Ported for Lean Pool: imports relocated and code adapted to Lean/Mathlib v4.34.0.
Product-coordinate consequences of the semiconvex matrix lemma (Embeddings) #
Part of the product-coordinate development connecting the semiconvex matrix
conclusion on the coordinate space R^(n+n) to the block-matrix notation on
R^n × R^n. Split from ProductCoordinates.lean; see the umbrella module
docstring.
A local maximum on a set is a maximum on a smaller relative neighborhood.
In quantified mathematical form, if a ∈ S and a is a local maximum point
of f relative to S, then there is a set P such that P is a
neighborhood of a relative to S, a ∈ P, P ⊆ S, and a is a maximum
point of f on P.
View a function on R^n × R^n as a function on the coordinate space
R^(n+n).
Equations
Instances For
The coordinate map from R^(n+n) to R^n × R^n is continuous.
The point (x, 0) in the product coordinate space R^n × R^n.
Equations
Instances For
The point (0, y) in the product coordinate space R^n × R^n.
Equations
Instances For
The point (x, 0), written in R^(n+n) coordinates.
Equations
Instances For
The point (0, y), written in R^(n+n) coordinates.
Equations
Instances For
The left R^n coordinate of a point of R^(n+n) = R^n × R^n.
Equations
Instances For
The right R^n coordinate of a point of R^(n+n) = R^n × R^n.
Equations
Instances For
The left R^n component of a vector written in R^(n+n) coordinates.
Equations
- ViscositySolns.leftPointGradient p i = p (Fin.castAdd n i)
Instances For
The right R^n component of a vector written in R^(n+n) coordinates.
Equations
- ViscositySolns.rightPointGradient p i = p (Fin.natAdd n i)
Instances For
The left-left block of a Hessian written in R^(n+n) coordinates.
Equations
Instances For
The right-right block of a Hessian written in R^(n+n) coordinates.
Equations
Instances For
The left coordinate projection of a jet on R^n × R^n.
Equations
- ViscositySolns.leftPointJet J = { gradient := ViscositySolns.leftPointGradient J.gradient, hessian := ViscositySolns.leftPointHessian J.hessian }
Instances For
The right coordinate projection of a jet on R^n × R^n.
Equations
- ViscositySolns.rightPointJet J = { gradient := ViscositySolns.rightPointGradient J.gradient, hessian := ViscositySolns.rightPointHessian J.hessian }
Instances For
Convergence of block jets implies convergence of their left-coordinate projected jets.
Convergence of block jets implies convergence of the negated right-coordinate projected jets.
In standard mathematical terms, if J_i -> J as jets on R^n × R^n, then
the jets obtained by taking the right coordinate component and negating both
the gradient and Hessian also converge:
-(right J_i) -> -(right J).
The left embedding vanishes on the right coordinate block.
The right embedding recovers the original right coordinate.
The mixed second difference of a quadratic model in a left direction and a right direction depends only on the mixed Hessian terms.
In standard mathematical terms, for a quadratic polynomial on R^n × R^n,
the expression
q(z + (x, 0) + (0, y)) - q(z + (x, 0)) - q(z + (0, y)) + q(z)
is
(1 / 2) * (⟪X(x, 0), (0, y)⟫ + ⟪X(0, y), (x, 0)⟫).
The cross term between a left increment and a right increment may be computed in block coordinates.
In standard mathematical terms, reindexing the Hessian Z from R^{2n} to
R^n × R^n does not change the bilinear expression
⟪Z(x, 0), (0, y)⟫.
The cross term between a right increment and a left increment may be computed in block coordinates.
In standard mathematical terms, reindexing the Hessian Z from R^{2n} to
R^n × R^n does not change the bilinear expression
⟪Z(0, y), (x, 0)⟫.
The left-right cross term against coordinate basis vectors is the corresponding matrix entry.
In standard mathematical terms, for a block Hessian A,
⟪A e_i^L, e_j^R⟫ = A_{R_j,L_i}.
The right-left cross term against coordinate basis vectors is the corresponding matrix entry.
In standard mathematical terms, for a block Hessian A,
⟪A e_j^R, e_i^L⟫ = A_{L_i,R_j}.