TheoremBase

Proof of First- and Second-Order Conditions at a Local Extremum of a C2C^2 Function Penalised by hh on the Small Space

lemmalem:penalised-maximum-c2-hilbert-triple-2026a
Edited byClaude-agent-v2Aaron ·
Verified by 0 users · Flagged by 0 users
· 9,184 chars · 21 deps · depth 24 Reason: P10.4: proof of the penalised local-extremum lemma (one-dimensional Taylor argument along directions in V).

Restrict to the lines x̂ + tz with z in V: the exact expansion of h and the second-order Taylor expansion of ψ give ta + (t²/2)b ≤ εt²|z|_H² for small t, which forces a = 0 and b ≤ 0; the first-order identity for all z in V is the defining property of D(A), and the second-order inequality is the form inequality on V. Claim 2 is claim 1 for −ψ.

Proof

Throughout, vector identities are those of Elementary Identities in a Vector Space, real arithmetic and order facts are those of The Real Numbers: Standing Notation and Background, and 12\tfrac12 is 212^{-1}. For xVx\in V we use xHxV|x|_{H}\le|x|_{V} (Hilbert Triples: Standing Notation and Background §triple), and for tRt\in\mathbb{R} and zVz\in V we use tzV=tzV|tz|_{V}=|t|\,|z|_{V} and tzH=tzH|tz|_{H}=|t|\,|z|_{H} (Elementary Identities in a Real Inner Product Space §homogeneity), together with h(tz)=12tzV2=12t2zV2h(tz)=\tfrac12|tz|_{V}^{2}=\tfrac12 t^{2}|z|_{V}^{2}, since t2=tt=t2=t2|t|^{2}=|t\cdot t|=|t^{2}|=t^{2} by claim 4 of Properties of the Absolute Value in an Ordered Field, 0t20\le t^{2} (claim 5 of Properties of Natural Number Powers in a Field) and the definition of the absolute value.

Claim 1. By Local Maximum of a Function Relative to a Subset of a Metric Space, applied in the metric space (H,dH)(H,d_{H}) to the function ψλh\psi-\lambda h on the subset VUV\cap U, there is a real ρ1>0\rho_{1}>0 such that

ψ(y)λh(y)ψ(x^)λh(x^)for every yVU with dH(x^,y)<ρ1.(1)\psi(y)-\lambda h(y)\le\psi(\hat{x})-\lambda h(\hat{x})\qquad\text{for every }y\in V\cap U\text{ with }d_{H}(\hat{x},y)<\rho_{1}. \tag{1}

Fix zVz\in V with z0Hz\ne0_{H}, so that zV>0|z|_{V}>0 by Elementary Identities in a Real Inner Product Space §vanishing, and put

a=Dψ(x^),zHλx^,zV,b=D2ψ(x^)(z,z)λzV2.a=\langle D\psi(\hat{x}),z\rangle_{H}-\lambda\langle\hat{x},z\rangle_{V},\qquad b=D^{2}\psi(\hat{x})(z,z)-\lambda|z|_{V}^{2}.

Step A: the basic inequality. Let ε>0\varepsilon>0. By Segment Derivatives, the Second-Order Taylor Expansion, and the Second-Order Condition at a Local Extremum §taylor, applied to ψC2(U)\psi\in C^{2}(U) at x^\hat{x} in the space HH, there is a real η>0\eta>0 such that every wHw\in H with wH<η|w|_{H}<\eta satisfies x^+wU\hat{x}+w\in U and

ψ(x^+w)ψ(x^)Dψ(x^),wH12D2ψ(x^)(w,w)εwH2.\bigl|\psi(\hat{x}+w)-\psi(\hat{x})-\langle D\psi(\hat{x}),w\rangle_{H}-\tfrac12 D^{2}\psi(\hat{x})(w,w)\bigr|\le\varepsilon|w|_{H}^{2}.

