Skip to content

v1.4.0

Latest
Compare
Choose a tag to compare
@timcallow timcallow released this 09 Nov 10:24

Changelog:

Bugfixes (code)

  • Error in calculation of Hartree potential fixed (incorrect synchronization of lower and upper integrals): means fewer grid points required for convergence (#196)
  • Error in the matrix solver which makes the initial guess for the eigenvalues fixed (#199)

Bugfixes (other - test suite, CI, etc.)

  • Rename tests so that pytest sees all tests (#204)
  • Make mixfrac explicitly re-defined in each test, as this was causing occasional failures (#206)
  • Update package requirements in docs to make them more up-to-date (#210)

New features and improvements (code)

  • Square-root grid implemented: usually requires many fewer points for converged calculations (#179)
  • Explicit use of diagonal matrices in numerov solver: yields massive speed-ups and reduced memory requirements (#194)
  • Speed-up for Hartree potential calculation (#198)
  • Update to a better set of defaults (#201)
  • Allow faster convergence after 2 SCF cycles (#205)
  • Add method B for virial pressure in kinetic energy calculation (#180)
  • Allow suppression of some warnings (#208)
  • Update stated temperature range of atoMEC (#209)

New features and improvements (other - test suite, CI, etc.)

  • Change installation of libxc: we now install from the official libxc latest stable tarball, which means atoMEC can be run on more Python versions (and we don't rely on an unofficial wheels distribution) (#223)
  • Update default Python version to 3.12: it was noticed during benchmarking that atoMEC runs ~20% faster with Python 3.12 compared to Python<=3.11 (#212 - just the final commit)
  • Introduce (optional) development tests to benchmark performance (#215)
  • Update out-of-date information in the README (#214) and improve it (#188)
  • Improve manual running of test suite (i.e. running without pytest) (#192)

Full Changelog: v1.3.0...v1.4.0