Name: Math Core
Author: gretjia

Skill: math_core

When To Use

Modifying omega_core/omega_math_core.py or omega_core/omega_math_vectorized.py
Adding/changing physics formulas, compression gain, topology calculations
Changing config.py parameters that affect mathematical output

Delta = 0.5 is universal (Sato 2025). The SRL exponent is hardcoded to 0.5. Do NOT make it configurable or "race" across exponents. The race was cancelled in v5.0.
Epiplexity gain ≥ 0 (MDL discipline). calc_compression_gain() must never return negative values. If model cost exceeds data cost, return 0.
Zero-variance → zero signal. Constant traces must produce 0.0 gain, not NaN/inf.
Topology sign reversal. calc_topology_area(x, y) reversed == -calc_topology_area(x_rev, y_rev). Green's theorem is antisymmetric.

Modifying omega_core/omega_math_core.py or omega_core/omega_math_vectorized.py
Adding/changing physics formulas, compression gain, topology calculations
Changing config.py parameters that affect mathematical output

Delta = 0.5 is universal (Sato 2025). The SRL exponent is hardcoded to 0.5. Do NOT make it configurable or "race" across exponents. The race was cancelled in v5.0.
Epiplexity gain ≥ 0 (MDL discipline). calc_compression_gain() must never return negative values. If model cost exceeds data cost, return 0.
Zero-variance → zero signal. Constant traces must produce 0.0 gain, not NaN/inf.
Topology sign reversal. calc_topology_area(x, y) reversed == -calc_topology_area(x_rev, y_rev). Green's theorem is antisymmetric.

Floor	Config Path	Purpose
`sigma_floor`	`L2SRLConfig.sigma_floor`	Prevents sqrt(Q/D) / 0
`depth_floor`	`L2SRLConfig.depth_floor`	Prevents division by zero depth
`price_scale_floor`	`L2TopologyRaceConfig`	Prevents topology normalization explosion
`min_trace_len`	`L2EpiplexityConfig`	Prevents OLS on < 3 data points