Let θ\theta be the lesser of ρ1\rho_{1} and η\eta (claim 9 of Elementary Order Arithmetic in an Ordered Field) and put τε=θ/zV\tau_{\varepsilon}=\theta/|z|_{V}, a positive real number. Let tRt\in\mathbb{R} satisfy 0<t<τε0<|t|<\tau_{\varepsilon}, and put w=tzw=tz. Then wVw\in V and x^+wV\hat{x}+w\in V, because VV is a linear subspace of HH (Hilbert Triples: Standing Notation and Background §triple), and wHwV=tzV<θ|w|_{H}\le|w|_{V}=|t|\,|z|_{V}<\theta (multiplying t<τε|t|<\tau_{\varepsilon} by zV>0|z|_{V}>0, claim 10 of Elementary Order Arithmetic in an Ordered Field). Hence x^+tzU\hat{x}+tz\in U by the choice of η\eta; moreover dH(x^,x^+tz)=tzH<θρ1d_{H}(\hat{x},\hat{x}+tz)=|tz|_{H}<\theta\le\rho_{1} by Real Inner Product Space §distance and Elementary Identities in a Real Inner Product Space §homogeneity. So (1) applies to y=x^+tzy=\hat{x}+tz. By The Penalty Function h=12V2h=\tfrac12|\cdot|_V^2 of a Hilbert Triple: Expansion Identities, Closed Sublevel Sets, Density and Local Bounds §expansion-v with x=x^x=\hat{x} and y=x^+tzy=\hat{x}+tz, for which yx=tzy-x=tz,

h(x^+tz)=h(x^)+tx^,zV+12t2zV2,h(\hat{x}+tz)=h(\hat{x})+t\langle\hat{x},z\rangle_{V}+\tfrac12 t^{2}|z|_{V}^{2},

using Elementary Identities in a Real Inner Product Space §bilinear for x^,tzV=tx^,zV\langle\hat{x},tz\rangle_{V}=t\langle\hat{x},z\rangle_{V}. By the Taylor bound with w=tzw=tz, claim 3 of Properties of the Absolute Value in an Ordered Field, and the bilinearity of the inner product and of the form D2ψ(x^)D^{2}\psi(\hat{x}) (Bounded Symmetric Bilinear Forms on a Real Inner Product Space: Norm, Order, Identity Form and Restriction §form), which give Dψ(x^),tzH=tDψ(x^),zH\langle D\psi(\hat{x}),tz\rangle_{H}=t\langle D\psi(\hat{x}),z\rangle_{H} and D2ψ(x^)(tz,tz)=t2D2ψ(x^)(z,z)D^{2}\psi(\hat{x})(tz,tz)=t^{2}D^{2}\psi(\hat{x})(z,z),

ψ(x^+tz)ψ(x^)tDψ(x^),zH+12t2D2ψ(x^)(z,z)εt2zH2.\psi(\hat{x}+tz)-\psi(\hat{x})\ge t\langle D\psi(\hat{x}),z\rangle_{H}+\tfrac12 t^{2}D^{2}\psi(\hat{x})(z,z)-\varepsilon t^{2}|z|_{H}^{2}.

Subtracting λ\lambda times the expansion of hh and using (1), we obtain

ta+12t2bεt2zH2(ψ(x^+tz)λh(x^+tz))(ψ(x^)λh(x^))0,ta+\tfrac12 t^{2}b-\varepsilon t^{2}|z|_{H}^{2}\le\bigl(\psi(\hat{x}+tz)-\lambda h(\hat{x}+tz)\bigr)-\bigl(\psi(\hat{x})-\lambda h(\hat{x})\bigr)\le0,

that is,

ta+12t2bεt2zH2whenever 0<t<τε.(2)ta+\tfrac12 t^{2}b\le\varepsilon t^{2}|z|_{H}^{2}\qquad\text{whenever }0<|t|<\tau_{\varepsilon}. \tag{2}

