openscvx 0.5.3.dev20__tar.gz → 0.5.3.dev21__tar.gz

{openscvx-0.5.3.dev20/openscvx.egg-info → openscvx-0.5.3.dev21}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: openscvx
-Version: 0.5.3.dev20
+Version: 0.5.3.dev21
 Summary: A general Python-based successive convexification implementation which uses a JAX backend.
 Author-email: Chris Hayner and Griffin Norris <haynec@uw.edu>
 License: Apache Software License

{openscvx-0.5.3.dev20 → openscvx-0.5.3.dev21}/openscvx/_version.py

@@ -18,7 +18,7 @@ version_tuple: tuple[int | str, ...]
 commit_id: str | None
 __commit_id__: str | None
 
-__version__ = version = '0.5.3.dev20'
-__version_tuple__ = version_tuple = (0, 5, 3, 'dev20')
+__version__ = version = '0.5.3.dev21'
+__version_tuple__ = version_tuple = (0, 5, 3, 'dev21')
 
-__commit_id__ = commit_id = 'g6b1cd4074'
+__commit_id__ = commit_id = 'g5960b8809'

{openscvx-0.5.3.dev20 → openscvx-0.5.3.dev21}/openscvx/propagation/post_processing.py

@@ -81,18 +81,23 @@ def propagate_trajectory_results(
     n_opt_states = x.shape[1]
     n_prop_states = settings.sim.x_prop.initial.shape[0]
 
-    # For a converged solution x[0, :] is the actual IC for every optimisation-state
-    # component regardless of initial_type (Fix / Free / Maximize).  Using the
-    # trajectory value directly avoids a subtle bug where "Fix" components would
-    # be seeded from settings.sim.x_prop.initial — the lowering-time default —
-    # rather than the value enforced by a param-fix pin for this particular solve.
+    # Seed from x[0, :] directly so param-fixed "Fix" components use this solve's
+    # pinned value rather than the lowering-time x_prop.initial default.
+    x0_opt = np.array(x[0, :], dtype=float)
+
+    # Seed from the pre-impulse state to avoid impulse duplication
+    if dynamics_discrete is not None and np.any(settings.sim.u._impulsive_mask()):
+        init_fixed = np.asarray(settings.sim.x.initial_type) == "Fix"
+        x_initial = np.asarray(settings.sim.x.initial, dtype=float)
+        x0_opt = np.where(init_fixed, x_initial, x0_opt)
+
     if n_opt_states == n_prop_states:
-        x_prop_for_propagation.initial = np.array(x[0, :], dtype=float)
+        x_prop_for_propagation.initial = x0_opt
     else:
         # Propagation has extra states appended beyond the optimisation states;
         # copy only the overlapping prefix and leave prop-only states untouched.
         x_prop_initial_updated = np.array(settings.sim.x_prop.initial, dtype=float)
-        x_prop_initial_updated[:n_opt_states] = np.array(x[0, :], dtype=float)
+        x_prop_initial_updated[:n_opt_states] = x0_opt
         x_prop_for_propagation.initial = x_prop_initial_updated
 
     # Temporarily replace x_prop with our modified copy for propagation

{openscvx-0.5.3.dev20 → openscvx-0.5.3.dev21/openscvx.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: openscvx
-Version: 0.5.3.dev20
+Version: 0.5.3.dev21
 Summary: A general Python-based successive convexification implementation which uses a JAX backend.
 Author-email: Chris Hayner and Griffin Norris <haynec@uw.edu>
 License: Apache Software License