compute Newton step as opposed to Gauss-Newton step when J is square and full rank

Author: FHoltorf

src/trustRegion.jl

@@ -140,7 +140,7 @@ function SciMLBase.__solve(prob::NonlinearProblem,
 
     for k in 1:maxiters
         # Solve the trust region subproblem.
-        δ = dogleg_method(H, g, Δ)
+        δ = dogleg_method(∇f, F, g, Δ)
         xₖ₊₁ = x + δ
         Fₖ₊₁ = f(xₖ₊₁)
         fₖ₊₁ = 0.5 * norm(Fₖ₊₁)^2

src/utils.jl

@@ -58,9 +58,9 @@ function init_J(x)
     return J
 end
 
-function dogleg_method(H, g, Δ)
+function dogleg_method(J, f, g, Δ)
     # Compute the Newton step.
-    δN = -H \ g
+    δN = J \ (-f)
     # Test if the full step is within the trust region.
     if norm(δN) ≤ Δ
         return δN