Pachyderm 3.2.2

Release to align the version tag here with the one on PyPI.

Code the same as 3.2.1, but with updated dependencies in the lock file to have a trivial change

Pachyderm 3.2.1

Same as 3.2, but trying to fix the publishing workflow

3.2

What's Changed

• Accumulated analysis updates (mostly on binned_data and plotting)
• update to use scientific python cookie for repo structure

Note that the 3.1 tag was pushed, but a release wasn't initiated. Since we have new changes since then, just jump to 3.2

Full Changelog: 3.0...3.2

Pachydem 3.0

[3.0] - 1 November 2020

Added

• BinnedData class to generalize the Histogram1D class to general sets of binned data. It supports arbitrary number of axes, a substantially improved interface, and better maintainability. Note that this class doesn't yet have every feature from Histogram1D, but many of them are not necessary with boost-histogram available (some BinnedData features are inspired by bh). For example, we can convert to a bh hist to project, and then convert back (or do the summing directly - it's usually not difficult, except for profiles). As a notable new features, it has conversion to and from a wide variety of formats. This has been the basis for most of my analysis work this year. Written and improved over a variety of commits, starting with d7e5bf8.
• Helper function for plotting error boxes (refactor from alice-jet-hadron). See: 77ae89e.
• A variety of ALICE datasets for download.

Changed

• Move from setup.py to poetry. Starting from: e7a9391, with a ton of bug fixes.
• Support JAlien for ALICE download scripts. This deprecates support for the older alien. See: fba152a.
• General maintenance tasks for typing, TravisCI, etc.

Fixed

• A ton of small bugs.

Pachyderm 2.5

[2.5] - 5 December 2019

Added

• Support for multiplying or dividing by a numpy array. See: c418722.
• Extracting TDirectory objects from a ROOT file. See: d698fb4.
• Support downloading pt hard binned train output. See: f41606d.
• Some additional documentation.

Changed

• Improved support for failed copying from AliEn. See: 76c7aa7.
• Improved tests output. See: b4f599f.
• Improved LHC16j5 dataset definition. See: 9e3b5bd.

Fixed

• Filelist generation didn't always include the right files. See: 684c42c.

Pachydem 2.4

[2.4] - 4 December 2019

Added

• Tools for downloading ALICE datasets and train outputs in parallel. It can download real or simulated datasets, run-by-run LEGO train output, or any list of input and output files (for simpler cases, such as meta-dataset trains). Derived and generalized from code from M. Fasel. See: 76ce114ccff10e59330b541e38d685fc4aa7fb2d to 51c469b078e8a9fe25a980bc7c8b61fb9633f83a.
• Convert HEPdata files directly into Histogram1D objects. Call via Histogram1D.from_hepdata(...). Note that it doesn't fit nicely into the from_existing_histogram function because additional information is needed when reading HEPdata files. See: c79f40b5a9c6cb5a8b31bd35275004196a5d8228.
• Ability to convert matplotlib color schemes to ROOT. See: 292106009c9e9d6d5d47bf5648a9481c89b1d20c.

Changed

• Support passing Path objects for opening ROOT files. See: a2d644b6cab7c35fa15ba3614fac027599484b24.
• Update pre-commit hook versions. See: 2ad77745008ce81a62ba37282a35177e74e868fb.

Pachyderm 2.3.2

[2.3.2] - 21 October 2019

Fixed

• Improved typing for generic_config iterators of possible values. See: 797848212b03e8c36174cbb1a6b9bc1fda7b7739.

Pachyderm 2.3.1

[2.3.1] - 20 October 2019

Fixed

• Fixed typing for Histogram1D scalar division. See: 7cc7a7e.

Pachyderm 2.3

[2.3] - 16 October 2019

Added

• Scalar division for Histogram1D. See: c4230c6.
• All arithmetic operations for PDFs when fitting. This expands from just adding PDFs, which was all that was available up to now. See: 1799c9e.

Changed

• Make gradient calculation available through the package API. It can be useful for more complex error propagation. See: 6454c86 and 3891fed.

Fixed

• Binned log likelihood fit values, error matrix, covariance, etc, now agrees with ROOT. This is important for error calculation! See: dc3e1ff.
• Explicit weighted histogram option for binned log likelihood. See: dc3e1ff.
• Calculation of chi squared for goodness of fit when there are empty bins (i.e. for calculating the effective chi squared when performing a log likelihood fit). The empty bins are ignored. See: 0e66393.
• Typing for a variety of functions.
• Typos in documentation.
• Check for Histogram1D when attempting to create a new one. See: 62d8613.

Pachyderm 2.2.1

[2.2.1] - 7 September 2019

Added

• Added effective chi squared for simultaneous fit. See: 90a98d9. Bug fix: 92ac7b0.
• Write numpy floats to YAML. This helps to avoid a bunch of difficult to debug errors. See: 474a242.

Changed

• Improve minimization settings. See: 8bb1602.

Fixed

• Don't convert the histogram metadata into a numpy array. See: 2fafe77.
• Ensure that metadata is copied when copying a histogram. See: f1d1fbf.