13.22 Highlights of What’s New in Version 34.3.0, June 14, 2018

Here is a summary of what’s changed since release 34.2.0.

13.22.1 Bug Fixes for Elements

• The CCBEND element had incorrect signs for the odd-order systematic multipoles when the bending angle was negative. There was also an issue with incorrect ordering of edge effects and coordinate transformations. R. Lindberg (APS) helped identify these problems.
• The KQUAD element had an issue with the order of the submatrices used for linear edge effects when tracking. This would cause small tune errors in tracking compared to the results of twiss_output.

13.22.2 Bug Fixes for Commands

• The beta-function scaling resulting from the twiss_scaling=1 setting for the elastic_scattering command was incorrect. Although the tracking results would be correct, this made it more difficult to optimize run time.
• The rf_setup command now handles α_c < 0, a deficiency that was reported by P. Piot (NIU/FNAL). It also now uses η = α_c - 1∕γ² instead of α_c, though this rarely makes a significant difference.
• The computation of exact normalized emittances, requested with the global_settings command, had several issues that were fixed. First, the values assigned to the horizontal and vertical corrected and uncorrected emittances were permuted. Second, in the serial version, the corrected emittances (with dispersive terms removed) were computed incorrectly. This would impact the sigma and final files from the run_setup command. J. Björklund Svensson (MAX-Lab) reported problems that helped find these bugs.

13.22.3 New and Modified Elements

• CCBEND has a new parameter, EDGE_ORDER, that allows controlling the order of edge kicks.
• BRANCH has a new parameter, DEFAULT_TO_ELSE, which allows determining how the element behaves when tracking for closed orbits and the like.

13.22.4 New and Modified Commands

• The configuration command-line argument was added, which allows specifying a configuration file to be read before processing the input file. This file can also be specified with the ELEGANT_CONFIGURATION environment variable.
• The floor_coordinates command now creates additional columns in the output file, giving data on the next element in the lattice. Given that data is only provided at the end of elements, this provides an easier way to determine information at the start of elements.
• The insert_elements command now has a parameter, insert_before, that allows controlling whether elements are inserted before or after (default) the specified locations.
• The elastic_scattering command now includes data that indicates warning conditions in the file specified by log_file.

13.22.5 Changes Specific to the MPI Parallel Version

• Read buffering was re-enabled for parallel I/O to avoid performance problems on GPFS file systems. Write buffering is still disabled, since this seems to prevent data corruption on some file systems. Users may wish to configure this for their file system using the new ELEGANT_CONFIGURATION environment variable.

13.22.6 Changes Specific to the GPU Version

The GPU version continues to be an alpha release and contains bugs. Users are encouraged to check results against the serial or parallel versions and report issues to the developers.

• None.

13.22.7 Changes to Related Programs and Files

• The scripts elasticScatteringAnalysis and inelasticScatteringAnalysis—which are used to analyze data from the elastic_scattering and inelastic_scattering commands—were replaced with compiled programs of the same name, giving a large reduction in run time.
• The script longitCalcs, which does rf calculations using an output file from twiss_output, now supports a commandline mode that is convenient for use in other scripts.
• The script computeQuadFringeIntegrals was added. It computes the fringe integrals and effective length for a quadrupole from gradient vs z data, producing a file suitable for configuring KQUAD elements.
• A bug was fixed in elegant2astra that would affect results for particles that are not highly relativistic. Forum user Biaobin reported the bug and provided the fix.
• A bug was fixed in smoothDist6s that resulted in strange longitudinal phase space when the average value of t was very large compared to the spread in t. Forum user Marcello reported the bug.