GammaLib

In the significance branch I linked to the li_ma_significance function works:

In [3]: gammalib.li_ma_significance(10, 20, 0.1)
Out[3]: 3.6850322319420274

I’ll now try to put it in a gamma namespace and see if the Makefiles work.

Christoph Deil wrote:

TODOs:

• Decide on naming convention for functions and how to expose them in C++ (namespace or prefix).
• Expose the function in Python (add Swig wrapper)
• Document function coding conventions.

TODOs are “Actions” in the Redmine tracker that are appended to a feature. Just created these actions, can now be easily see in the Agile taskboard: https://cta-redmine.irap.omp.eu/rb/taskboards/23

I just updated https://github.com/cdeil/gammalib/compare/significance .
It was just for me to play around with tough ... we don’t have to merge it.
I can add this function properly once the functions in gammalib issues have been decided.

One question I have is whether e.g. the gammln function in GNumerics should be in the gammalib namespace and if it should be part of the public GammaLib API. I think it currently is included by GammaLib.hpp, but maybe it shouldn’t, because it is an internal implementation detail?

I would make it public using the gammalib namespace, as somebody may need it for his own calculations. For the moment there is just a single function in GNumerics, but the idea was to add more ... (for example your Li and Ma function).

Created branch function-namespace where all GTools constants are now in gammalib namespace.

Port now the GTools into the gammalib namespace.

Christoph Deil wrote:

I just updated https://github.com/cdeil/gammalib/compare/significance .
It was just for me to play around with tough ... we don’t have to merge it.
I can add this function properly once the functions in gammalib issues have been decided.

One question I have is whether e.g. the gammln function in GNumerics should be in the gammalib namespace and if it should be part of the public GammaLib API. I think it currently is included by GammaLib.hpp, but maybe it shouldn’t, because it is an internal implementation detail?

I now placed all constants and functions in the gammalib namespace. Pushed in devel.

You may now add your function, the only question being: where? In fact, I renamed GNumerics to GMath, as the only function that was in there was an elementary mathematical functions. I also added a bunch of other mathematical functions that were in GTools and GWcslib, so that they are now centralized. I also moved all mathematical constants from GTools to GMath.

The Li and Ma function would not be a mathematical function. It could go in GTools (there are still other functions in there, mostly string and conversion function, but also functions for analytical power law integrations, for example).

Is it possible to make the C++ and Python API consistent by putting the functions and constants also in a gammalib namespace?
At the moment they are in the top-level namespace (and not reachable via `from gammalib import *` I believe):
```
In [8]: import gammalib

In [9]: gammalib.gammln(42)
Out<sup>9</sup>: 114.03421178150786

In [10]: gammalib.gammalib
Out<sup>10</sup>: <module 'gammalib’ from '/usr/local/gamma/lib/python2.7/site-packages/gammalib/__init__.pyc’>
```

Christoph Deil wrote:

Is it possible to make the C++ and Python API consistent by putting the functions and constants also in a gammalib namespace?
At the moment they are in the top-level namespace (and not reachable via `from gammalib import *` I believe):
```
In [8]: import gammalib

In [9]: gammalib.gammln(42)
Out<sup>9</sup>: 114.03421178150786

In [10]: gammalib.gammalib
Out<sup>10</sup>: <module 'gammalib’ from '/usr/local/gamma/lib/python2.7/site-packages/gammalib/__init__.pyc’>
```

This is done now, and merged in the devel branch. Have a look at the new matrix_howto.py file in the examples/python folder to see this at work.