GaussNewtonModel.gradient

t3toolbox.fitting.GaussNewtonModel.gradient()

def gradient(self) -> t3m.T3Tangent:  # g = Π 𝒥ᵀ r (+ g_R), a tangent at frame

The gradient g = geometry.project(𝒥ᵀ r) (the Gauss-Newton Jᵀr), a tangent at frame, plus the regularizer gradient g_R when set.

On the manifold geometry this is the gauged Riemannian gradient; on the corewise geometry it is the raw core gradient 𝒥ᵀr (a tangent at (U,G,G,G), no Π).

Return type:

T3Tangent