Reason: Initial publication of the proof that a C^1 map is differentiable: telescoping along coordinate segments, one mean-value point per coordinate, and an epsilon/n estimate against continuity of the partials. Step 3 now fixes the auxiliary radius once as rho=(r-||h||)/2 and derives the bound ||h||+rho<r on the distance from a of every point of the coordinate slice. Claim-9 references point at lem:absolute-value-properties-2026b.
As in earlier arguments we use that squares are strictly monotone on nonnegative reals: if 0≤α, 0≤β and α<β, then 0<β by claim 2, so βα<ββ by claim 10 and αα≤βα, whence α2<β2 by claim 2; consequently α≤β whenever α2≤β2. In particular ∣zi∣≤∥z∥ for each i, since zi2≤∥z∥2 and ∣zi∣2=zi2 by claim 4 of Properties of the Absolute Value in an Ordered Field.
Step 1 (a ball inside U). Since a∈U and U is open, there is r with 0<r such that every point of Rn at Euclidean distance less than r from a lies in U.
Step 2 (telescoping). Let h=(h1,…,hn)∈Rn with ∥h∥<r. Define points p0,…,pn by p0=a and pm=pm−1[m:am+hm] for m∈{1,…,n}, so that pm has lth coordinate al+hl for l≤m and al for l>m; in particular pn=a+h.
More generally, for m∈{1,…,n} and u between am and am+hm inclusive, the point pm−1[m:u] differs from a only in coordinates l≤m, by hl for l<m and by u−am for l=m, and ∣u−am∣≤∣hm∣; so the square of its distance to a is at most ∑lhl2=∥h∥2, whence that distance is at most ∥h∥<r and the point lies in U. In particular every pm lies in U, and
f(a+h)−f(a)=m=1∑n(f(pm)−f(pm−1)).
Step 3 (a mean value point in each coordinate). Fix m. If hm=0 then pm=pm−1 and f(pm)−f(pm−1)=0=∂mf(qm)hm with qm=pm−1.
Suppose hm=0. Since ∥h∥<r, claim 1 gives 0<r−∥h∥, so by claim 8 the element ρ=(r−∥h∥)⋅2−1 satisfies 0<ρ and ρ+ρ=r−∥h∥, that is ∥h∥+ρ+ρ=r; adding ∥h∥+ρ to 0<ρ gives ∥h∥+ρ<r, again by claim 1. Let J be the open interval consisting of those u with am−(∣hm∣+ρ)<u and u<am+(∣hm∣+ρ).
Let u∈J. Adding −am to both inequalities, claim 1 gives −(∣hm∣+ρ)<u−am and u−am<∣hm∣+ρ, so ∣u−am∣<∣hm∣+ρ by claim 9 of Properties of the Absolute Value in an Ordered Field. The point pm−1[m:u] agrees with a in the coordinates l>m and differs from it by hl in the coordinates l<m and by u−am in the coordinate m; hence the square of its Euclidean distance to a equals ∑l<mhl2+(u−am)2. Here ∑l<mhl2≤∥h∥2−hm2, since the omitted terms hl2 with l>m are nonnegative, and (u−am)2=∣u−am∣2≤(∣hm∣+ρ)2 by claim 4 of that lemma and the monotonicity of squares. Moreover ∣hm∣ρ≤∥h∥ρ: this is claim 10 when ∣hm∣<∥h∥, and an equality when ∣hm∣=∥h∥. Using ∣hm∣2=hm2, again by claim 4, we get
By the monotonicity of squares the distance from pm−1[m:u] to a is therefore at most ∥h∥+ρ, hence less than r; so pm−1[m:u]∈U for every u∈J.
Let G:J→R be given by G(u)=f(pm−1[m:u]). For each u0∈J, apply Slice Function and the Partial Derivative to f at the point pm−1[m:u0] of U in the mth variable: it gives a positive radius ρ0 and identifies the slice function on (u0−ρ0,u0+ρ0), which agrees with G there, as differentiable at u0 with derivative ∂mf(pm−1[m:u0]), the partial derivative existing because f is of class C1. By An Open Interval is an Interval All of Whose Points Are Interior both intervals have all points interior, so shrinking the δ in the defining condition confines the increments to the overlap, where the two functions agree; hence G is differentiable at u0 with G′(u0)=∂mf(pm−1[m:u0]).
Both am and am+hm lie in J: indeed 0≤∣hm∣ and 0<ρ give 0<∣hm∣+ρ and ∣hm∣<∣hm∣+ρ by claim 1, so claim 9 of Properties of the Absolute Value in an Ordered Field, applied to 0 and to hm with c=∣hm∣+ρ, yields the two pairs of strict inequalities required. Applying Mean Value Theorem on an Open Interval to G on J with these two points in increasing order yields ξm strictly between them with G(am+hm)−G(am)=G′(ξm)hm, that is, with qm=pm−1[m:ξm],
f(pm)−f(pm−1)=∂mf(qm)hm.
Since ∣ξm−am∣<∣hm∣, Step 2 shows qm is at distance at most ∥h∥ from a.
Let ε∈R with 0<ε. Since n≥1, the element n, a sum of copies of 1, is positive by claims 6 and 3, so ε′=εn−1 is positive by claims 7 and 5. Each ∂mf is continuous at a, by the definition of class C1; taking the least of the finitely many radii by repeated use of claim 9, there is θ with 0<θ such that every z∈U at distance less than θ from a satisfies ∣∂mf(z)−∂mf(a)∣<ε′ for every m.
Let δ be the least, by claim 9, of r and θ, so 0<δ. Suppose h∈Rn satisfies 0<∑ihi2<δ2 and a+h∈U. Then ∥h∥<δ by the monotonicity of squares, so ∥h∥<r and Steps 2 and 3 apply, and each qm is at distance at most ∥h∥<θ from a, so ∣∂mf(qm)−∂mf(a)∣<ε′.
This is exactly the defining condition of differentiability of f at a, with m=1 there, the required partial derivatives existing because f is of class C1.