Sweep complementary magnetic tangent weighting

Author: mgrouch

tools/ou_iii_mag_tangent_sweep.py

@@ -149,17 +149,19 @@ def main() -> None:
         base = run(1.0, 1.0)
         print("BASELINE", flush=True)
 
-        # Start with yaw-only down-weighting, because isotropic mag inflation
-        # improved 44/56 but paid its main cost in horizontal X. Then add only
-        # mild tilt changes around the most plausible yaw range, plus two
-        # stronger-tilt-information probes (<1) to test the opposite direction.
-        candidates = []
-        for sy in (1.05, 1.10, 1.15, 1.20, 1.30, 1.45):
-            candidates.append((sy, 1.00))
-        for sy in (1.10, 1.15, 1.20):
-            for st in (1.05, 1.10):
-                candidates.append((sy, st))
-        candidates.extend(((1.15, 1.15), (1.15, 0.95), (1.15, 0.90)))
+        # First sweep showed that weakening the yaw-sensitive tangent direction
+        # moves the stubborn horizontal-X metrics in the wrong direction, while
+        # weakening the orthogonal (tilt-sensitive) tangent direction produces
+        # most of the broad gains.  Search the complementary quadrant: slightly
+        # *more* yaw information with slightly *less* tilt information.
+        candidates = [
+            (0.88, 1.10),
+            (0.92, 1.06), (0.92, 1.10), (0.92, 1.14),
+            (0.96, 1.06), (0.96, 1.10), (0.96, 1.14),
+            (1.00, 1.06), (1.00, 1.10), (1.00, 1.14),
+            (1.04, 1.06), (1.04, 1.10), (1.04, 1.14),
+            (0.96, 1.18), (0.96, 1.22),
+        ]
 
         results = []
         for sy, st in candidates: