Streaming Polyphase Factor Decomposition / SKAT

Download original file

This is the original source text. Reading it does not run or verify the program.

Read the typeset paper

1 from pathlib import Path
2 import pandas as pd
3 
4 HERE = Path(__file__).resolve().parent
5 
6 part = pd.read_csv(HERE/'partition_invariance.csv')
7 hier = pd.read_csv(HERE/'hierarchical_refinement.csv')
8 lc = pd.read_csv(HERE/'polyphase_psd_light_cone.csv')
9 rank = pd.read_csv(HERE/'phase_local_rank_audit.csv')
10 qr = pd.read_csv(HERE/'qr_core_update_audit.csv')
11 mem = pd.read_csv(HERE/'memory_asymptotic_sweep.csv')
12 
13 assert part['gram_rel_error'].max() < 1e-12
14 assert part['action_rel_error'].max() < 1e-12
15 assert hier['coarse_vs_fine_rel_error'].iloc[0] < 1e-12
16 assert lc.loc[lc['outside_light_cone'], 'norm'].max() == 0.0
17 assert (rank['constructed_rank'] == rank['svd_numerical_rank']).all()
18 assert qr['relative_l2_singular_difference'].iloc[0] < 1e-12
19 assert mem['stream_kernel_action_MiB'].max() == mem['stream_kernel_action_MiB'].min()
20 assert mem['full_factor_MiB'].iloc[-1] > 90 * mem['full_factor_MiB'].iloc[0]
21 print('PASS: v0.2 formal regression invariants')

Back to source controls

Back to Streaming Polyphase Factor Decomposition / SKAT