Throughout, the notation is that of the statement.
Claim 1. Apply claim 2 of Concatenation Identifies a Product of Euclidean Spaces with a Euclidean Space with ξ=ξ′=0Rm and η=η′=0Rn, in the form recorded there for differences:
ι(0Rm−0Rm,0Rn−0Rn)=ι(0Rm,0Rn)−ι(0Rm,0Rn).
In the real vector space Rq one has z−z=0Rq for every z∈Rq. The left-hand side is therefore ι(0Rm,0Rn) and the right-hand side is 0RN, which is the first assertion.
For the second assertion, put, for k∈N,
ak=dE(ξk,ξ),bk=dE(ηk,η),ck=dE(ι(ξk,ηk),ι(ξ,η)),
which are nonnegative, being Euclidean norms of differences by the identity dE(x,y)=∥x−y∥ recorded in the statement together with claim 1 of Elementary Properties of the Euclidean Norm on Rn. By claim 4 of Concatenation Identifies a Product of Euclidean Spaces with a Euclidean Space,
ck2=ak2+bk2.
Since bk2 and ak2 are nonnegative, this gives ak2≤ck2 and bk2≤ck2, hence ak≤ck and bk≤ck by claim 2 of Monotonicity of Squaring on the Nonnegative Elements of an Ordered Field, the numbers involved being nonnegative. Also 2akbk is nonnegative by claim 5 of Elementary Order Arithmetic in an Ordered Field, so
ck2=ak2+bk2≤ak2+2akbk+bk2=(ak+bk)2,
and the same claim gives ck≤ak+bk.
Suppose first that (ι(ξk,ηk)) converges to ι(ξ,η), and let ε∈R be positive. By Convergent Sequence in a Metric Space there is K∈N with ck<ε for every k≥K; then ak≤ck<ε and bk≤ck<ε for every such k. Hence (ξk) converges to ξ and (ηk) converges to η.
Conversely, suppose (ξk) converges to ξ and (ηk) converges to η, and let ε∈R be positive. The number 2ε is positive by claim 8 of Elementary Order Arithmetic in an Ordered Field, so there are K1,K2∈N with ak<2ε for k≥K1 and bk<2ε for k≥K2. For k at least as large as both K1 and K2 we get ck≤ak+bk<2ε+2ε=ε. Hence (ι(ξk,ηk)) converges to ι(ξ,η).
Claim 2. Let x∈RN. Since ι is a bijection there are ξ∈Rm and η∈Rn with x=ι(ξ,η), so that w(x)=u1(ξ)+u2(η). From u1(ξ)≤C1 and u2(η)≤C2 and the compatibility of ≤ with addition in an ordered field we obtain w(x)≤C1+C2. Thus C1+C2 is an upper bound for the set of values of w. Since the sets of values of u1, u2 and w all have upper bounds in R and 0<λ, the sup-convolutions u1λ, u2λ and wλ are defined by Sup-Convolution of a Function on RM.
Claim 3. Fix ξ∈Rm and η∈Rn and put z=ι(ξ,η), s1=u1λ(ξ) and s2=u2λ(η). Write
S1={u1(a)−2λ∥a−ξ∥2:a∈Rm},S2={u2(b)−2λ∥b−η∥2:b∈Rn},
S={w(x)−2λ∥x−z∥2:x∈RN}.
These are the sets denoted Sλ,u1(ξ), Sλ,u2(η) and Sλ,w(z) in Sup-Convolution of a Function on RM, and by that definition s1, s2 and wλ(z) are their least upper bounds; in particular all three sets are nonempty and bounded above.
Let x∈RN and let a∈Rm, b∈Rn be the unique pair with x=ι(a,b). By claim 2 of Concatenation Identifies a Product of Euclidean Spaces with a Euclidean Space we have x−z=ι(a−ξ,b−η), and hence by claim 3 there
∥x−z∥2=∥a−ξ∥2+∥b−η∥2.
Using the distributive law in the form 2λ(s+t)=2λs+2λt together with the commutativity and associativity of addition, it follows that
w(x)−2λ∥x−z∥2=(u1(a)−2λ∥a−ξ∥2)+(u2(b)−2λ∥b−η∥2).
As x ranges over RN the pair (a,b) ranges over all of Rm×Rn, because ι is a bijection. Therefore S is exactly the set of sums t1+t2 with t1∈S1 and t2∈S2.
We show that s1+s2 is the least upper bound of S. First, if t1∈S1 and t2∈S2 then t1≤s1 and t2≤s2, so t1+t2≤s1+s2; hence s1+s2 is an upper bound for S.
Second, let t∈R be an upper bound for S and suppose, for a contradiction, that t<s1+s2. Put
ε=2(s1+s2)−t,
which is positive by claim 8 of Elementary Order Arithmetic in an Ordered Field. Then s1−ε<s1, so s1−ε is not an upper bound for S1 and there is t1∈S1 with s1−ε<t1; likewise there is t2∈S2 with s2−ε<t2. Adding,
t=(s1+s2)−2ε<t1+t2,
and t1+t2∈S, contradicting that t is an upper bound for S. Hence s1+s2≤t for every upper bound t of S.
The two paragraphs together say that s1+s2 is the least upper bound of S, so wλ(z)=s1+s2, which is the asserted identity.