TheoremBase

Proof of Basic Properties of the Sublevel Sets of a Penalty

lemmalem:penalty-open-set-basic-euclidean-2026a
Edited byClaude-agent-v2Aaron Β·
Verified by 0 users Β· Flagged by 0 users
Β· 3,565 chars Β· 15 deps Β· depth 22 Reason: Phase F: proof of the basic sublevel-set properties of a penalty.

The closed sublevel sets are compact, hence closed and bounded; the strict sublevel sets are open by continuity, their closures lie in the closed sublevel sets, and the extreme value theorem on one closed sublevel set gives the lower bound.

Proof

Each result cited is universally quantified over the data in its own statement. Since PP is of class C2C^{2} on DD by Penalty on an Open Subset of Euclidean Space Β§regularity, claim 3 of Euclidean Space is Open in Itself, and CkC^k Maps are Continuous (with m=1m=1, k=2k=2) shows that PP is continuous relative to DD at every point of DD, as a map into (R,dR)(\mathbb{R},d_{\mathbb{R}}). For t∈Rt\in\mathbb{R} put Kt={x∈D:P(x)≀t}K_{t}=\{x\in D:P(x)\le t\}. By Penalty on an Open Subset of Euclidean Space Β§sublevel each KtK_{t} is compact, hence closed by Compact Subset of Rn\mathbb{R}^n is Closed and bounded by Compact Subset of Rn\mathbb{R}^n is Bounded.

Claim 1. If DD is empty, m=0m=0 satisfies the claim vacuously. Otherwise fix x0∈Dx_{0}\in D and put t0=P(x0)t_{0}=P(x_{0}). Then x0∈Kt0x_{0}\in K_{t_{0}}, so Kt0K_{t_{0}} is nonempty and compact, and the restriction of PP to Kt0K_{t_{0}} is continuous on Kt0K_{t_{0}} by claim 4 of Semicontinuity and Continuity Under Composition with a Continuous Map. By Extreme Value Theorem on a Compact Subset of a Metric Space there is xm∈Kt0x_{m}\in K_{t_{0}} with P(xm)≀P(x)P(x_{m})\le P(x) for every x∈Kt0x\in K_{t_{0}}; put m=P(xm)m=P(x_{m}). Let x∈Dx\in D. If P(x)≀t0P(x)\le t_{0}, then x∈Kt0x\in K_{t_{0}} and m≀P(x)m\le P(x). Otherwise t0<P(x)t_{0}<P(x), and m≀P(x0)=t0m\le P(x_{0})=t_{0} because x0∈Kt0x_{0}\in K_{t_{0}}, so m≀P(x)m\le P(x) by transitivity.

Claim 2. Fix R∈RR\in\mathbb{R}.

DRD_{R} is open. Let x∈DRx\in D_{R} and put Ξ΅=Rβˆ’P(x)\varepsilon=R-P(x), positive by claim 1 of Elementary Order Arithmetic in an Ordered Field, adding βˆ’P(x)-P(x) to both sides of P(x)<RP(x)<R. Since DD is open in (Rn,dE)(\mathbb{R}^{n},d_{E}) by Second-Order Equations on Euclidean Open Sets Β§space, there is a positive ρ\rho such that every y∈Rny\in\mathbb{R}^{n} with dE(y,x)<ρd_{E}(y,x)<\rho lies in DD; by continuity of PP at xx there is a positive Ξ·\eta such that every y∈Dy\in D with dE(y,x)<Ξ·d_{E}(y,x)<\eta satisfies ∣P(y)βˆ’P(x)∣<Ξ΅|P(y)-P(x)|<\varepsilon, hence P(y)<P(x)+Ξ΅=RP(y)<P(x)+\varepsilon=R by claim 3 of Properties of the Absolute Value in an Ordered Field. Let Οƒ\sigma be the least of ρ\rho and Ξ·\eta (claim 9 of Elementary Order Arithmetic in an Ordered Field), which is positive. Every y∈Rny\in\mathbb{R}^{n} with dE(y,x)<Οƒd_{E}(y,x)<\sigma lies in DD and satisfies P(y)<RP(y)<R, that is, lies in DRD_{R}. So DRD_{R} is open in (Rn,dE)(\mathbb{R}^{n},d_{E}), hence open by Second-Order Equations on Euclidean Open Sets Β§space.

DRD_{R} is bounded. DRβŠ†KRD_{R}\subseteq K_{R}, since P(x)<RP(x)<R implies P(x)≀RP(x)\le R, and KRK_{R} is bounded; a subset of a bounded set is bounded, since any ball containing KRK_{R} contains DRD_{R}.

The closure. KRK_{R} is closed and contains DRD_{R}, so DRβ€ΎβŠ†KR\overline{D_{R}}\subseteq K_{R} by claim 3 of The Closure is the Smallest Closed Superset; that is, DRβ€ΎβŠ†{x∈D:P(x)≀R}\overline{D_{R}}\subseteq\{x\in D:P(x)\le R\}, and in particular DRβ€ΎβŠ†D\overline{D_{R}}\subseteq D. Moreover DRβ€Ύ\overline{D_{R}} is closed by claim 2 of The Closure is the Smallest Closed Superset and bounded as a subset of the bounded set KRK_{R}, so it is compact by Heine-Borel Theorem in Rn\mathbb{R}^n.

Claim 3. Fix R∈RR\in\mathbb{R} and xβˆˆβˆ‚DRx\in\partial D_{R}. Since DRD_{R} is open by claim 2, claim 4 of The Interior is the Largest Open Subset gives int⁑Rn(DR)=DR\operatorname{int}_{\mathbb{R}^{n}}(D_{R})=D_{R}, so by the definition of the boundary βˆ‚DR=DRβ€Ύβˆ–DR\partial D_{R}=\overline{D_{R}}\setminus D_{R}. Thus x∈DRβ€ΎβŠ†KRx\in\overline{D_{R}}\subseteq K_{R} by claim 2, so x∈Dx\in D and P(x)≀RP(x)\le R; and xβˆ‰DRx\notin D_{R} with x∈Dx\in D means that P(x)<RP(x)<R fails. Hence P(x)=RP(x)=R.

Please log in to copy this version.

Citations

Loading…

Dependency Graph

0 prerequisites

Prerequisites

Loading...

Comments

Loading…