ChatGPT suggested a couple of more improvements to the Kawasaki Metropolis sampling for transverse field Ising model sample generation: add a "thinning" multiple proportional to qubit count to reduce sample-to-sample correlation, and construct the zeros and ones indices lists once as sets, instead, then add and remove from them.
Full Changelog: v9.11.1...v9.11.2
sha1sum results:
96a34a72f30f7b7b198c853157442d972888a1bb pyqrackising-9.11.2-py3-none-macosx_14_0_arm64.whl
76b6d3968e918908c20c541334a5fb0462d56e6c pyqrackising-9.11.2-py3-none-macosx_15_0_arm64.whl
ee4badb9030ca6ffe4ba50eaac8f5687dab74f78 pyqrackising-9.11.2-py3-none-manylinux_2_35_x86_64.whl
d487702ee77e0877713e5ec86435ecb65dad11ea pyqrackising-9.11.2-py3-none-manylinux_2_39_x86_64.whl
5250257223c4272f3629350ded86e8a8618d1141 pyqrackising-9.11.2-py3-none-win_amd64.whl
ca4e46c3ae207445692935f2edadbccc77f44f8f pyqrackising-9.11.2.tar.gz