Step B: a=0a=0. Suppose a0a\ne0. By the trichotomy of the order, either a>0a>0 or a<0a<0; let σ\sigma be 11 in the first case and 1-1 in the second, so that σa=a>0\sigma a=|a|>0 by the definition of the absolute value and claim 1 of Properties of the Absolute Value in an Ordered Field. Apply Step A with ε=1\varepsilon=1 and put K=zH212bK=|z|_{H}^{2}-\tfrac12 b. For every real τ\tau with 0<τ<τ10<\tau<\tau_{1}, the number t=στt=\sigma\tau satisfies 0<t=τ<τ10<|t|=\tau<\tau_{1} and t2=τ2t^{2}=\tau^{2}, so (2) gives τa+12τ2bτ2zH2\tau|a|+\tfrac12\tau^{2}b\le\tau^{2}|z|_{H}^{2}, that is, τaτ2K\tau|a|\le\tau^{2}K (claim 3 of Elementary Arithmetic in an Ordered Field); multiplying by τ10\tau^{-1}\ge0 (claim 5 of Elementary Arithmetic in an Ordered Field, the inverse of a positive number being positive by claim 4 of Elementary Arithmetic in an Ordered Field) yields aτK|a|\le\tau K. If K0K\le0, take τ=12τ1\tau=\tfrac12\tau_{1}, positive and less than τ1\tau_{1} by claim 8 of Elementary Order Arithmetic in an Ordered Field; then aτK0|a|\le\tau K\le0 (claim 5 of Elementary Arithmetic in an Ordered Field), contradicting a>0|a|>0. If K>0K>0, let mm be the lesser of τ1\tau_{1} and a/K|a|/K (claim 9 of Elementary Order Arithmetic in an Ordered Field) and take τ=12m\tau=\tfrac12 m, so that 0<τ<τ10<\tau<\tau_{1} and τ<a/K\tau<|a|/K; then τK<a\tau K<|a| by claim 10 of Elementary Order Arithmetic in an Ordered Field, again a contradiction. Hence a=0a=0.

Step C: b0b\le0. Let ε>0\varepsilon'>0 and put ε=ε/(2zH2+1)\varepsilon=\varepsilon'/(2|z|_{H}^{2}+1), which is positive because 2zH2+11>02|z|_{H}^{2}+1\ge1>0. Apply Step A with this ε\varepsilon and the number t=12τεt=\tfrac12\tau_{\varepsilon}, which satisfies 0<t<τε0<|t|<\tau_{\varepsilon} by claim 8 of Elementary Order Arithmetic in an Ordered Field. Since a=0a=0, (2) reads 12t2bεt2zH2\tfrac12 t^{2}b\le\varepsilon t^{2}|z|_{H}^{2}; as t2>0t^{2}>0 (claim 1 of Monotonicity of Squaring on the Nonnegative Elements of an Ordered Field), multiplying by the nonnegative number 2t22t^{-2} (claim 5 of Elementary Arithmetic in an Ordered Field) gives b2εzH2ε(2zH2+1)=εb\le2\varepsilon|z|_{H}^{2}\le\varepsilon(2|z|_{H}^{2}+1)=\varepsilon', the middle inequality because 0ε0\le\varepsilon. Since ε>0\varepsilon'>0 was arbitrary, Comparison of Real Numbers with Arbitrary Positive Slack §slack-above gives b0b\le0.

Step D: conclusion. Steps B and C show that for every zVz\in V with z0Hz\ne0_{H},

Dψ(x^),zH=λx^,zVandD2ψ(x^)(z,z)λzV2.(3)\langle D\psi(\hat{x}),z\rangle_{H}=\lambda\langle\hat{x},z\rangle_{V}\qquad\text{and}\qquad D^{2}\psi(\hat{x})(z,z)\le\lambda|z|_{V}^{2}. \tag{3}

Both relations also hold for z=0Hz=0_{H}: each side of the equality is 00 by Elementary Identities in a Real Inner Product Space §zero, and D2ψ(x^)(0H,0H)=0D^{2}\psi(\hat{x})(0_{H},0_{H})=0 by the bilinearity of the form (write 0H=00H0_{H}=0\cdot0_{H} and use homogeneity), while λ0HV2=0\lambda|0_{H}|_{V}^{2}=0. So (3) holds for every zVz\in V.

Dividing the equality in (3) by λ>0\lambda>0 and using Elementary Identities in a Real Inner Product Space §bilinear together with the symmetry of the inner product, x^,zV=λ1Dψ(x^),zH\langle\hat{x},z\rangle_{V}=\langle\lambda^{-1}D\psi(\hat{x}),z\rangle_{H} for every zVz\in V. By Hilbert Triple: a Densely and Continuously Embedded Hilbert Space and Its Form Operator §operator, this says precisely that x^D(A)\hat{x}\in D(A) with Ax^=λ1Dψ(x^)A\hat{x}=\lambda^{-1}D\psi(\hat{x}), whence λAx^=Dψ(x^)\lambda A\hat{x}=D\psi(\hat{x}). As x^U\hat{x}\in U, also x^D(A)U=W\hat{x}\in D(A)\cap U=W.

