v1.5.0

New features:

• Added option OUTPUT_PRECISION to control the numerical precision in the output catalog.

v1.4.0 -- line spread function and spectrum template error

Warning: this release changes the behavior of the program compared to previous releases. Previous releases applied the same template error function to the photometry and spectrum, unless AUTO_SCALE=1 in which case the error function was only applied to the photometry. Now, the value of AUTO_SCALE is no longer relevant. Whether or not the spectrum has a template error function applied is entirely controlled by TEMP_ERR_SPEC_FILE (not set by default). This brings the default FAST++ behavior closer to that of the original FAST, where the template error function was never used when fitting the spectrum.

New features:

• Added option TEMP_ERR_SPEC_FILE to control the template error function for the spectrum
• The spectrum no longer inherits the same template error function as the photometry
• AUTO_SCALE=1 no longer automatically disables the template error for the spectrum
• Added option SPEC_LSF_FILE to apply a line spread function to the models
• Added option LSF_BEST_FIT for appending the LSF-convolved model to the best fit output

Bug fixes:

• Fixed crash when saving extra outputs and using an absolute path in CATALOG
• Fixed invalid cache and grid files if using extremely large number of models or galaxies (more than 4 billion)

v1.3.2 -- Fix compilation with GCC 11.x

This is a bugfix release, addressing only a compilation issues when using GCC 11.x (not affecting GCC 10 or 12).

v1.3.1 -- Fix compilation issues, grid cache hash, and grid2fits bug

This is a bugfix release, addressing mostly compilation issues (in particular, compiling with the latest versions of XCode on MacOS). It also fixes a bug in the grid2fits tool, where the extracted data would be incorrect when more than one galaxy is present in the grid.

There is otherwise no change to the fitting algorithm or its output.

v1.3.0 -- Added differential attenuation model

This new version includes a new major feature: differential attenuation. In the original dust model of FAST, all stars suffer the same amount of extinction by dust. With differential attenuation enabled, young stars can be given an extra amount of extinction, since they can be assumed to still reside in their birth clouds (this is the Charlot & Fall 2000 prescription). This feature is disabled by default but can be turned on in the .param file, and the amount of extra extinction in birth clouds can be varied on a grid (or fixed to a single value).

A consequence of this addition is that it is now possible to apply the dust luminosity prior either on the whole galaxy, only on the birth clouds, or only on the cirrus (diffuse ISM) components.

v1.2.2 -- Bug fix release

This release contains a small bug fix when leaving the template error function file path empty, where the code would use the default template error function instead of none at all.

v1.2.1 -- Bug fix release

This release only contains two minor bug fixes, one regarding the Milky Way attenuation curve, and another regarding best fit SEDs when BEST_FROM_SIM=1 is enabled.

v1.2 -- Major update with new features

This version of the code contains all the features that were used in Schreiber et al. (2018). The list below gives the full list of changes.

Features:

• Added the APPLY_VDISP opion, to convolve SEDs with a fixed velocity dispersion.
• Added the REST_MAG option to produce rest-frame magnitudes.
• Added the CUSTOM_SFH_LOOKBACK option to have time running backwards in the custom SFH expression
• Added the CONTINUUM_INDICES option to produce absorption line EWs, Dn4000, etc.
• Added fast++-grid2fits tool to convert .grid files into .fits files.
• Added fast++-sfh2sed tool to produce SED for a given SFH.
• Added 'nmodel' as a possible output column, showing the number of models fit to each galaxy.
• Added the NO_IGM option to disable IGM absorption.
• Added the SFH_QUANTITIES option to output non-parametric quantities of the SFH.
• When not using EAzY photometric redshifts, confidence intervals can now be freely chosen.
• Added option to fit log(LIR) instead of LIR, better for galaxies with uncertain Tdust.
• Added the INTERVAL_FROM_CHI2 option to compute conf intervals from chi2 grid directly.

Bug fixes:

• Made sure that attenuation law never goes negative.
• Added explicit check for galaxies with no valid data.
• Added the DEBUG option to help identify invalid models.
• Added explicit check to catch incorrect wavelength units in filters.
• Fixed crash for galaxies with spectra but no valid photometry (thanks to Hugh Dickinson).
• Fixed issues with unexpected line endings in input files.
• Fixed bug where it was not possible to use the 'mw' dust law (thanks brojonat).
• Fixed incorrect MC outputs for unsorted grids (e.g., metallicity, thanks Themiya Nanayakkara).
• Fixed missing column names on the .input_res.fit files (thanks Themiya Nanayakkara).
• Fixed MC conf intervals for gridded values close to grid edges (thanks Francesco Valentino).

Performances:

• Added the 'generators' parallel execution scheme, which should be faster in all cases.
• Improved performances of dust attenuation.
• Faster implementation of the fit progress bar.
• Improved speed of MAKE_SEDS by about a factor 5, and also enabled multithreading for this stage.
• Simplified IGM absorption formula.

v1.1 -- Improved robustness of filter reading code and fixed BEST_FROM_SIM

This release contains two main changes. First, the filter file is now read in a more robust way, so your custom FILTER.RES files that worked with FAST are also guaranteed to work with FAST++. Second, the BEST_FROM_SIM option now works correctly. In the previous version, using this option would produce correct fit parameters, but the chi2 would be infinite and the best-fit SEDs would be invalid. This is now fixed.

A smaller change concerns the MAKE_SED option. When enabling this option, the fitting stage is now disabled and the program only outputs the SEDs requested by the user. Grids and other output files are not modified.

v1.4 [LEGACY] -- Improved FAST compatibility and fixed BEST_FROM_SIM

The main changes in this version concerns how the program behave compared to the original FAST. The filter file is now read in a more robust way, so your custom FILTER.RES files that worked with FAST are also guaranteed to work with FAST++. The default AB zero point is switched back to 25 (the default in FAST, even though it does not make sense...). And finally, the handling of EAzY photo-zs is now by default the same as FAST (but it can still be changed).

The second most important change concerns the BEST_FROM_SIM option. In the previous version, using this option would produce correct fit parameters, but the chi2 would be infinite and the best-fit SEDs would be invalid. This is now fixed.

In addition, the input/output were vastly optimized (more than ten time faster), and several small fixes were applied to the terminal output messages.

NB: This is a "legacy" release. These versions of FAST++ are meant to reproduce the behavior of the old FAST and not introduce any other feature. If you want more fancy options and updated stuff, look for a non-legacy release.