SirStone/SirRoboGarage
1
0
Fork 0
Code Issues 28 Pull Requests Actions Projects Releases Wiki Activity
Files
a01141c959b3c5b17c47a6957e45d96a43a6f684
SirRoboGarage/common_libs/tests
T
History
SirStone a01141c959 j151 (default-off): hard arrival bound TR_TFIL_ARRIVE_TICKS + the offline per-pick ruler that found it 
Measures the owner's two complaints per pick (turn angle, heat on the path
vs at the destination, time-to-arrive, and whether the destination is hot on
the recorded true future when we would arrive), split by empty vs non-empty
safe set. Result: the path IS scored (hard filter + least-hot fallback), but
there was NO time term at all -- 65% of picks outran the 15-tick commitment
and the chosen tile was reached 6.5% of the time. The bound refuses a
candidate we cannot reach in TR_TFIL_ARRIVE_TICKS (default 0 = off = today).

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-09-27 09:57:25 +02:00
..
fixtures
j134 fire fix: share ONE fire_tracker across tfil/ring/strafe/learned/surf (TR_FIRE_FIX, default on); env-gated TR_FIRE_DIAG alignment trace
2026-09-26 12:40:19 +02:00
probe_bots/RulesProbe
harness: upgrade to server 1.3.1, keep 0.35.5 selectable, re-baseline
2026-09-21 22:41:06 +02:00
acceptance_offline_vs_online.nim
Gate virtual-bullet spawn on rack admission: +68% tick rate, selected gun unchanged
2026-09-22 02:24:18 +02:00
analyze_drussgt_dodge_vs_power.py
DrussGT dodge vs fired power: no movement response once range is controlled
2026-09-24 21:22:38 +02:00
analyze_lead_capture_by_range.py
lead capture by range: we under-lead (0.40->0.135) — a RANGE effect, not a power one
2026-09-24 22:47:18 +02:00
analyze_melee_ab.py
MELEE A/B: BitBrain vs shipped Pattern rack — repo's first melee measurement (null)
2026-09-26 00:35:44 +02:00
analyze_selector.nim
feat(gun_harness): scale-aware selector thresholds; default = path + relative
2026-09-21 04:33:52 +02:00
analyze_tmcomposites.py
TMComposites gate: per-gun confidence faithful for 3 guns; no pair composes
2026-09-25 22:04:39 +02:00
audit_virtual_guns.nim
gun audit: the virtual system is sound; the rack is redundant, not broken
2026-09-22 00:44:19 +02:00
audit_wave_pairing.nim
wave pairing: 36-58% of GF/DecayGF/KNN learning samples were MISLABELLED
2026-09-22 01:33:31 +02:00
compare_pairing.nim
wave pairing: 36-58% of GF/DecayGF/KNN learning samples were MISLABELLED
2026-09-22 01:33:31 +02:00
config.nims
fix(guns): speed-sensitive caches, dead stop-shot branch, exact TM trace pairing
2026-09-20 22:47:26 +02:00
diag_automata_validation.nim
Automata metrics: settledness alone does NOT separate learning from fidgeting
2026-09-22 21:40:27 +02:00
diag_synthetic.nim
TM diagnostics kit: VALIDATED (finds a known dead input), and it found a real bug
2026-09-22 21:24:27 +02:00
diag_tm_pattern_offline.nim
s settled from OUR code (higher = LONGER clauses), and it cannot rescue the gun
2026-09-22 21:56:06 +02:00
exact_geometry_gate.py
j131 learned movement: real bullet-endpoint resolution (TR_LEARNED_REAL_EVENTS, default off) + exact-geometry Gate A/B (inversion NOT fixed; state still the constraint)
2026-09-26 11:54:11 +02:00
gate2_tm_miss_shrink_results.txt
GATE 2: STOP - the TM learns the side but the correction cannot buy hits
2026-09-22 22:06:01 +02:00
gate2b_hit_optimal_results.txt
Gate 2b: the shift form WORKS - but break-even needs ~80% side accuracy and the
2026-09-22 22:21:32 +02:00
gen_synthetic_fixtures.nim
feat(gun_harness): offline gun range, proven equivalent to live play
2026-09-20 23:44:42 +02:00
label_inversion_three_way.py
j133 strafe fire detection: correct the enemy energy delta for the server's +3*power hit bonus and our own damage, never drop a too-large drop (TR_STRAFE_FIRE_FIX, default on); catch 98.89%->100% of enemy fires on the 70-battle corpus
2026-09-26 12:18:21 +02:00
learned_surfer_gate.py
learned movement: danger-map alignment diagnostic (corr(P(arrival bin), P(hit|bin)) = -0.342) in the gate + ledger
