TheoremBase

Proof

Throughout, R\mathbb{R} carries the operations and the order ≤\le of its ordered field structure, where s−ts-t abbreviates s+(−t)s+(-t) and s<ts<t means that s≤ts\le t and s≠ts\ne t; the order ≤\le is a total order and is in particular reflexive. Regard Rn\mathbb{R}^n as a metric space through the Euclidean distance dEd_E, which is a metric by Euclidean Distance is a Metric on Rn\mathbb{R}^n; this is the metric used in the local extremum notions below. For x∈Ux\in U write Du(x)Du(x) for the gradient of uu at xx and D2u(x)D^2u(x) for the Hessian matrix of uu at xx.

Step 1 (The test function uu itself). Take φ=u\varphi=u; this function is of class C2C^2 on UU by hypothesis, so it is admissible as a test function in either viscosity notion, and Dφ(x)=Du(x)D\varphi(x)=Du(x) and D2φ(x)=D2u(x)D^2\varphi(x)=D^2u(x) for every x∈Ux\in U. For this choice the function u−φ:U→Ru-\varphi:U\to\mathbb{R} of that definition has value u(y)−u(y)u(y)-u(y) at y∈Uy\in U, and u(y)−u(y)=0u(y)-u(y)=0 by claim 3 of Additive Cancellation and Elementary Additive Identities in a Field. Write w:U→Rw:U\to\mathbb{R} for this function, so that w(y)=0w(y)=0 for every y∈Uy\in U.

Step 2 (Every point of UU is both a local maximum and a local minimum of ww). Let x∈Ux\in U. By claim 6 of Elementary Order Arithmetic in an Ordered Field we have 0<10<1 in R\mathbb{R}, so δ=1\delta=1 is an admissible radius. Every y∈Uy\in U with dE(x,y)<1d_E(x,y)<1 satisfies w(y)=0w(y)=0 and w(x)=0w(x)=0, whence w(y)≤w(x)w(y)\le w(x) and w(x)≤w(y)w(x)\le w(y) by reflexivity of ≤\le. Therefore ww has a local maximum at xx relative to UU and also a local minimum at xx relative to UU.

Step 3 (Proof of claim 1). Assume that uu is a viscosity subsolution of FF on UU and let x∈Ux\in U. By Steps 1 and 2 the function u−φu-\varphi associated with the test function φ=u\varphi=u has a local maximum at xx relative to UU, so the defining condition of a viscosity subsolution, applied to this φ\varphi and this xx, gives

F(x,u(x),Dφ(x),D2φ(x))≤0,F(x,u(x),D\varphi(x),D^2\varphi(x))\le 0 ,

that is, F(x,u(x),Du(x),D2u(x))≤0F(x,u(x),Du(x),D^2u(x))\le 0. Since x∈Ux\in U was arbitrary and uu is of class C2C^2 on UU, this is exactly the statement that uu is a classical subsolution of FF on UU.

Step 4 (Proof of claim 2). Assume that uu is a viscosity supersolution of FF on UU and let x∈Ux\in U. By Steps 1 and 2 the same function u−φu-\varphi, with φ=u\varphi=u, has a local minimum at xx relative to UU, so the defining condition of a viscosity supersolution gives

0≤F(x,u(x),Dφ(x),D2φ(x)),0\le F(x,u(x),D\varphi(x),D^2\varphi(x)) ,

that is, 0≤F(x,u(x),Du(x),D2u(x))0\le F(x,u(x),Du(x),D^2u(x)). Since x∈Ux\in U was arbitrary and uu is of class C2C^2 on UU, uu is a classical supersolution of FF on UU.

Citations

Loading…

Dependencies

Uses0

Loading…

Comments

Log in to comment.

Loading…