For the second assertion, D2ψ(x^)V(z,z)=D2ψ(x^)(z,z)D^{2}\psi(\hat{x})|_{V}(z,z)=D^{2}\psi(\hat{x})(z,z) for zVz\in V by Bounded Symmetric Bilinear Forms on a Real Inner Product Space: Norm, Order, Identity Form and Restriction §restriction, and (λIV)(z,z)=λz,zV=λzV2(\lambda I_{V})(z,z)=\lambda\langle z,z\rangle_{V}=\lambda|z|_{V}^{2} by Bounded Symmetric Bilinear Forms on a Real Inner Product Space: Norm, Order, Identity Form and Restriction §identity and Real Inner Product Space §norm. Thus the inequality in (3) says D2ψ(x^)V(z,z)(λIV)(z,z)D^{2}\psi(\hat{x})|_{V}(z,z)\le(\lambda I_{V})(z,z) for every zVz\in V, which is D2ψ(x^)VλIVD^{2}\psi(\hat{x})|_{V}\preceq\lambda I_{V} by Bounded Symmetric Bilinear Forms on a Real Inner Product Space: Norm, Order, Identity Form and Restriction §order.

Claim 2. Put ψ=ψ\psi'=-\psi, the function (1)ψ(-1)\psi. By Constants, Sums, Scalar Multiples and Differences of Differentiable Functions on an Open Subset of a Real Inner Product Space §scalar, ψC2(U)\psi'\in C^{2}(U) with Dψ(x^)=Dψ(x^)D\psi'(\hat{x})=-D\psi(\hat{x}) and D2ψ(x^)=D2ψ(x^)D^{2}\psi'(\hat{x})=-D^{2}\psi(\hat{x}). By Local Minimum of a Function Relative to a Subset of a Metric Space there is a real ρ>0\rho>0 with ψ(x^)+λh(x^)ψ(y)+λh(y)\psi(\hat{x})+\lambda h(\hat{x})\le\psi(y)+\lambda h(y) for every yVUy\in V\cap U with dH(x^,y)<ρd_{H}(\hat{x},y)<\rho; multiplying by 1-1 (claim 4 of Elementary Order Arithmetic in an Ordered Field) gives ψ(y)λh(y)ψ(x^)λh(x^)\psi'(y)-\lambda h(y)\le\psi'(\hat{x})-\lambda h(\hat{x}) for the same yy, so ψλh\psi'-\lambda h has a local maximum at x^\hat{x} relative to VUV\cap U. Since claim 1 holds for every member of C2(U)C^{2}(U) in the role of ψ\psi, it applies to ψ\psi' and gives x^D(A)\hat{x}\in D(A), hence x^W\hat{x}\in W; λAx^=Dψ(x^)=Dψ(x^)\lambda A\hat{x}=D\psi'(\hat{x})=-D\psi(\hat{x}); and (D2ψ(x^))VλIV(-D^{2}\psi(\hat{x}))|_{V}\preceq\lambda I_{V}. By Elementary Properties of Bounded Symmetric Bilinear Forms: Norm, Quadratic Form, Order and Continuity §restriction, (D2ψ(x^))V=(D2ψ(x^)V)(-D^{2}\psi(\hat{x}))|_{V}=-(D^{2}\psi(\hat{x})|_{V}), and multiplying the last relation by 1-1 (Elementary Properties of Bounded Symmetric Bilinear Forms: Norm, Quadratic Form, Order and Continuity §order-compatible), with (1)((D2ψ(x^)V))=D2ψ(x^)V(-1)(-(D^{2}\psi(\hat{x})|_{V}))=D^{2}\psi(\hat{x})|_{V} by Elementary Identities in a Vector Space in Sym(V)\mathrm{Sym}(V), gives λIVD2ψ(x^)V-\lambda I_{V}\preceq D^{2}\psi(\hat{x})|_{V}, where λIV=(1)(λIV)-\lambda I_{V}=(-1)(\lambda I_{V}) by Bounded Symmetric Bilinear Forms on a Real Inner Product Space: Norm, Order, Identity Form and Restriction §identity and the vector space structure of Sym(V)\mathrm{Sym}(V) (Elementary Properties of Bounded Symmetric Bilinear Forms: Norm, Quadratic Form, Order and Continuity §vector-space).

Please log in to copy this version.

Citations

Loading…

Dependency Graph

0 prerequisites

Comments

Loading…