fix(lag computation): fixing lag computation by switching from delta states - raw actions correlation to delta states - delta actions correlation

src/app/[org]/[dataset]/[episode]/fetch-data.ts

@@ -2082,14 +2082,12 @@ export async function loadCrossEpisodeActionVariance(
 
       for (let pi = 0; pi < pairs.length; pi++) {
         const [ai, si] = pairs[pi];
-        const aVals = actions.slice(0, n).map((r) => r[ai] ?? 0);
-        const sDeltas = Array.from(
-          { length: n - 1 },
-          (_, t) => (states[t + 1][si] ?? 0) - (states[t][si] ?? 0),
-        );
-        const effN = Math.min(aVals.length, sDeltas.length);
-        const aM = aVals.slice(0, effN).reduce((a, b) => a + b, 0) / effN;
-        const sM = sDeltas.slice(0, effN).reduce((a, b) => a + b, 0) / effN;
+        const aDeltas = Array.from({ length: n - 1 }, (_, t) => (actions[t + 1][ai] ?? 0) - (actions[t][ai] ?? 0));
+        const sDeltas = Array.from({ length: n - 1 }, (_, t) => (states[t + 1][si] ?? 0) - (states[t][si] ?? 0));
+        const effN = aDeltas.length;
+        if (effN < 4) continue;
+        const aM = aDeltas.reduce((a, b) => a + b, 0) / effN;
+        const sM = sDeltas.reduce((a, b) => a + b, 0) / effN;
 
         for (let li = 0; li < numLags; li++) {
           const lag = -maxLag + li;
@@ -2098,8 +2096,8 @@ export async function loadCrossEpisodeActionVariance(
             sV = 0;
           for (let t = 0; t < effN; t++) {
             const sIdx = t + lag;
-            if (sIdx < 0 || sIdx >= sDeltas.length) continue;
-            const a = aVals[t] - aM,
+            if (sIdx < 0 || sIdx >= effN) continue;
+            const a = aDeltas[t] - aM,
               s = sDeltas[sIdx] - sM;
             sum += a * s;
             aV += a * a;

src/components/action-insights-panel.tsx

@@ -1281,15 +1281,14 @@ function StateActionAlignmentSection({
     }
     if (pairs.length === 0) return null;
 
-    // Per-pair cross-correlation
+    // Per-pair cross-correlation (Δaction vs Δstate)
     const pairCorrs: number[][] = [];
     for (const [aKey, sKey] of pairs) {
-      const aVals = data.map((row) => row[aKey] ?? 0);
-      const sDeltas = data
-        .slice(1)
-        .map((row, i) => (row[sKey] ?? 0) - (data[i][sKey] ?? 0));
-      const n = Math.min(aVals.length, sDeltas.length);
-      const aM = aVals.slice(0, n).reduce((a, b) => a + b, 0) / n;
+      const aDeltas = data.slice(1).map((row, i) => (row[aKey] ?? 0) - (data[i][aKey] ?? 0));
+      const sDeltas = data.slice(1).map((row, i) => (row[sKey] ?? 0) - (data[i][sKey] ?? 0));
+      const n = Math.min(aDeltas.length, sDeltas.length);
+      if (n < 4) { pairCorrs.push(Array(2 * maxLag + 1).fill(0)); continue; }
+      const aM = aDeltas.slice(0, n).reduce((a, b) => a + b, 0) / n;
       const sM = sDeltas.slice(0, n).reduce((a, b) => a + b, 0) / n;
 
       const corrs: number[] = [];
@@ -1299,8 +1298,8 @@ function StateActionAlignmentSection({
           sV = 0;
         for (let t = 0; t < n; t++) {
           const sIdx = t + lag;
-          if (sIdx < 0 || sIdx >= sDeltas.length) continue;
-          const a = aVals[t] - aM,
+          if (sIdx < 0 || sIdx >= n) continue;
+          const a = aDeltas[t] - aM,
             s = sDeltas[sIdx] - sM;
           sum += a * s;
           aV += a * a;
@@ -1407,7 +1406,7 @@ function StateActionAlignmentSection({
           </h3>
           <InfoToggle>
             <p className="text-xs text-slate-400">
-              Per-dimension cross-correlation between action<sub>d</sub>(t) and
+              Per-dimension cross-correlation between Δaction<sub>d</sub>(t) and
               Δstate<sub>d</sub>(t+lag), aggregated as
               <span className="text-orange-400"> max</span>,{" "}
               <span className="text-slate-200">mean</span>, and