-
Notifications
You must be signed in to change notification settings - Fork 17
Release Notes
The release of each version of the Fermitools will be announced on the fermi-soft mailing list. Please see the FSSC's instructions on subscribing to it and other Fermi mailing lists that may be of interest.
This release is primarily aimed at solving problems that have made the Fermitools difficult to build and led to installation problems and lengthy installation times due to dependency resolution and dependency conflicts. These goals were primarily achieved by removing CERN's ROOT package from the Fermitools, which was the source of many of these issues.
-
Removed Root and Root-dependent features
- Standalone MINUIT2 optimizer implemented
- Skymaps package removed
- Pulsar and GRBObs models have been removed from gtobssim
- Astropy will default to the latest version (currently 5.1 on release)
- Python upgraded to 3.9
- Native ARM builds for Apple M1 processors available
- libCLHEP: Cannot open shared object file error fixed (Issue #107)
- ModelEditor Tk callback error fixed (Issue #97)
- Padding of PSF containment for diffuse sources adjusted to address (Issue #103
- Fixed issue with gtburst overwriting files when downloading multiple files from the LAT data server (gtburst PR #7)
- PFILES environment variable now sets correctly in csh activation script
- All spectral models available in Likelihood are now accessible in pyLikelihood (https://github.com/fermi-lat/Fermitools-conda/issues/66)
- New recommendations in [Troubleshooting] (https://github.com/fermi-lat/Fermitools-conda/wiki/Troubleshooting)have been made to resolve issues with dependency management hanging (https://github.com/fermi-lat/Fermitools-conda/issues/100)
- Import errors related to a CLHEP version mismatch have been fixed (https://github.com/fermi-lat/Fermitools-conda/issues/100)
- An issue with GuassianError in logLikelihood apply an incorrect offset has been fixed (https://github.com/fermi-lat/Likelihood/issues/89)
- Floating point exception that occurred when gtsrcmaps encountered a pixel with zero exposure has been fixed (https://github.com/fermi-lat/Likelihood/issues/83)
- Issue with setting PFILES path in the csh activation script has been fixed (https://github.com/fermi-lat/Fermitools-conda/issues/96)
- The Fermitools have been updated to use Python 3. Python 2 has been sunset and is no longer supported.
- The CALDB has been updated to add P8R3_V3 IRFs.
- These IRFs use an in-flight calibration method to change the PSF event type effective area (as a function of energy and incidence angle) in order to improve the agreement between the observed and predicted numbers of events.
- Please see that caveats at https://fermi.gsfc.nasa.gov/ssc/data/analysis/LAT_caveats.html.
- Removed behavior that would automatically change selected IRFs to the latest version if only the version string had changed (e..g., change a call to P8R3_V2 IRFs to P8R3_V3). Added a warning if you are using an old version of the IRFs.
- Added the PLSuperExpCutoff4 model.
- ROOT has been updated to ROOT6. ROOT5 support has been deprecated.
- Fermitools now compiled with Conda provided compilers (
gcc 7.5.*
andClang 9.*
)-
condaforge/label/cf201901
label replaced withcondaforge
in the Fermitools install command
-
- Unnecessary warning messages about unspecified parameters have been fixed
- Error thrown when attempting to use weighted likelihood with single energy bin analysis has been fixed
- Conda FFTW3 headers are now handled properly
- A bug creating extra unneeded keyword headers has been fixed
- Overflow handling in the PLSuperExpCutoff3 model has been improved
- Fixed error caused by nonvirtual destructor definition in FitScanner (https://github.com/fermi-lat/Likelihood/pull/78)
- Par files for gteffbkg, gttsmap, gttscube, and gthealcube now properly include the edisp_bins parameter (https://github.com/fermi-lat/Likelihood/issues/55)
- Bug fixed in Gaussian Error that was causing an issue with offsets in logLikelihood (https://github.com/fermi-lat/Likelihood/issues/89)
- Issue with creating SEDs in unbinned mode fixed (https://github.com/fermi-lat/pyLikelihood/issues/16)
- Problem restoring likelihood objects in Python interface fixed (https://github.com/fermi-lat/Likelihood/issues/76 and https://github.com/fermi-lat/pyLikelihood/issues/19)
- Failure of gtexpmap fixed (https://github.com/fermi-lat/Likelihood/issues/70 and https://github.com/fermi-lat/Likelihood/issues/84)
- NewMinuit can now report fit convergence status similar to that for Minuit (https://github.com/fermi-lat/optimizers/issues/6)
- Added new PLSuperExpCutoff3 model.
