
Hopefully there'll be no error message here. Now try running the
tests (which may require you to conda install pytest first):
% pytest
=========================================== test session starts ============================================
platform linux -- Python 3.7.5, pytest-5.3.2, py-1.8.0, pluggy-0.13.1
rootdir: /somewhere/over/the/rainbow/xspeclmpodels
collected 15 items
src/tests/test_xspeclmodels.py ........ [ 53%]
src/tests/test_xspeclmodels_ui.py ....... [100%]
============================================= warnings summary =============================================
/home/ahappycamper/anaconda/envs/ciao412/lib/python3.7/site-packages/pytransform/__init__.py:13
/home/ahappycamper/anaconda/envs/ciao412/lib/python3.7/site-packages/pytransform/__init__.py:13: DeprecationWarning: the imp module is deprecated in favour of importlib; see the module's documentation for alternative uses
import imp
-- Docs: https://docs.pytest.org/en/latest/warnings.html
====================================== 15 passed, 1 warning in 1.85s =======================================
Note that the DeprecationWarning about pytransform can be ignored.
If there are any failures here then please check the
issues list to see
if it's known about and, if not, add an issue.
Once the tests have passed (note that they are very limited, and are only designed to check the models are "doing something", not that they are doing the right thing!), you can install the module with:
% python setup.py install
After installation, you should be able to start Sherpa and
import the xspeclmodels.ui module, which will add the
XSPEC local models to Sherpa:
% sherpa
-----------------------------------------------------
Welcome to Sherpa: CXC's Modeling and Fitting Package
-----------------------------------------------------
Sherpa 4.12.0
Python 3.7.5 (default, Oct 25 2019, 15:51:11)
Type 'copyright', 'credits' or 'license' for more information
IPython 7.10.2 -- An enhanced Interactive Python. Type '?' for help.
IPython profile: sherpa
Using matplotlib backend: Qt5Agg
sherpa In [1]: import xspeclmodels.ui
Adding XSPEC local model: xsagnslim
Adding XSPEC local model: xszkerrbb
Adding XSPEC local model: xsthcompc
sherpa In [2]: src = xsphabs.gal * xszkerrbb.mdl
sherpa In [3]: print(src)
(xsphabs.gal * xszkerrbb.mdl)
Param Type Value Min Max Units
----- ---- ----- --- --- -----
gal.nH thawed 1 0 100000 10^22 atoms / cm^2
mdl.eta frozen 0 0 1
mdl.a thawed 0.5 -0.99 0.999
mdl.i frozen 30 0 85 degree
mdl.Mbh frozen 1e+07 3 1e+10 M_sun
mdl.Mdd thawed 1 1e-05 10000 M0yr
mdl.z frozen 0.01 0 10
mdl.fcol frozen 2 -100 100
mdl.rflag thawed 1 -3.40282e+38 3.40282e+38
mdl.lflag thawed 1 -3.40282e+38 3.40282e+38
mdl.norm thawed 1 0 1e+24
sherpa In [4]: dataspace1d(0.1, 6, 0.01)
sherpa In [5]: set_source(src)
sherpa In [6]: plot_source(xlog=True, ylog=True)
For those that just want the model class, and do not use the
Sherpa "UI" layer, you can just import xspeclmodels.
The screen messages created by xspeclmodels.ui are from the
Sherpa logging instance, so you can turn them off when
importing the model with the following (don't forget to
change the level back to INFO afterwards!):
sherpa In [1]: import logging
sherpa In [2]: logger = logging.getLogger('sherpa')
sherpa In [3]: logger.setLevel(logging.WARNING)
sherpa In [4]: import xspeclmodels.ui
sherpa In [5]: logger.setLevel(logging.INFO)