Skip to content

Releases: william-dawson/NTPoly

Version 3.1.1

01 Oct 08:13
20f55fe
Compare
Choose a tag to compare

Minor release for compatibility with other packages.

Features:

  • Subroutine to scale a matrix by a diagonal matrix.
  • Matrices to help with preserving symmetry (test symmetry deviation, symmetrize matrix)
  • Added a subroutine test_ntpoly_link that sits outside of any module, that can be used to test linking.

Bug Fixes:

  • unused be_verbose argument in process grid constructor removed

Version 2.7.2

29 Feb 12:12
Compare
Choose a tag to compare

Minor update to version 2.7.2. This release backports some critical fixes from the version 3 series (#231).

Version 3.1.0

29 Feb 12:33
b1ca2f5
Compare
Choose a tag to compare

Features:

  • Small benchmark driver
  • Some automatic convergence detection
  • Rudimentary method to estimate homo-lumo gap
  • Python documentation

Improvements:

  • Memory optimizations in multiply
  • Early exit for WOM
  • Intermediate energies printed by WOM (for heat capacity estimate)
  • Write comment feature for the logger
  • Merge triplet lists function

Bug Fixes:

  • Fix use of fractional charge in dense eigensolver
  • Extra warnings added to the debug setting, with appropriate fixes
  • When number of matrix elements couldn't fit in int4, there was a bug
  • TRS4 had some issues with division by zero for the exact solution
  • Enforce F2003 standard

Check BigDFT 3.1.0 Pre-release

15 Nov 08:13
95b3ba3
Compare
Choose a tag to compare
Pre-release

Pre-release for testing with BigDFT

Version 3.0.0

23 May 13:53
19afc5f
Compare
Choose a tag to compare

Version upped to 3 because of a significant change to the density matrix solver interfaces (not because there are a ton of new features). Basically, instead of an integer for the number of electrons, it now takes in a float which is the trace of the density matrix you want. This clarifies some of the factors of two issues, and leaves room for partial occupancy.

Features:

  • Floating point number for number of electrons
  • IsIdentity routine
  • Similarity transform helper
  • McWeeny step helper
  • Reduce dimension feature
  • Maps with integer optional data
  • Should avoid assign alloc everywhere, including data types, using proper copy routines for all derived types.
  • Print the process grid information.
  • Fix integer out of bounds potential issue in I/O.
  • Eigensolver based dense fallback for most routines.
  • Wave operator minimization method.

Check BigDFT 2.8-4

26 Apr 06:04
b5fa45d
Compare
Choose a tag to compare
Check BigDFT 2.8-4 Pre-release
Pre-release
Code uniformity changes (#200)

* Cleanup for triplets/timers/loggins/errors

Credit to ChatGPT

* Copy/construct pattern for solver parameters and permutations

* Cleanup + Examples

* Linting

* Example fixes

* Variable fixes

* Remove unused variable

* Comment typo fix

credit chatgpt

* Fix interface issue

* Additional cleanup

* Remove timers

* Fix missing in/out intent

* Explicit copy routine for triplet lists

* Additional include

* Don't used fixed shape

* Split up the multiply

* Generic error should be public

* Cleaner implementation of dense conversion

* Spacing

* Cleanup the matrix reduction module

* More assorted cleanup

* More cleanup

* Spacing fixes

* Last set of spacing issues

* lint

* lint

Check bigdft 2.8-3

07 Apr 06:37
Compare
Choose a tag to compare
Check bigdft 2.8-3 Pre-release
Pre-release
ntpoly_2.8_bigdft3

Reinforce the pattern of destructing at the start of a constructor

Check bigdft 2.8-2

01 Apr 06:15
Compare
Choose a tag to compare
Check bigdft 2.8-2 Pre-release
Pre-release
ntpoly_2.8_bigdft_2

Fix cmake

Version 2.7.1

16 Feb 02:22
1bc7159
Compare
Choose a tag to compare

Minor update to version 2.7.1. To be synchronized with ELSI.

Features:

  • Utilities for generating ELSI releases
  • Export pkg-conf
  • Fill dense matrix of 1s

Bug Fixes:

  • MPI datatypes fixes

check bigdft 2.8

07 Dec 06:54
Compare
Choose a tag to compare
check bigdft 2.8 Pre-release
Pre-release

This is a test release to help understand the role of finite temperature in BigDFT calculations.