chore(K2): purge all patent content from DiffCFD

Remove patent claim references (C1/C2/C3) from code comments, examples,
and README. Replace with neutral technical descriptions.

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

Author: telleroutlook

README.md

@@ -415,7 +415,7 @@ src/ (Rust via PyO3/maturin, at repo root)
 
 ## Contributing
 
-This repository is currently in a patent-sensitive phase. Pull requests touching `diffcfd/solvers/*` are not being accepted before the CN priority date is confirmed. Discussion issues and benchmark proposals are welcome.
+This repository is currently in an early-development phase. Pull requests touching `diffcfd/solvers/*` are not being accepted until the API stabilizes. Discussion issues and benchmark proposals are welcome.
 
 ---
 

diffcfd/envs/base.py

@@ -1,8 +1,8 @@
 """Base gymnasium.Env for DiffCFD reinforcement learning environments (v0.3+).
 
-C2 patent claim: step_differentiable() preserves the autograd graph (no .detach()
+step_differentiable() preserves the autograd graph (no .detach()
 on outputs), enabling policy_gradient() to return exact analytical gradients via
-implicit diff (C1).
+implicit differentiation.
 
 Mode A: single-step contextual bandit (geometry/BC optimization).
 Mode B: sequential quasi-steady-state episode (flow control, SB3/PPO compatible).

diffcfd/envs/cylinder_wake.py

@@ -1,4 +1,4 @@
-"""Cylinder wake environment (Mode B) — C2 patent embodiment.
+"""Cylinder wake environment (Mode B).
 
 Quasi-steady-state flow control around a circular cylinder at Re=100.
 Action: cylinder rotation rate (dimensionless, α = ω·D/(2·U∞)).

diffcfd/solvers/implicit_diff.py

@@ -1,6 +1,6 @@
 """Fixed-point implicit differentiation through SIMPLE-converged steady state.
 
-C1 patent claim core: the dual-function architecture.
+Dual-function architecture:
 - Forward: under-relaxed SIMPLE (stable convergence, not used in backward).
 - Backward: matrix-free GMRES on unrelaxed physics residual R(u*, theta) = 0.
 

diffcfd/workflows/topology.py

@@ -1,11 +1,11 @@
 """Topology optimization workflow using differentiable Helmholtz filter + Brinkman.
 
-C3 patent embodiment: coupled geometry and boundary condition optimization with
+Coupled geometry and boundary condition optimization with
 manufacturing constraints (minimum feature size).
 
 The optimization chain is:
-    design variables ρ → Helmholtz filter → smooth Heaviside projection →
-    Brinkman penalization mask → SIMPLE solve → objective → implicit diff backward
+    design variables rho -> Helmholtz filter -> smooth Heaviside projection ->
+    Brinkman penalization mask -> SIMPLE solve -> objective -> implicit diff backward
 
 All within a single PyTorch autograd computational graph.
 

examples/01_implicit_differentiation.py

@@ -1,11 +1,11 @@
 """Example: implicit differentiation through Poiseuille channel flow.
 
-Demonstrates the C1 patent claim:
-    solve SIMPLE to steady state → compute pressure drop →
-    backward via matrix-free GMRES on unrelaxed physics residual →
-    exact analytical gradient dΔP/dU_inlet
+Demonstrates:
+    solve SIMPLE to steady state -> compute pressure drop ->
+    backward via matrix-free GMRES on unrelaxed physics residual ->
+    exact analytical gradient dDP/dU_inlet
 
-Compares against the analytical solution: dΔP/dU = 12μL/h² for
+Compares against the analytical solution: dDP/dU = 12*mu*L/h^2 for
 fully-developed Poiseuille flow.
 """
 

examples/02_topology_optimization.py

@@ -1,8 +1,8 @@
 """Example: topology optimization for minimum pressure drop.
 
-Demonstrates the C3 patent embodiment:
-    design variables → Helmholtz filter → smooth Heaviside →
-    Brinkman penalization → SIMPLE solve → pressure drop → implicit diff
+Demonstrates:
+    design variables -> Helmholtz filter -> smooth Heaviside ->
+    Brinkman penalization -> SIMPLE solve -> pressure drop -> implicit diff
 
 All within a single PyTorch autograd computational graph.
 """

examples/03_gymnasium_cylinder.py

@@ -1,8 +1,8 @@
 """Example: CylinderWakeEnv gymnasium environment.
 
-Demonstrates the C2 patent claim:
+Demonstrates:
     standard gymnasium.Env with gradient-attached step()
-    → policy_gradient() returns exact analytical gradients via implicit diff
+    -> policy_gradient() returns exact analytical gradients via implicit diff
 
 The environment supports two modes:
     Mode A: single-step contextual bandit (geometry optimization)

tests/validation/test_gradients.py

@@ -1,6 +1,6 @@
 """Gradient verification tests for implicit differentiation.
 
-C1 patent claim: matrix-free GMRES implicit differentiation gives exact
+Verifies that matrix-free GMRES implicit differentiation gives exact
 gradients. Verified by:
 1. torch.autograd.gradcheck on implicit diff backward
 2. Cross-validation: implicit diff vs finite differences

tests/validation/test_grid_convergence.py

@@ -1,8 +1,7 @@
 """Grid convergence study: Richardson extrapolation on lid-driven cavity.
 
-Runs the lid-driven cavity at Re=100 on 16², 32², 64² grids and verifies
-that the solution converges monotonically. This prevents the "numbers were
-tuned" critique during patent examination.
+Runs the lid-driven cavity at Re=100 on 16-squared, 32-squared, 64-squared grids and verifies
+that the solution converges monotonically.
 
 Uses Ghia et al. 1982 reference data for the centerline u-velocity profile.
 """