TheoremBase

Sketch: homogeneity at λ = 0 gives p(0) = 0, subadditivity at x and −x gives the reflection, and subadditivity across x − y gives the Lipschitz bound.

Proof

For (i): 0V=0⋅0V0_{V}=0\cdot0_{V} by Elementary Identities in a Vector Space (ii), so Sublinear Functional (ii) with λ=0\lambda=0 gives

p(0V)=p(0⋅0V)=0⋅p(0V)=0.p(0_{V})=p(0\cdot0_{V})=0\cdot p(0_{V})=0 .

For (ii), let x∈Vx\in V. Since x+(−x)=0Vx+(-x)=0_{V} (The Zero Vector, Negatives and Differences in a Vector Space (ii)), the value at zero and Sublinear Functional (i) give

0=p(0V)=p(x+(−x))≤p(x)+p(−x),0=p(0_{V})=p\bigl(x+(-x)\bigr)\le p(x)+p(-x),

that is, −p(−x)≤p(x)-p(-x)\le p(x).

For (iii), let MM and ff be as there and let x∈Mx\in M. The upper inequality f(x)≤p(x)f(x)\le p(x) is the hypothesis. For the lower one, −x=(−1)x-x=(-1)x by Elementary Identities in a Vector Space (iv), so −x∈M-x\in M by Linear Subspace of a Real Vector Space (iii), and f(−x)=−f(x)f(-x)=-f(x) by Linear Functional on a Linear Subspace (ii). The hypothesis at −x-x gives −f(x)=f(−x)≤p(−x)-f(x)=f(-x)\le p(-x), that is, −p(−x)≤f(x)-p(-x)\le f(x).

For (iv), let ∥⋅∥\lVert\cdot\rVert be a norm on VV. It is subadditive by Norms and Real Normed Spaces (iv), and for every real λ≥0\lambda\ge0 and x∈Vx\in V, Norms and Real Normed Spaces (iii) gives ∥λx∥=∣λ∣ ∥x∥=λ∥x∥\lVert\lambda x\rVert=|\lambda|\,\lVert x\rVert=\lambda\lVert x\rVert. So ∥⋅∥\lVert\cdot\rVert is a sublinear functional.

For (v), let ∥⋅∥\lVert\cdot\rVert and CC be as there and let x,y∈Vx,y\in V. Since x=y+(x−y)x=y+(x-y) (Elementary Identities in a Vector Space (viii)), Sublinear Functional (i) and the hypothesis give

p(x)≤p(y)+p(x−y)≤p(y)+C∥x−y∥.p(x)\le p(y)+p(x-y)\le p(y)+C\lVert x-y\rVert .

Exchanging the roles of xx and yy gives p(y)≤p(x)+C∥y−x∥p(y)\le p(x)+C\lVert y-x\rVert, and ∥y−x∥=∥x−y∥\lVert y-x\rVert=\lVert x-y\rVert: indeed y−x=−(x−y)=(−1)(x−y)y-x=-(x-y)=(-1)(x-y) by Elementary Identities in a Vector Space (xi) and Elementary Identities in a Vector Space (iv), and Norms and Real Normed Spaces (iii) applies with λ=−1\lambda=-1. The two inequalities together give ∣p(x)−p(y)∣≤C∥x−y∥|p(x)-p(y)|\le C\lVert x-y\rVert.

For (vi), let ∥⋅∥\lVert\cdot\rVert, CC, MM and ff be as there and let x∈Mx\in M. By (iii) and the hypothesis on pp,

f(x)≤p(x)≤C∥x∥,−f(x)≤p(−x)≤C∥−x∥=C∥x∥,f(x)\le p(x)\le C\lVert x\rVert,\qquad -f(x)\le p(-x)\le C\lVert -x\rVert=C\lVert x\rVert,

where ∥−x∥=∥(−1)x∥=∥x∥\lVert -x\rVert=\lVert(-1)x\rVert=\lVert x\rVert by Elementary Identities in a Vector Space (iv) and Norms and Real Normed Spaces (iii). Hence ∣f(x)∣≤C∥x∥|f(x)|\le C\lVert x\rVert.

Citations

Loading…

Dependencies

Uses0

Loading…

Comments

Log in to comment.

Loading…