Releases: viperleed/viperleed-betatest
v0.12.0
Note: This release is mainly for archival purposes and can be considered a "Golden Master" version before the soon-to-be released v1.0.0.
Note: With the release of viperleed.calc v1.0.0, this beta tester repository will be considered deprecated in favor of the main repository. Note also, that this will be the last release for which executables for viperleed.calc
and the bookkeeper
utility are provided as going forward it will be possible to install (and update) ViPErLEED via pip install viperleed
.
What's Changed
- The Installable viperleed branch by @amimre has been merged (see viperleed/viperleed#94). This makes viperleed pip installable and brings a major refactoring of the project structure.
- Improve utilities by @michele-riva in viperleed/viperleed#196
- Compatibility with new TensErLEED organization, version numbers and TensErLEED v2.0.0 by @amimre in viperleed/viperleed#207
- Major updates to the docs with the installable branch.
- TensErLEED is no longer contained in the main viperleed repository. Instead, it is hosted on https://github.com/viperleed/viperleed-tensorleed
Full Changelog: viperleed/viperleed@v0.11.0...v0.12.0
Release v0.11.0 (tleedm, TensErLEED)
Note: This release is mainly for archival purposes. We have been developing the installable version of ViPErLEED (#90) in parallel and will soon merge that branch (#94) into master. A release v0.12.0 will then follow as a "Golden Master" before publication.
tleedm v0.11.0
- Refactor and improve POSCAR read/write (#91)
- Refactor and improve Slab (#114)
- Update license (#93)
- warn for ambiguous PHI (#71)
- utility to write file for VASP (#98, expanded on in the installable branch)
- More general PoscarReader/Writer and some refactor (#100)
- Fix beamgen (#106)
- Refactor tests (#102)
- Add some more structure to classes.layer (#110)
- Partial cleanup of Atom (#112)
- Ensure compatibility with python v3.7 (#126, #173)
- Fix search segfault by (#128)
- Faster tests by (#127)
- Refactor parameters file (#130)
- Fe2O3 Example (#177)
- Fix #187
TensErLEED v1.75 & v1.76
Note: TensErLEED and everything in the viperleed/tensorleed directory will be moved to a separate repository with the next release.
- Fixes and error handling (writing error messages to where one would expect them).
Release v0.10.0 (tleedm, TensErLEED, Spotracker)
tleedm v0.10.0
- fixed slab transformation (#76)
- input files saved to history are now unmodified version from work/original_inputs
- fixed race condition in structure optimization when control.chem is read during write
- implemented more consistency/syntax checks for PARAMETERS
- unrecognized parameters raise error instead of being silently ignored
- implement a style sheet with font type 42 by default (#85, #77)
- better unit tests for ASE interface
- unit tests for parameters
- replaced the Fortran beamgen script by a Python version (#60)
- updated viperleed conda environments
TensErLEED v1.74
This release contains a minor patch to TensErLEED v1.74 which only relates to error messages.
The previous TensErLEED version will work, but some rare errors may not be caught by ViPErLEED.
ImageJ Spotracker
- TODO: changelog
tleedm v0.9.0
tleedm v0.9.0:
- reworked the output of the error calculation and added estimates of statistical errors of parameters.
- added parameter "ZIP_COMPRESSION_LEVEL".
- various clarifications in the documentation.
- catch IEEE errors in superpos output.
- changed SD.TL output interval to fit parameter dgen.
- improved PHASESHIFTS consistency checks.
- suppress warning for non-available f2py package.
- various bug fixes.
Release v0.8.0 (tleedm)
tleedm v0.8.0:
- completely overhauled the documentation. Added missing pages, sections and example systems.
- updates to the ASE interface.
- various fixes related to plotting.
- fix level R+var(R) in Errors.pdf.
- fixed issue with gfortran compiler options.
- add option for checksum computation of TensErLEED code (and parameter TL_IGNORE_CHECKSUM).
- fix issue with LMAX > 16 in delta-amplitudes and search.
- fix various issues for importing tleedm.
- better handling of compile logs.
- fix mix ups in symmetry reduction.
- improved log messages.
- removed case-sensitivity in element names for consistency.
Release v0.7.2 (tleedm, TensErLEED, ImageJ plugin)
Edit: add TensErLEED source to assets.
- job.py can accept ViPErLEED source code and working directories via command line arguments.
- Include .yml files with recommended conda environments for running ViPErLEED calculation.
- New version of R-factor calculation with a compiled python module. This is in testing stage and can be enabled with the option R_FACTOR_LEGACY = False.
- Added a Makefile for beamgen and EASiSSS.
- Added an API to call ViPErLEED refcalc from ASE atoms objects and calculate R-factor between two THEOBEAMS.csv files. (not part of tleedm)
- Enabled more general unit cell operations on slab objects.
