QStabilizerHybrid` has been fundamentally improved, particularly for the gate set "Clifford+RZ" (or "Clifford+T"). The entire unitary portion of circuit simulation, before measurement, now has a polynomial-complexity simulation algorithm, in space and time requirements. If measuring across the full width of the simulator, or sampling, measurement (alone) scales exponentially in space requirements proportional to (less than or up to) the number of non-Clifford `RZ` (or `T`) gates, and exponentially in time requirements proportional to base logical qubit count in the simulator instance.
No special considerations are necessary to engage this simulation mode: simply restrict your gate set to Clifford+RZ, when using any simulator that properly includes the `QStabilizerHybrid` layer, such as the default optimal simulator stack.
(UPDATE: Actually, it turns out that the default optimal stack, with `QUnit`, will apparently work for Clifford+T, but not Clifford+RZ. For Clifford+RZ, use `QStabilizerHybrid` _without_ `QUnit`.)
File SHA-1 sums:
872e50ec0ab5ec55fed3ec792371fe2ff7e5797e pyqrack-1.14.0-py3-none-any.whl
22efe1b97f5a33b7fda0eefb5ef4d612fef6c331 pyqrack-1.14.0-py3-none-linux_aarch64.whl
c660e8880c1c1d152e53670658c515ba64020f77 pyqrack-1.14.0-py3-none-linux_armv7l.whl
abf8b6d5cb70ffada16a9217e449312722e086ca pyqrack-1.14.0-py3-none-macosx_10_4_x86_64.whl
fd73db64b66b07c3c977eaba0119e8651bec2d7b pyqrack-1.14.0-py3-none-manylinux2014_x86_64.whl
a57f12343acd669e69b80b1c201346317d29ae8c pyqrack-1.14.0-py3-none-manylinux_2_35_x86_64.whl
f635d177713c60385843380dfa0c9db93b082c73 pyqrack-1.14.0-py3-none-win32.whl
7d0b1dd20b529bc844eabd6873c26cb3cd8f54ff pyqrack-1.14.0-py3-none-win_amd64.whl
9e7d24b63767102fc6e5a680f7fd85cb11fb9a91 pyqrack-1.14.0.tar.gz