Version 1.5.0

New features

• Support for outputting to HDF5 format, specifically for initialising Gadget (and variants) or Swift simulations (#123, #124)
• Allows genetIC to be used with float32 (not double aka float64) throughout; pass -f on command line; #118
• Adds memory reporting #118
• Add possibility to use cuFFT to perform FFTs by @cphyc in #119

Full Changelog: v1.4.0...v1.5.0

Version 1.4.0

What's new

• Improvements to writing out different gadget particle types: #106, #113
• Slightly more descriptive error when insufficient transfer function points are available: #108

Full Changelog: v1.3.5...v1.4.0

Version 1.3.5

Bug fixes

This release fixes a performance issue with some network file systems, where file writing could become extremely slow. #103

Version 1.3.4

What's Changed

The baryon_tf_on command has been disabled. This is due to discovering a significant bug in the redshift evolution when using a separate baryonic transfer function. A fix for this will be provided in a new version in due course.

Version 1.3.3

The code for v1.3.3 is unchanged with respect to v1.3.2.

The only difference is that the default optimisation in the Makefile is now -O3.

Docker images use the default Makefile flags, and therefore were running very slowly.

Full Changelog: v1.3.2...v1.3.3

Version 1.3.2

Fixes

• Fix bug that prevented opening a zoom grid with zoom_fac=1 (@apontzen)
• Output gadget internal energy array for SPH particles when OmegaB > 0 (@apontzen)
• Fix bug in gadget particle ID output, where sometimes dark matter and SPH particles would have the same IDs (leading gadget to crash on start-up) (@apontzen)

Enhancements

• Clearer warning for when the number of gadget_num_files is too small (@lluciesmith)

Full Changelog: v1.3.1...v1.3.2

Version 1.3.1

Bug fixes

• Fixes performance issue when using lustre (and possibly other network) filesystems.

Enhancements

• Minor improvements to output when stdout is not a TTY, automatically replacing progress bar with logging.

Version 1.3

New features

• Introduce 'gene splicing', i.e. the ability to cut together different realisations of the overdensity field. An accompanying paper will be posted to arXiv shortly (Cadiou, Pontzen, Peiris & Lucie-Smith 2021).
• Add ability to output multiple gadget files (essential for large ICs due to limitations of the gadget format)

Bug fixes

• Addresses issue #90, where placement of zoom windows near the edge of the box could result in errors
• Addresses issue #85, where Zel'dovich displacement calculations could cause a bounds overrun error

Version 1.2

New features

• Ability to modify angular momentum (modify Lx, Ly, Lz)
• Ability to select elliptical region (select_ellipse)
• Ability to take gradients for the Zel'dovich approximation using a local 4th-order stencil instead of in Fourier space (disable the ZELDOVICH_GRADIENT_FOURIER_SPACE flag)

See the updated user manual for more information.

Bug fixes & enhancements

• Fix a segmentation fault which arose in recent versions of GCC with -O2 or higher due to accidentally returning a reference to a local variable
• Remove some unused variables which led to compiler warnings (@cphyc)

Version 1.1.1

Enhancements

• Significantly improves readability of logging output
• Now ensures unused parts of gadget headers are always initialised to zero
• Cosmetic improvements to outputs of the check_zoom_accuracy.py script (used for development purposes only)

Bug fixes

• When creating an input mapper, some commands in the input paramfile were executed when they should be ignored (fixed by @Martin-Rey)
• When opening double zooms near the edge of the fundamental domain, the second-level zoom was sometimes wrapped out of the coordinates of the first-level zoom, leading to an error
• Fix misnomer in deprecation warning for append_idfile