pyfracval.experimental#

Archived CCA sticking experiments that did not outperform the vanilla Fibonacci baseline in head-to-head benchmarks.

See docs/source/experiments.md for the data and reasoning. Kept here (rather than deleted) in case a different angle on the same idea proves useful later. Nothing in pyfracval imports this package by default except through the existing opt-in config flags (CCA_STICKING_METHOD, CCA_SOFT_RELAXATION_ENABLED), both of which default to off.