2026-09-26 10:45:26 +02:00
measure_bitbrain_gate_results.txt
BitBrain gate test: fine-grained aim correction vs naive + Pattern (offline)
2026-09-24 21:23:29 +02:00
measure_bitbrain_gate.nim
BitBrain gate test: fine-grained aim correction vs naive + Pattern (offline)
2026-09-24 21:23:29 +02:00
measure_bridge_battle.nim
harness: upgrade to server 1.3.1, keep 0.35.5 selectable, re-baseline
2026-09-21 22:41:06 +02:00
measure_cornering_guns.nim
cornering/ramming premise REFUTED on three independent measurements
2026-09-21 22:49:56 +02:00
measure_cornering_ram.py
cornering/ramming premise REFUTED on three independent measurements
2026-09-21 22:49:56 +02:00
measure_corpse_melee.nim
tracker: corpses do not exist - revert the fix and retire the workaround
2026-09-21 22:41:11 +02:00
measure_counted_sbc.nim
BitBrain SBC: counted mode + global decay (forgetting, probabilities)
2026-09-25 08:39:10 +02:00
measure_fire_ghost_lag.py
j147: the 1-tick fire-detection lag is OURS — measure it, then back-date it (TR_FIRE_LAG)
2026-09-26 23:08:57 +02:00
measure_horizon_headroom.py
Horizon headroom: QUALIFIED PASS at h=10..50, but the signal is one feature
2026-09-22 21:46:05 +02:00
measure_hysteresis.nim
gun selector: hysteresis measured NEGATIVE, shipped at the lightest setting
2026-09-21 22:41:01 +02:00
measure_melee_bitbrain_ab.nim
MELEE A/B: BitBrain vs shipped Pattern rack — repo's first melee measurement (null)
2026-09-26 00:35:44 +02:00
measure_melee_strong_field.nim
j124 fair melee: add strong-field melee harness + driver (legacy champion field)
2026-09-26 04:08:44 +02:00
measure_parallel_vbullet_cost.nim
cost: parallel per-enemy virtual bullets are NOT affordable as proposed
2026-09-22 01:26:12 +02:00
measure_pattern_radial.nim
Radial offset: STRUCTURALLY incapable of helping, and Pattern does not overshoot
2026-09-22 02:23:41 +02:00
measure_physics.nim
harness: upgrade to server 1.3.1, keep 0.35.5 selectable, re-baseline
2026-09-21 22:41:06 +02:00
measure_power_policy.nim
Energy economy: the cliff becomes a SLOPE, plus a finishing cap. 11% less energy.
2026-09-23 00:12:04 +02:00
measure_strafe_fire_catch.py
j134 fire fix: share ONE fire_tracker across tfil/ring/strafe/learned/surf (TR_FIRE_FIX, default on); env-gated TR_FIRE_DIAG alignment trace
2026-09-26 12:40:19 +02:00
measure_strafe_gates.nim
STRAFE: draw the heat field, real bullet danger, corrected ring comment
2026-09-25 23:25:34 +02:00
measure_strafe_heat_display.nim
j132 strafe display check: also assert the chosen tile's ramp border and value plate reach the SVG
2026-09-26 11:54:14 +02:00
measure_strafe_range_stall.nim
STRAFE: range control (tilt) + corner-stall escape; shipped TFIL default untouched
2026-09-25 23:39:28 +02:00
measure_strafe_wings.nim
STRAFE: curved wings + guaranteed wall/corner escape; shipped TFIL default untouched
2026-09-25 23:51:12 +02:00
measure_tfil_arrival.nim
j146 ledger: the field shape restores a real safe set offline (filter broken 63.5%->30.4%) and is a live null on damage/run and round wins; 375 battles, 5 arms
2026-09-26 22:35:17 +02:00
measure_tfil_commit_ab.nim
TFIL: measure the commitment A/B on 490 live DrussGT rounds - no effect
2026-09-24 20:24:34 +02:00
measure_tfil_heat_field.nim
TFIL heat field: the safety model keeps us ~400px away, which is our worst range
2026-09-21 23:39:14 +02:00
measure_tfil_heat_time.nim
TFIL: time-indexed bullet heat (TR_TFIL_HEAT_TIME, DEFAULT OFF)
2026-09-25 21:47:54 +02:00
measure_tfil_pick_defects.nim
j151 (default-off): hard arrival bound TR_TFIL_ARRIVE_TICKS + the offline per-pick ruler that found it
2026-09-27 09:57:25 +02:00
measure_tfil_picker_loss.nim
j150 (diag only): picker loss histogram + sweepable heat cutoff
2026-09-27 09:51:19 +02:00
measure_tm_hit_optimal.nim
Gate 2b: the shift form WORKS - but break-even needs ~80% side accuracy and the
2026-09-22 22:21:32 +02:00
measure_tm_miss_shrink.nim
GATE 2: STOP - the TM learns the side but the correction cannot buy hits
