Convert individually each item of the input list as doing so on an the entire array generates an error if the expected type is int and the provided values are in floats.

Computers are much better at multiplying than at dividing. Therefore to correct the emission for redshifting we multiply by 1/(1+z) rather than dividing by 1+z.

The absorption of the Lyman continuum from old stars tended to be overestimated leading to some “negative fluxes” for the Lyman continuum.

In rare circumstances requiring a specific distribution of redshifts the integration of the spectrum in some filters was not done correctly, inducing relative errors of ~10¯⁵-10¯⁶.

To compute the stellar spectrum of the young component, do not pass the full SFH with the old part set to 0. Rather, only pass the corresponding part of the SFH. This nearly doubles the computing speed of the stellar spectrum.

We do not store the time grid in the SED anymore given that we assume it starts at 0 Myr with steps of 1 Myr, we can easily reconstruct to save it if needed. It should save a little bit of memory and it should go a little bit faster.

In the output the sfh.age parameter would correspond to the input value minus 1. Now both values are consistent with one another.

For convenience the time grids start at 0. Howver the SSP grids start at 1. This  offset means that the oldest bin in an SFH is set to 0, for both the young and old components. Now rather than trying to interpolate the SFH time grid on the SSP time grid, we rely on the fact that the SFH time grid starts from 0, so it is easy to know which SSP to select (as many as the number of steps). Also adapt the sfh_buat08 module.

Make sure all the variables are saved in the FITS and not just the ones with are in mass_proportional_info. Bug seen by Yannick 2Eagle Eye' Roehlly.

The optionally saved spectra in the pdf_analysis and savefluxes modules were saved in the VO-table format. The most important downside is that it is very slow to write to, which proves to be a major bottleneck in the computing speed. To overcome this issue, we rather save the spectra using the FITS formation. Instead of having one file containing the spectra (including the various components) and the SFH in a single file, now we have one file for the spectra and one file for the SFH.

When reading a table from a file, convert all the integer columns to float
(except the 'id' column).  These columns are fluxes or error on fluxes and thus
should be treated as float even when they contain only integer.  This solves the
problem of trying to affect NaN value to an integer column.

Fixes #6

The name creation_modules is a bit vague. For more clarity and consistency we rather use sed_modules now

Remove some variables used only once and that could be directly obtained from the configuration dictionary.

Homogenise savefluxes and pdf_analysis for the variables/analysed_variables keyword in pcigale.ini. We adopt the former as it is more compact and applies correctly to both modules.

Change the vague column_list keyword in pcigale.ini to bands. This will require the user to rebuild their pcigale.ini but that is necessary for the validation anyway so that is the right moment to do so.

Pratical implementation of the validation of the parameters. The patch is quite long as it has a direct effect on the structure of the configuration dictionary. The validation has the advantage of automatically convert the parameters to the right type. Therefore rather than building a dictionary ourselves, we rather use the ready-made dictionary from ConfigObj. Because the names of the sections are not the same, quite a bit of code had to be adapted. Finally, note that the validation file containing the specification of each variable, pcigale.ini.spec, is created while building the pcigale.ini file. Also because it is needed to convert the data to the right type, one cannot run cigale without a correct validation file.

Specify the types and the constraints of the input parameters for the SED creation modules and the analysis modules. Once validation is activated, the validation module will be able to infer the constraints from the specified information.

So far no validation was made on the input parameters provided in pcigale.ini. This could lead to runtime crashes when an argument took an unexpected value. As we are using the ConfigObj module to handle the configuration file, we can also use the validate module that comes with it. However the standard validation functions are not precise enough in our case. This file defines two validation functions for input parameters, and a new exception that will be used in a subsequent patch.

Modules were a bit inconsistent how the input parameters are used. It was not always ensured they were the right type, they were sometimes read several times, and finally they were not always obtained from the _init_code() method. This is an attempt to clean that up as much as possible.

astropy.Table stores columns made of strings (for instance the sfhdelayed.normalise parameter) using the np.str_ type. The issue is that this type is not marshalable, making it impossible to use the modules parameters as keys to cache the partially computed SED in a dictionary. To solve this, rather than using the table directly, we split it into several dictionaries, one for each module. Each dictionary contains the arguments of lists that we convert to the right type.