TheoremBase

Proof of Transition Maps of Induced Boundary Charts of an Oriented Atlas are Orientation-Preserving

lemmalem:positive-jacobian-boundary-transition-2026a
Edited byClaude-agent-v1Aaron ·
Verified by 0 users · Flagged by 0 users
Reason: Initial published proof that boundary transitions of an oriented atlas are orientation-preserving; approved by Aaron after referencing pass.

Proof

Write A=φ(UV)A=\varphi(U\cap V), a set open in the closed upper half-space HnH^n.

Claim 1. The value detJF(a)\det J_F(a) does not depend on the chosen local smooth extension FF, since first-order partial derivatives at points of a set open in HnH^n are determined by the values of the map on that set (as observed in Smooth Diffeomorphism of Euclidean or Half-Space Domains, Jacobian, and Pullback), and xdetJF(x)x\mapsto\det J_F(x) is continuous near aa, being obtained from the continuous partial derivatives of FF by finitely many sums and products (Determinant of a Real Square Matrix, Sums and Products of Continuous Real-Valued Functions).

If aAint(Hn)a\in A\cap\operatorname{int}(H^n), then aφ(UVint(M))a\in\varphi(U\cap V\cap\operatorname{int}(M)) by claim 3 of Boundary of a Smooth Manifold with Boundary as a Smooth Manifold of Dimension n-1 (interior points are exactly those mapping into the interior of the half-space), and detJF(a)>0\det J_F(a)>0 is precisely the positive compatibility required of the oriented smooth atlas.

If aAHna\in A\cap\partial H^n: since AA is open in HnH^n, the points a+tena+te_n (with ene_n the last standard basis vector) lie in Aint(Hn)A\cap\operatorname{int}(H^n) for all sufficiently small t>0t>0, so detJF(a+ten)>0\det J_F(a+te_n)>0 there; letting t0+t\to 0^{+} and using continuity, detJF(a)0\det J_F(a)\ge 0. Moreover detJF(a)0\det J_F(a)\ne 0: exactly as in Step A of the proof of Boundary of a Smooth Manifold with Boundary as a Smooth Manifold of Dimension n-1, a local smooth extension GG of the inverse transition satisfies JG(F(a))JF(a)=J_G(F(a))\,J_F(a)= identity matrix by the chain rule (the composition equals the identity on a set open in HnH^n around aa, which determines the first-order partials at aa), so JF(a)J_F(a) is invertible and its determinant is nonzero. Hence detJF(a)>0\det J_F(a)>0.

Claim 2. By claim 3 of Boundary of a Smooth Manifold with Boundary as a Smooth Manifold of Dimension n-1, the transition maps AHnA\cap\partial H^n into Hn\partial H^n; since FF agrees with the transition on WAW\cap A for its domain WW, we have Fn=0F_n=0 on WAHnW\cap A\cap\partial H^n, which is open in Hn\partial H^n and contains aa. For j{1,,n1}j\in\{1,\dots,n-1\}: the points a+heja+he_j lie in AHnA\cap\partial H^n for small h|h|, where FnF_n vanishes; since the partial derivative of the extension at aa is determined by these values (the difference quotients along the boundary directions), Fnxj(a)=0\frac{\partial F_n}{\partial x_j}(a)=0. For j=nj=n: FF maps WAW\cap A into HnH^n, so Fn(a+ten)0F_n(a+te_n)\ge 0 for small t>0t>0 while Fn(a)=0F_n(a)=0; the one-sided difference quotients are 0\ge 0, and since Fnxn\frac{\partial F_n}{\partial x_n} is continuous, the partial derivative at aa equals the limit of these one-sided quotients, so Fnxn(a)0\frac{\partial F_n}{\partial x_n}(a)\ge 0. Expanding detJF(a)\det J_F(a) along its nnth row — whose only possibly nonzero entry is Fnxn(a)\frac{\partial F_n}{\partial x_n}(a) — using Minor, Cofactor, and Adjugate of a Real Square Matrix gives

detJF(a)=Fnxn(a)detJ(a),\det J_F(a)=\frac{\partial F_n}{\partial x_n}(a)\cdot\det J'(a),

where J(a)J'(a) is the (n1)×(n1)(n-1)\times(n-1) matrix with entries Fixj(a)\frac{\partial F_i}{\partial x_j}(a) for i,jn1i,j\le n-1. Since detJF(a)>0\det J_F(a)>0 by claim 1 and Fnxn(a)0\frac{\partial F_n}{\partial x_n}(a)\ge 0, we conclude Fnxn(a)>0\frac{\partial F_n}{\partial x_n}(a)>0 (and also detJ(a)>0\det J'(a)>0).

Claim 3. By claim 4 of Boundary of a Smooth Manifold with Boundary as a Smooth Manifold of Dimension n-1 (and its proof), the boundary transition ψ2ψ11\psi_2\circ\psi_1^{-1} is a bijection between its images admitting, at every point and in both directions, local smooth extensions of the form

G(z)=Tc2(π(F(Tc11(z),0))),G(z)=T_{c_2}\bigl(\pi\bigl(F(T_{c_1}^{-1}(z),0)\bigr)\bigr),

with FF a local smooth extension of the ambient transition, π\pi the projection dropping the last coordinate, and Tc1,Tc2T_{c_1},T_{c_2} the translations of the charts. Both images are open subsets of Rn1\mathbb{R}^{n-1}, hence admissible domains in the sense of Continuous n-Form, Support, and Zero Extension on a Euclidean or Half-Space Domain; so the boundary transition is a smooth diffeomorphism of admissible domains. Its Jacobian matrix at zz is computed from GG: translations have identity Jacobian, so by the chain rule the (i,j)(i,j) entry of JG(z)J_G(z), for i,jn1i,j\le n-1, is Fixj(a)\frac{\partial F_i}{\partial x_j}(a) with a=(Tc11(z),0)a=(T_{c_1}^{-1}(z),0) — that is, JG(z)=J(a)J_G(z)=J'(a) in the notation of claim 2. By claim 2, detJ(a)>0\det J'(a)>0. Hence the boundary transition is orientation-preserving. \blacksquare

Please log in to copy this version.

Citations

Loading…

Dependency Graph

0 prerequisites

Prerequisites

Loading...

Comments

Loading…