paper: integrate calibrated oral-B-v2 evidence

2 files changed, 49 insertions, 3 deletions

diff --git a/experiments/audit_manuscript.py b/experiments/audit_manuscript.py
index 5743f24..555a50e 100644
--- a/experiments/audit_manuscript.py
+++ b/experiments/audit_manuscript.py
@@ -15,7 +15,7 @@ EXPECTED_SECTIONS = [
     "## 3. What residualization guarantees—and what it does not",
     "## 4. Experimental protocol",
     "## 5. Results",
-    "## 6. Biological-signature test and negative evidence",
+    "## 6. Biological-signature test and outcome-surprise evidence",
     "## 7. Related work",
     "## 8. Limitations and discussion",
     "## 9. Reproducibility statement",
diff --git a/experiments/finalize_accept.sh b/experiments/finalize_accept.sh
index b39a714..db3f9bd 100755
--- a/experiments/finalize_accept.sh
+++ b/experiments/finalize_accept.sh
@@ -16,14 +16,28 @@ experiments/finalize_claims.sh
 /home/yurenh2/miniconda3/envs/ep_pascal/bin/python3 \
   experiments/bci_td_protocol_smoke.py
 /home/yurenh2/miniconda3/envs/ep_pascal/bin/python3 \
+  experiments/bci_v2_smoke.py
+/home/yurenh2/miniconda3/envs/ep_pascal/bin/python3 \
+  experiments/bci_v2_recovery_smoke.py
+/home/yurenh2/miniconda3/envs/ep_pascal/bin/python3 \
+  experiments/bci_v2_calibrated_smoke.py
+/home/yurenh2/miniconda3/envs/ep_pascal/bin/python3 \
   experiments/oral_a_dynamic_scaling_smoke.py
 /home/yurenh2/miniconda3/envs/ep_pascal/bin/python3 -m py_compile \
   experiments/bci_td_run.py experiments/analyze_bci_td_development.py \
   experiments/bci_td_confirmation.py \
   experiments/analyze_bci_td_confirmation.py \
+  experiments/bci_v2_run.py experiments/analyze_bci_v2_development.py \
+  experiments/bci_v2_recovery_run.py \
+  experiments/analyze_bci_v2_recovery_development.py \
+  experiments/bci_v2_calibrated_run.py \
+  experiments/analyze_bci_v2_calibrated_development.py \
+  experiments/bci_v2_calibrated_confirmation.py \
+  experiments/analyze_bci_v2_calibrated_confirmation.py \
   experiments/oral_a_dynamic_scaling.py \
   experiments/analyze_oral_a_dynamic_scaling.py \
   experiments/plot_resnet_confirmation.py \
+  experiments/plot_bci_v2_confirmation.py \
   experiments/audit_manuscript.py
 /home/yurenh2/miniconda3/envs/ep_pascal/bin/python3 \
   experiments/analyze_kp_dynamic_projection.py >/dev/null
@@ -58,6 +72,38 @@ if [ -f results/bci_td_confirmation_gate.json ]; then
   /home/yurenh2/miniconda3/envs/ep_pascal/bin/python3 \
     experiments/analyze_bci_td_confirmation.py >/dev/null
 fi
+if [ -f results/bci_v2_dev_gate.json ]; then
+  PYTHONPATH=. /home/yurenh2/miniconda3/envs/ep_pascal/bin/python3 \
+    experiments/analyze_bci_v2_development.py >/dev/null
+fi
+if [ -f results/bci_v2_recovery_dev_gate.json ]; then
+  PYTHONPATH=. /home/yurenh2/miniconda3/envs/ep_pascal/bin/python3 \
+    experiments/analyze_bci_v2_recovery_development.py >/dev/null
+fi
+if [ -f results/bci_v2_calibrated_dev_gate.json ]; then
+  PYTHONPATH=. /home/yurenh2/miniconda3/envs/ep_pascal/bin/python3 \
+    experiments/analyze_bci_v2_calibrated_development.py >/dev/null
+fi
+if [ -f results/bci_v2_calibrated_confirmation_gate.json ]; then
+  PYTHONPATH=. /home/yurenh2/miniconda3/envs/ep_pascal/bin/python3 \
+    experiments/analyze_bci_v2_calibrated_confirmation.py >/dev/null
+  /home/yurenh2/miniconda3/bin/python \
+    experiments/plot_bci_v2_confirmation.py >/dev/null
+  jq -e '
+    .strict == true and
+    .gate_status == "passed" and
+    .task_seeds == [30, 31, 32, 33, 34, 35] and
+    .model_seeds == [0, 1, 2, 3, 4] and
+    .record_count == 30 and
+    .statistics.intact_final_mean >= 0.99 and
+    .statistics.residual_soma_corr_mean <= 0.10 and
+    .statistics.surrounding_accuracy_lower >= 0.50 and
+    .statistics.terminal_accuracy_lower >= 0.75 and
+    .statistics.outcome_lesion_drop_lower >= 0.15 and
+    .statistics.critic_expectedness_lower >= 0.02 and
+    .statistics.positive_sign_count == 30
+  ' results/figs/figure5_bci_v2_manifest.json >/dev/null
+fi
 if [ -f results/oral_a_dynamic_scaling_gate.json ]; then
   /home/yurenh2/miniconda3/envs/ep_pascal/bin/python3 \
     experiments/analyze_oral_a_dynamic_scaling.py >/dev/null
@@ -75,8 +121,8 @@ fi
 jq -e '
   .strict == true and
   .status == "passed" and
-  (.audited_numeric_claims | length) == 34 and
-  (.gates | map(select(.status == "passed")) | length) == 1 and
+  (.audited_numeric_claims | length) == 51 and
+  (.gates | map(select(.status == "passed")) | length) == 2 and
   (.gates | map(select(.status == "failed")) | length) == 1 and
   (.gates[] | select(.status == "failed") | .false_checks | length) == 7
 ' paper/manuscript_audit.json >/dev/null