DspPipeline

Phase 2: symbol pipeline conventions, reproducibility and estimators

6/6 passed in 3519ms

Test	What it does	Result
test_chunk_invariance	A 200k run reproduces a 100k run's frames bit for bit but for the last, neighbourless block, symbol $i$ being a pure function of (seed, stage, $i$).	✅ pass
test_frames_parity	v_err, t_hat and sigma2_hat are bit-identical with the recovered frames kept or dropped.	✅ pass
test_pilot_snr	Pilot SNR against the adjacent empty DFT bin is positive and tracks pilot power one for one in dB.	✅ pass
test_residual_conditioning	Moments taken about the nominal slope, not the origin, keep sigma2_hat converged up to $V_A={10}^{12}$ while v_err falls as $1/V_A$.	✅ pass
test_thread_invariance	Fixed index-order folding makes one and eight workers agree bit for bit, frames and scalars, at a 4 MHz offset.	✅ pass
test_vacuum_normalization	With no modulation, unit transmittance and no excess or phase noise, Bob's per-quadrature variance is $1+v_{el}$ at every $\eta$.	✅ pass