2026-09-22 22:06:01 +02:00
measure_tm_pattern_cost.nim
TM radial gun: registered (default OFF) + label-bias fix that removes the bias but
2026-09-22 01:58:33 +02:00
measure_tm_readapt_results.txt
Re-adaptation: the sliding WINDOW is a control-validated +9.3pp on late accuracy
2026-09-23 00:29:37 +02:00
measure_tm_readapt.nim
Re-adaptation: the sliding WINDOW is a control-validated +9.3pp on late accuracy
2026-09-23 00:29:37 +02:00
measure_tmcomposites.nim
TMComposites gate: per-gun confidence faithful for 3 guns; no pair composes
2026-09-25 22:04:39 +02:00
measure_vbullet_admit_gate.nim
Gate virtual-bullet spawn on rack admission: +68% tick rate, selected gun unchanged
2026-09-22 02:24:18 +02:00
outcome_label_gate.py
j130 learned movement: outcome label (P(hit|state,g)) mode + Gate A; pre-registered outcome arms
2026-09-26 11:27:03 +02:00
pattern_radial_results.md
Radial offset: STRUCTURALLY incapable of helping, and Pattern does not overshoot
2026-09-22 02:23:41 +02:00
prediction_quality_results.txt
BitBrain campaign phase 1: the missing gain sweep and BitBrain as a lead-gain corrector
2026-09-25 00:10:19 +02:00
range_guns.nim
Offline harness: audited, calibrated against live, and one real bug fixed
2026-09-24 21:40:38 +02:00
run_melee_bitbrain_ab.sh
MELEE A/B: BitBrain vs shipped Pattern rack — repo's first melee measurement (null)
2026-09-26 00:35:44 +02:00
run_melee_strong_field.sh
j124 fair melee: add strong-field melee harness + driver (legacy champion field)
2026-09-26 04:08:44 +02:00
run_prediction_quality.nim
j142 retire the ADE+SBC gun (rack id 17): the owner watched it, it does not learn, throw it away
2026-09-26 19:20:23 +02:00
run_range.nim
feat(gun_harness): offline gun range, proven equivalent to live play
2026-09-20 23:44:42 +02:00
state_window_gate.py
State-window gate: a temporal window of wave-relative states does NOT beat a single state
2026-09-25 08:43:36 +02:00
sweep_pattern_radial.nim
Radial offset: STRUCTURALLY incapable of helping, and Pattern does not overshoot
2026-09-22 02:23:41 +02:00
sweep_radial_offset.nim
Ablation: the radial TM is replaceable by a CONSTANT, and its avenue is dead on the
2026-09-22 02:08:38 +02:00
sweep_tm_pattern.nim
TM verdict, settled: it loses LIVE and sits at/below its majority class - (c)
2026-09-22 08:21:49 +02:00
sweep_tsetlin.nim
Tsetlin gun: NO configuration adapts faster than random feedback
2026-09-21 22:45:28 +02:00
test_adaptive_radar.nim
tracker: corpses do not exist - revert the fix and retire the workaround
2026-09-21 22:41:11 +02:00
test_bitbrain_mnist.nim
BitBrain: generic clean-room ADE + SBC library with MNIST acceptance
2026-09-24 20:52:45 +02:00
test_bitbrain.nim
BitBrain SBC: counted mode + global decay (forgetting, probabilities)
2026-09-25 08:39:10 +02:00
test_enemy_tracker_death.nim
tracker: corpses do not exist - revert the fix and retire the workaround
2026-09-21 22:41:11 +02:00
test_env_dotenv.nim
j139 dotenv: strip trailing inline comments and warn on non-token values
2026-09-26 15:18:04 +02:00
test_env_report.nim
j142 retire the ADE+SBC gun (rack id 17): the owner watched it, it does not learn, throw it away
2026-09-26 19:20:23 +02:00
test_event_drop_mechanism.nim
tracker: corpses do not exist - revert the fix and retire the workaround
2026-09-21 22:41:11 +02:00
test_gun_harness.nim
gun selector: hysteresis measured NEGATIVE, shipped at the lightest setting
2026-09-21 22:41:01 +02:00
test_lead_gain_legacy.nim
j142 retire the ADE+SBC gun (rack id 17): the owner watched it, it does not learn, throw it away
2026-09-26 19:20:23 +02:00
test_lead_gain_registration.nim
j142 retire the ADE+SBC gun (rack id 17): the owner watched it, it does not learn, throw it away
2026-09-26 19:20:23 +02:00
test_pattern_radial_offset.nim
Radial offset: STRUCTURALLY incapable of helping, and Pattern does not overshoot
2026-09-22 02:23:41 +02:00
test_power_policy.nim
Energy economy: the cliff becomes a SLOPE, plus a finishing cap. 11% less energy.