- Added support for RadialDisk model to ModelEditor (https://github.com/fermi-lat/Fermitools-conda/issues/51)
- Fixed convergence issues reported in (https://github.com/fermi-lat/Likelihood/issues/51)
- RHEL6 support restored (https://github.com/fermi-lat/Fermitools-conda/issues/50)
- Parameter files have been updated to properly handle Energy Dispersion update options.
- A bug which was preventing updated parameter files from installing has been fixed.
- The release contains a major update to the implementation of the energy dispersion correction method in the Fermitools. It can now take into account the interplay between the energy dispersion and PSF. Please read the usage recommendations and details and the caveats below.
- A new tool called gtdrm has been added. It estimates the energy dispersion matrix for the energy binning that you are using and can be used to evaluate the size of the effect and help you decide how many bins to use.
- Updated the fhelp files to add missing parameters.
- Restored the "modeleditor" wrapper around ModelEditor.py that was missing from previous Fermitools releases.
- The gtobssim cannot use the energy dispersion correction method implemented in this release but can still use the previous correction method.
- Any gtlike and pyLikehood analysis with edisp_bins > 0 using a source map file or exposure map file that does not include the extra energy bins will abort and give you a message explaining that you need to remake the offending file.
- You may encounter a "missing parameter" error due to the addition of the edisp_bins parameter in several likelihood tools. You should remove the old parameter files (usually in your $HOME/pfiles directory). Running the tool again will install an updated parameter file.
- Added PLSuperExpCutoff2 model to modeleditor (https://github.com/fermi-lat/modelEditor/issues/2)
- Fixed critical interface issue with Fermitools version reporting in pyLikelihood (https://github.com/fermi-lat/Fermitools-conda/issues/37)
- Fixed compilation error in the SolarSystemTools (https://github.com/fermi-lat/SolarSystemTools/issues/2)
- Fixed ST_Version method to work correctly with Fermipy in pyLikelihood (https://jira.slac.stanford.edu/browse/LK-160)
- Added PLSuperExpCutoff2 model to gtexposure (https://github.com/fermi-lat/pyExposure/issues/1)
- Fixed "Wrong number or type of arguments" error in BinnedAnalysis (https://github.com/fermi-lat/Likelihood/issues/41)
- Fixed activation problem for tcsh and csh with recent conda release (>4.6.12) (https://github.com/fermi-lat/Fermitools-conda/issues/34)
- gtsrcmaps has been patched to prevent PSF extrapolation from going outside of range near the edge of large regions of interest
Due to conda-forge changes, the label cf201901 is currently required during installation to avoid mismatched symbol errors. See the Installation Instructions and User Notes for more details. We hope to make this unnecessary for the next release of the Fermitools.
- Likelihood:
- SummedLikelihood handling of priors has been improved.
- ModelMap file writing has been made more efficient.
- apply_psf and apply_exposure options moved to XML SpatialModel block.
- Added GaussianError class to as a possible implementation of priors on parameters.
- Switches added to turn on/off PSF convolution and exposure on a source by source basis.
- gtmodel : Added phased_expmap parameter.
- pointlike: A number of updates to bring it up to date with the latest development changes.
- Implemented various C++17 standard compliance updates.
- gtsrcmaps now generates properly filled source maps.
- Likelihood: EblAtten bug fixed. Now the state of the underlying parameters is latched befored cloning.
- gtexpcube2: HEALPix ordering scheme in has been fixed.
- Astro package: Fixed error in Geodetic coordinate calculation.
- Initial release
- Name change: The Fermi Science Support Center has opted to rename the version of the Fermi ScienceTools that is released via the Conda package management system 'Fermitools'. This is to deliniate between the old binary/source tarball download releases (which will remain 'ScienceTools') and the Fermitools conda package.
- The primary goals of this release are to A) Transition to the Conda-based distribution system and B) Release required irfs and isotropic templates for use with the concurrently released P305 data set.
- New IRFs and Isotropic Templates are included in this release for use with P305 data.
- This release does NOT have updated Galactic Diffuse Models. The FSSC recommends continuing to use the current galactic diffuse model (gll_iem_v06.fits). This model can be obtained from the FSSC website or from the fermitools-data conda package.
- A number of compatibility changes and bug fixes were made to the codebase to accomodate the Conda distribution system transition.
- The ScienceTools/Fermitools codebase is now accessible via fermi-lat github.