- Fixes issue in symmetry reduction for hexagonal to cm.
ImageJ plugin:
- Added macro scripting capabilities.
- Enabled reading I_00 from dark frames.
- More help buttons.
- Ability to set parameters from log file of previos Spot Tracker output.
- Various bug fixes.
Release v0.7.1 (tleedm)
Updates tleedm. ImageJ plugins, bookkeeper, documentation and GUI are included for completeness, but are identical to previous versions.
- Switched default separator in CSV input/output files back to comma (from semicolon). Semicolons will still be accepted in input files.
- Fixes a major bug in applying unit cell scaling during full-dynamic optimization
- Fixes some issues with saving fortran-compile.log files for diagnosis
- Reduced angle tolerance for unit cell type detection and increased code consistency. This prevents some incorrect assignment of rhombic cells as hexagonal. However, this should not have caused any problems aside from incorrect cell types reported in the log files.
- Fixed an issue that made it impossible to use the control.chem file as a starting configuration for a new search, because old output files were cleared from the working directory at the beginning of tleedm execution. Output files that can also serve as input (control.chem, refcalc-fd.out and superpos-spec.out) are now kept until needed, but still deleted at the start of the respective segments to avoid mistaking them for new output.
- DISPLACEMENTS direction statements are now more tolerant to whitespace, e.g. '[ 1 -1]' or '[ 1 -1 ]' are accepted in place of '[1 -1]'.
Release v0.7.0 (tleedm, bookkeeper)
Update tleedm, bookkeeper and documentation. ImageJ plugins and GUI are included for completeness, but are identical to previous versions.
- Added segment "full-dynamic optimization", which iterates "reference calculations" (without Tensor output) to optimize parameters not accessible by Tensor LEED search (i.e. beam incidence angles, V0i, unit cell size). Behaviour is controlled by new parameter OPTIMIZE. See documentation for details.
- THEOBEAMS.csv are now also output as plots in THEOBEAMS.pdf (raw data without smoothing or interpolation).
- Slightly modified storage of original input files and fortran compile logs in both tleedm and bookkeeper. This should reduce the potential for user behaviour affecting the "history" folder. For example, "original input" from tleedm execution is now stored in a dedicated subfolder in SUPP such that even if input files are modified before the bookkeeper is run, it is still possible to reconstruct the input files used for generating a specific output.
Release v0.6.2 (tleedm)
Patches tleedm and updates documentation. Bookkeeper, ImageJ plugins and GUI are included for completeness, but are identical to previous versions.
Features:
- Added parameter SYMMETRY_BULK, which allows manually setting a bulk symmetry other than the automatically detected one. See documentation for details.
Debugging:
- fixed a bug that prevented mpifort compiler from being recognized
- fixed an issue that sometimes crashed the search if SD.TL was written at an inconvenient time (mainly on macOS)
- fixed an issue with cleanly terminating the search through user intervention (sometimes leaves zombie processes)
- fixed an issue when reading refcalc output: TensErLEED sometimes produces unusual formatting (seen only on macOS)
Release v0.6.1 (tleedm)
Patches tleedm and updates documentation. Bookkeeper, ImageJ plugins and GUI are included for completeness, but are identical to previous versions.
Features:
- added an optional parameter BULK_LIKE_BELOW that can be used to automatically detect N_BULK_LAYERS, LAYER_CUTS and BULK_REPEAT from the POSCAR file. See documentation for details.
- added a parameter AVERAGE_BEAMS that allows switching off the averaging of equivalent beams, or to average according to a different beam incidence than given by BEAM_INCIDENCE. Note: Correct averaging for off-perpendicular beam incidence is currently not supported, but will be fixed in an upcoming release!
- RUN parameter now allows string aliases for the different segments, e.g. "ref" for reference calculation, "init" for initialization, etc. See documentation for details.
Minor changes:
- turned PARABOLA_FIT off by default (as it affected performance and was not very reliable)
- made POSCAR output preserve the scaling factor of the input POSCAR, instead of always setting scaling to 1.
- Input files PARAMETERS, DISPLACEMENTS and VIBROCC now accept characters "%" and "#" to indicate the beginning of a comment, in addition to "!"
- changed default of ATTENTUATION_EPS to 0.001
- checked the non-parallelized search modules of TensErLEED, and made ViPErLEED use those for single-node jobs if no fortran MPI compiler is present. Therefore, it is now possible to run ViPErLEED for simple systems without installing an MPI compiler, although this is not recommended for serious operation.
Debugging:
- fixed a bug in ViPErLEED that led to variations of concentrations being ignored in DISPLACEMENTS if the initial site occupation was mixed
- fixed a bug that led to incorrect comparison of experimental and theoretical beams in R-factor calculation if experimental beams were dropped (due to missing theoretical counterpart)