2026-09-23 00:12:04 +02:00
test_power_selection.nim
feat(guns): scale-aware power selection (+52% damage); TM classifier gun built, measured, DISABLED
2026-09-21 05:19:07 +02:00
test_rack_membership.nim
j142 retire the ADE+SBC gun (rack id 17): the owner watched it, it does not learn, throw it away
2026-09-26 19:20:23 +02:00
test_rack_stat_width.nim
j142 retire the ADE+SBC gun (rack id 17): the owner watched it, it does not learn, throw it away
2026-09-26 19:20:23 +02:00
test_ram_decision.nim
Ram: finisher-only by default, and the bullet-rain abort now measures real energy
2026-09-22 08:14:03 +02:00
test_range_rack_parity.nim
Offline harness: audited, calibrated against live, and one real bug fixed
2026-09-24 21:40:38 +02:00
test_selector_tiebreak.nim
selector: arrival-accuracy tie-break measured NEGATIVE; randomness is load-bearing
2026-09-22 01:07:12 +02:00
test_strafe_fire_fix.nim
j134 fix: tracker must take the mover's switch per-call (instance flag was silently off on the object-literal construction path) + apply the correction on the reading that carries the server's turn-N+1 energy change (two-slot buffer); live-verified alignment
2026-09-26 12:56:26 +02:00
test_tfil_commit_env.nim
j151 (default-off): hard arrival bound TR_TFIL_ARRIVE_TICKS + the offline per-pick ruler that found it
2026-09-27 09:57:25 +02:00
test_tfil_ring_weights.nim
movement: range-weighted TFIL variant + tamed heat field (opt-in, default unchanged)
2026-09-21 23:46:47 +02:00
test_tm_automata_diag.nim
Automata metrics: settledness alone does NOT separate learning from fidgeting
2026-09-22 21:40:27 +02:00
test_tm_clause_shape.nim
s settled from OUR code (higher = LONGER clauses), and it cannot rescue the gun
2026-09-22 21:56:06 +02:00
test_tm_diag.nim
TM diagnostics kit: VALIDATED (finds a known dead input), and it found a real bug
2026-09-22 21:24:27 +02:00
test_tm_horizon.nim
Re-adaptation: the sliding WINDOW is a control-validated +9.3pp on late accuracy
2026-09-23 00:29:37 +02:00
test_tm_pattern_learning.nim
test(range): restore the 12/12 offline==online proof; measure TM clause readability
2026-09-21 00:14:46 +02:00
test_tm_pattern_rack_live.nim
TM radial gun: registered (default OFF) + label-bias fix that removes the bias but
2026-09-22 01:58:33 +02:00
test_tm_pattern_registration.nim
j142 retire the ADE+SBC gun (rack id 17): the owner watched it, it does not learn, throw it away
2026-09-26 19:20:23 +02:00
test_tm_selector.nim
feat(guns): scale-aware power selection (+52% damage); TM classifier gun built, measured, DISABLED
2026-09-21 05:19:07 +02:00
test_tsetlin_gun.nim
feat(gun_harness): scale-aware selector thresholds; default = path + relative
2026-09-21 04:33:52 +02:00
test_tsetlin_live.nim
fix(tsetlin): make the TM actually learn - saturation 714 -> 13.8 literals/clause
2026-09-20 23:59:32 +02:00
test_vbullet_admit_gate.nim
Gate virtual-bullet spawn on rack admission: +68% tick rate, selected gun unchanged
2026-09-22 02:24:18 +02:00
test_vbullet_metric.nim
feat(gun_harness): scale-aware selector thresholds; default = path + relative
2026-09-21 04:33:52 +02:00
test_wave_pairing.nim
wave pairing: 36-58% of GF/DecayGF/KNN learning samples were MISLABELLED
2026-09-22 01:33:31 +02:00
test_wavesurfer_velocity.nim
fix(adversaries): repair four of the five sparring bots
2026-09-21 08:21:47 +02:00
tm_pattern_sweep_results.md
TM verdict, settled: it loses LIVE and sits at/below its majority class - (c)
2026-09-22 08:21:49 +02:00