CHANGELOG

SIRE changelog
==============

devel branch:

    [2021.1.0] Feb 2021 - Added support for multiple combining rules in SOMD
               (@SofiaBariami). Added support for triclinic simulation boxes.
               Convert Ryckaert-Bellememans form dihedral functions from
               GROMACS to Fourier series to allow conversion to AMBER format.
               Updated search functionality to enable searching for objects
               within an arbitrary distance of a point. Fixed PDB2 parser bug
               to ensure that residue names are fixed width. Ensure that
               NUMEXTRA pointer is written so that AMBER topology files can
               be read by tools such as ParmEd. Write NATYP pointer and
               correct number of SOLTY flags. Even though these aren't used,
               incorrect values break external tools, e.g. ParmEd. Added
               support for AMBER TIP5P water topology conversion. Correctly
               flag OPLS style force fields when creating MMDetail object so
               that users can reconstruct OPLS systems written to AMBER format.
               Made build Python 3.8+ compliant. (Python libraries are now
               ABI compatible.) Switched to using std::atomic since
               tbb/atomic.h is now deprecated. Switched to using HTTPS for
               sending analytics. Updated build to be able to link against
               conda version of libcpuid. Added support for generating PDB
               CONECT records from a Sire.Mol.Connectivity object.

    [2020.1.0] July 2020 - Fixed bug in WaterView program to ensure that a
               molecule is extracted from the returned list. Stable sorting
               of dihedrals and other potential terms to allow reproducible
               writing of input files for SOMD (@ptosco). Updated the
               FreeEnergyAnalysis script to support different versions of the
               pymbar API. Significant performance improvement to the GroTop
               parser by looping over cut-groups during non-bonded matrix
               evaluation. Updated Miniconda and conda dependencies to latest
               cross-compatible versions. Fixed minor copiler and runtime
               issues (@nigel-cresset).

    [2019.3.0] November 2019 - Added functionality to restrict the search space
               when finding paths between atoms or searching for rings. Fixed
               performance issue in GroTop parser caused by an N^2 loop over
               atoms when searching the intrascale matrix. We now loop over
               cut-groups, which is far more efficient. Fixed issues with
               Python wrapper generation caused by issues with missing define
               symbols and a bug in the scanheaders.py script.

    [2019.2.1] October 2019 - Updated the Conda recipe to pin the dependencies
               of dependencies that are used at run time since Conda doesn't
               automatically do this for you. Added instructions detailing the
               Azure Pipeline build process and how to create a new release.

    [2019.2.0] September 2019 - Updated the Gromacs topology writer to support
               perturbable molecules containing a variable number of bonds.
               Created a Docker container for building wrappers and updated
               to using CastXML. Added support for running background
               processes on Windows (@ptosco). Updated SOMD Python wrapper
               to write restart files every cycle to simplify system monitoring
               in BioSimSpace. Fixed macOS build issue by not linking against
               libpython. Made sure that Conda dependencies are pinned
               correctly to avoid compatibility issues. Fixed bug that
               prevented upload statistics being sent and added support for
               tracking BioSimSpace usage.

    [2019.1.0] May 2019 - Updates to the Gromacs topology writer to support
               free energy perturbation simulations. The MCS matching
               functionality has been extended to allow matches between heavy
               and light atoms, and the ability to return all current matches,
               rather than just the most recent. Temporarily disabled
               parallelisation in the AmberPrm parser to avoid a threading
               issue. Switched to using Azure Pipelines for continuous
               integration to enable a fully automated build, testing, and
               deployment pipeline. In addition, we finally have created
               a Sire Conda package to simplify the installation and update
               process.

    [2018.2.0] July 2018 - Improvements to the Gromacs topology reader/writer,
               addition of code to improve matching of atoms in proteins,
               fixing compile issues on modern Ubuntu, bugfixes for crashes
               in the AmberPrm reader, added in text-based searching for
               atoms, residues etc. from Systems, MoleculeGroups, Molecules,
               etc. based on boost::spirit, updated boost to latest version,
               bugfixes for quantomm infinite rotation bug for ions,
               general bugfixes.

