/bio/skills/blog/math

The Invariance of Boundary Theorem

The Invariance of Boundary Theorem

Given, MM, a manifold with a boundary, the interior and boundary are disjoint. (from Mathematical Foundations of Artificial Intelligence: Basics of Manifold Theory*)

Proof

Assume, for contradiction that pp is in Int MM and on M∂M. Note that by definition, for a point pp to be in the interior of MM means there's a homeomorphism φ\varphi: Int M{(x1,,xn)R+nxn>0}M \to \{(x_1, \dots, x_n) \in \mathbb{R}^n_{+} | x_n > 0 \}. Recall that a point pp is on the boundary of MM means there's a homeomorphism ψ:M{(x1,,xn)R+nxn=0}\psi: ∂M \to \{(x_1, \dots, x_n) \in \mathbb{R}^n_{+} | x_n = 0 \}.

At pp there must be a transition map, ψ(φ1)\psi(\varphi^{-1}) that maps from {(x1,,xn)R+nxn>0}\{(x_1, \dots, x_n) \in \mathbb{R}^n_{+} | x_n > 0 \} to {(x1,,xn)R+nxn=0}\{(x_1, \dots, x_n) \in \mathbb{R}^n_{+} | x_n = 0 \}. Since the transition map is a composition of bijections, it too, is a bijection.

Let y,z>0y,z \gt 0 and yzy \neq z. Consider the points p1=(x1,,xn1,y)p_1 = (x_1, \dots, x_{n-1}, y) and p2=(x1,,xn1,z)p_2 = (x_1, \dots, x_{n-1}, z). Note that p1p2p_1 \neq p_2, but our transition map sends both p1p1 and p2p_2 to the same point (x1,,xn1,0)(x_1, \dots, x_{n-1}, 0). Thus we have our contradiction and reject the assumption that a point pp can be both in Int MM and on M∂M. This concludes the proof. \blacksquare

*Xiong, M. (2026). Mathematical Foundations of Artificial Intelligence: Basics of Manifold Theory (1st ed.). page 29. Chapman and Hall/CRC. https://doi.org/10.1201/9781003641452

2026 Stefano De Vuono