SabaPivot/repro-on-the-computational-complexity-of-performative-prediction

Reproduction: On the Computational Complexity of Performative Prediction

An open experiment logbook, published with Trackio.

The canonical Space includes a rigorous v2 claim-native audit:

python upgrade_performative_v2.py

It runs the affine and quadratic reductions at larger dimensions, a working central-cut ellipsoid solver, seven convex domains, the exact ERM-query bound, and the complete strategic classifier/neighbor domain. Hashed results are in outputs_v2/.