With apologies for releasing twice in a day after the point we intended for churn to stop, for DOI archival, there is a point of implementation "errata" on the v9.8.0 release: during the "reheat" phase of the Gray code search in spin_glass_solver(G) variants, differential improvements were identified correctly, but the associated absolute cut or energy was calculated incorrectly. This release debugs that single issue. Full Changelog: https://github.com/vm6502q/PyQrackIsing/compare/v9.8.0...v9.8.1 sha1sum results: 2f144312c9c2d333b4e64e417a8aa2c8b3f66525 pyqrackising-9.8.1-py3-none-macosx_14_0_arm64.whl ee29fe2f5ed66c6b8cd88d3600b145aed879d429 pyqrackising-9.8.1-py3-none-macosx_15_0_arm64.whl 3580f563ed8ba62f6d1cc36f1f0fdaf7569883db pyqrackising-9.8.1-py3-none-manylinux_2_35_x86_64.whl b3d57d4df82a9696369a3bc42c431dd748fe8645 pyqrackising-9.8.1-py3-none-manylinux_2_39_x86_64.whl c241d9842a5a1bd04c7cb2b47626a38a5098fba7 pyqrackising-9.8.1-py3-none-win_amd64.whl 58ed8ceea7d9b989e838cb60c0c0374e9f3c2a03 pyqrackising-9.8.1.tar.gz