    [2018.1.1] May 2018 - Small bug fixes to allow single-atom solutes
               and also to fix small issues with some parsers for BioSimSpace

    [2018.1.0] March 2018 - Signficantly improved the Gromacs and Charmm
               parser and  fixed bugs. Can now write with both :-). Fixed
               compilation on Windows 7 and above. Small changes to
               the API of AtomProperties to make them easier to work
               with from Python. Added a script to automatically color
               code swap-based free energies. Fixed localisation
               problems for the PDB writer. Improved mbar analysis
               code. Added code to track forcefield data of a molecue.
               Added code to better manage processes, including redirection
               of standard output and error.

    [2017.3.0] December 2017 - Added new PDB (PDB2), Mol2, Gro87, and CharmmPSF
               parsers, as well as a GroTop parser, all part of the new
               MoleculeParser framework. Updated all of the swap-based
               methods to use this.

               Removed the ViewsOfMol Python wrapper and now have the
               code automatically return the correct python object for
               the molecule (or part of molecule) that is returned from
               the system. This makes simple scripts easier to write.

               General bugfixes and optimisations, including fixing
               bugs with the way that PropertyMap worked, cleaning
               up to/from converters from python objects to automatic
               Property wrappers, and fixing Process so that it can
               redirect to stdout and that isRunning works without the
               user having to call "wait" first!

    [2017.2.0] September 2017 - The MoleculeParser framework has
               been created to support reading and writing of molecules
               in lots of different formats. The first set of formats
               that have been completed are Amber Prmtop, Amber Rst7
               and Amber Rst/Trj. The parsers work in parallel, with
               file formats automatically detected by the parser,
               e.g. system = MoleculeParser.read( "file.prm", "file.rst" )
               will automatically do the right thing.

               Improved automatic compilation on Arch linux.

               Fixed temperature checking and general bugfixing for mbar code.

    [2017.1.0] April 20 2017 - First 2017 release. Included new
            parallel MoleculeParser code for reading molecules,
            and moved fully over to C++-14 style coding for new code.
            Included AVX-512 vectorisation for Intel KNL and can
            now successfully compile and run using GCC 5 and GCC 6,
            as well as Intel 2017 compilers and Clang.

    [2016.3.1] January 9 2017 - Minor patch release that fixes bugs:
        (1) Correctly sets MACOSX_DEPLOYMENT_TARGET to 10.8 so executable works
            on OS X 10.8 (Mountain Lion) and above
        (2) Fixed a parsing bug in Parameter that prevented integer or float
            parameters from being passed to ligandswap, waterswap etc.
        (3) Fixed a small bug in MultiDouble that meant it lost precision when
            swapping individual values
        (4) Fixed a parsing bug in Parameter that meant that windows path names
            were not interpreted correctly
        (5) Fixed the build scripts so that they placed bundled libraries into
            bundled/lib rather than bundled/lib64 (affected SUSE-based distributions)

    [2016.3.0] December 22 2016 - Public release containing full LigandSwap. Uses
     new optimised forcefields for energy calculations, built on top of Intel Threaded
     Building Blocks for parallelisation. New code is significantly faster with better
     scaling.

    [2016.2.0] June 3 2016 - Semi-private release for the CCP-BioSim workshops. Included
     the first version of LigandSwap and general bug fixes

    [2016.1.0] April 29 2016 - Merge of Bristol and Edinburgh codes, moved to miniconda
     and clean packaging system, including OpenMM fully, added in nautilus, somd etc.,
     added optimised forcefields, added a proper unit testing suite.


    [OLD] Updated gradient compuation in openmmfreenergst to finite differece gradients
    Allow the computation of reduced perturbed energies in openmmfreenergst of all computed lambda values
    Separated minimization and equilibration from production run.
    Implemented mass repartitioning for hydrogens atoms to allow for larger integration timesteps
    Added nautilus scripts