Simplify the cache architecture as we always use the in-memory one. The removes one level of indirection and makes the model generation faster by 1%.

Implement the possibility of not caching a module in order to save memory. A typical use case is when having many different stellar populations at different metallicities. It can take a substantial amount of memory, especially when using many parallel processes.

Correct a bug that was preventing the module restframe_parameters from being listed among the available modules.

Fix a crash in the computation of the χ² because previous optimisations had been been not fully propagated to the case of upper limits.

Give a more reasonable amplitude for the UV bump with dustatt_powerlaw. Otherwise it is a factor Rv too large compared to dustatt_calzleit.

Add the new restframe_param module. It replaces most of the functionality provided in the unofficial param module, which has been trimmed down to only computing fluxes in the observed frame and has been renamed fluxes. Unlike param, the restframe_param module is to be called before the redshifting module.

When calling np.sum(), the axis should be an integer, not a float, otherwise it triggers a very disgracious Numpy warning. Now it is gone.

Use np.take() rather than fancy indexing on large arrays. This accelerates the computation of the models by 10%.

Optimise the interpolation of the luminosity components 1. computing the interpolation of all the components in one step rather than interpolating separately for each component, and 2. caching some intermediate computations that only depend on the grid sampling and therefore apply to all physical components.

Speed up the computation of the models using the compiled interp function rather than the regular numpy one so we bypass checks on the size and type of the arrays.

Update the nebular models from Akio Inoue. Now log U is sampled in steps of 0.1 dex down to -4. Also deviations from the 10000K case B assumption are taken into account.

Limit the number of threads to 1 per process in order to prevent MKL from creating numerous threads leading to a strong oversubscription of the CPU/RAM as pcigale already does its own parallelisation.

Implement the computation of the mass-weighted age for the bc03 and m2005 module. This is done in the corresponding convolve() functions. The information is returned in a new variable called info_all, which also contains all the information when the entire SFH is included (rather than separating old and young populations).

Do not return the break strengths in the bc03 module as they are not computed corrected by a convolvution.

In order to release a new version correcting some bugs, we remove the
work in progress correction for upper limits.

Correct the function adjusting the data and implements a new way to indicate
upper limits.

FIXME: This breaks other parts of the code that must be corrected because now
upper limits are indicated with a negative flux.

When on a grid, make sure that the models are always computed in the same order to facilitate comparisons between different runs.

Remove the output_file and output_format parameters from the savefluxes module. Now results are systematically saved as ASCII and FITS tables.

As we loop over the bands to compute the χ², it is smarter that the bands are the slower varying index and the models the faster varying index. That way it is more likely that models at the correct redshift are prefetched into the cache.

Use a multiplication rather than a division to convert luminosities into fluxes. This should yield a small speedup.

When giving the list of parameters through a file, cigale did not compute properly what parameter changed between to successive models.

With the new sanity check of the input parameters, cigale did not handle the fact that the redshift could be given in the parameters file. Now this is handled properly.