For a planar convex body K, let θ(K) denote the expected Euclidean distance between two independent points uniformly distributed on ∂K with respect to arclength. We prove the sharp inequality perKθ(K)>61. The constant is approached by rectangles degenerating to a segment.
Introduction
Let K⊂R2 be a convex body, and let B1,B2 be independent random points uniformly distributed on ∂K with respect to arclength. Put
θ(K)=E∣B1−B2∣,
and denote the perimeter of K by perK.
According to Tokmachev [3 Eq. (5)], Gusakova and Zaporozhets proposed the pair of sharp inequalities
61<perKθ(K)≤π22.
The upper bound, with equality only for the disk, was proved by Tokmachev [2]. The purpose of the present paper is to establish the complementary lower bound.
Theorem 1.1. For every planar convex body K,
θ(K)>6perK.
The constant 1/6 is sharp.
Our result is closely related to the Zaporozhets–Tarasov conjecture, which concerns the comparison of mean distances between independent uniform points inside a convex body and on its boundary. Bonnet, Gusakova, Thäle and Zaporozhets [1] proved that the mean distance Δ(K) between two independent uniform points in a planar convex body satisfies
Δ(K)<6perK.
Consequently, Theorem 1.1 gives another proof of the planar Zaporozhets–Tarasov inequality Δ(K)<θ(K). This inequality was previously established by a different approach in [5], where a stronger directional comparison was obtained.
Our proof is based on a comparison of one-dimensional distributions associated with portions of ∂K visible from a certain direction, which we call half-boundaries. We consider the tangent angle of a uniformly sampled point on a half-boundary and the median of three independent such angles. Using the concavity of the Gini mean difference under mixtures, we first bound θ(K) from below by an integral involving the median distribution. A probabilistic comparison then allows us to replace the median by a single tangent angle, after which Cauchy’s perimeter formula gives the desired constant 1/6. Finally, we examine the equality conditions to establish strictness.
We first recall the central notion and its basic properties.
Definition 1.2. For a probability measure ρ on R, write
D(ρ)=∬R2∣x−y∣dρ(x)dρ(y).
This is the Gini mean difference of ρ.
We use the following standard identities for the Gini mean difference; see also [4 Sec. 2.1.1].
Lemma 1.3. Let ρ,σ be probability measures on R with finite first moments, and let F,G be their distribution functions. Then
The argument we use to prove Theorem 1.1 is a comparison of one-dimensional distributions. The relevant laws are obtained by choosing either one random tangent direction or the median of three random tangent directions on a moving half-boundary.
Orient ∂K counterclockwise and parametrize it by arclength s∈[0,P], where P=perK:
r:[0,P]⟶∂K.
r′(s)=(cosα(s),sinα(s))for almost every s.
Its tangent angle α(s) is defined almost everywhere and admits a nondecreasing lift. Let μ be the pushforward of arclength under s↦α(s), and put
P=μ(S1),S(t)=μ([t,t+π]).
For each t, let It∈[0,π] have the normalized restriction of μ to [t,t+π], shifted by −t, and let Mt be the median of three independent copies of It. If ∂K is C2 with positive curvature, then μ(dα)=τ(α)dα, where τ is the radius of curvature.
Proposition 2.1.With the preceding notation,
θ(K)≥6P1∫02πS(t)2EsinMtdt.(4)
Proof. Fix t such that μ({t})=μ({t+π})=0. Such t form a set of full Lebesgue measure, since μ has at most countably many atoms. Project the body onto
ut=(−sint,cost).
The boundary ∂K splits into two arcs joining the two points where the projection onto ut attains its minimum and maximum. Let Γt and Γt+π be the two boundary arcs whose tangent angles range over [t,t+π] and [t+π,t+2π]. Their arclengths are S(t) and S(t+π)=P−S(t). Let μt,μt+π be the pushforwards of normalized arclength measures on these arcs under projection onto ut. Finally, let νt be the pushforward of normalized arclength measure on the whole boundary ∂K under projection onto ut. Since
∂K=Γt∪Γt+π,S(t)+S(t+π)=P,
::: we have
νt=PS(t)μt+PS(t+π)μt+π.
As the Gini functional is concave under mixtures,
D(νt)≥PS(t)D(μt)+PS(t+π)D(μt+π).(5)
Now
∣z∣=21∫0π∣⟨z,ut⟩∣dt,
and hence
θ(K)=21∫0πD(νt)dt.
Using 2.2,
θ(K)≥2P1∫0π[S(t)D(μt)+S(t+π)D(μt+π)]dt.
Projection of Γt+π onto ut+π=−ut reflects the law obtained by projection onto ut. Since the Gini mean difference is invariant under reflection, the change of variable r=t+π gives
where in the last step we used Cauchy’s perimeter formula. Therefore, the following Proposition together with Proposition 2.1 would conclude the proof of Theorem 1.1 with non-strict inequality.
Proof. Put b=π−a and let, as before, μ(dα) be the tangent-angle measure. Set
C:=∫02πS(t)2dt,fT(t):=CS(t)2,
and choose T∈[0,2π) with density fT. Conditional on T=t, choose three independent points X,Y,Z of the angle circle from normalized μ on the half-arc [t,t+π]. Write m=med(X,Y,Z), taking real representatives in this interval. Conditional on T=t, the laws of X−t and m−t are those of It and Mt, respectively. Define the events
Hence, conditional on X,Y, the start T is uniform among the half-windows containing both points.
If X=Y, then m=X, and the conditional comparison (12) is an equality. Fix a pair X=x, Y=y, order its points along their shorter arc, rotate the coordinates and μ, and lift the points to 0,d, where 0<d<π. Thus X,Y take the values 0,d in either order. Using the corresponding lift of the start, conditional on this fixed pair, T is uniform on
J=[d−π,0].
In particular, events with X and Y antipodal occur with probability zero, so we assume d<π.
For each t∈J, let Zt have normalized μ-law on [t,t+π], and put
mt=med(0,d,Zt),p(t)=P{mt∈[t+a,t+b]}.
Conditional on T=t, the third sample is independent of X,Y, so conditioning on the pair leaves its normalized μ-law on [t,t+π] unchanged. Consequently,
where the two interval intersections have the same length δ=max{0,min{b−a,b−d}}. Since X is one of these two fixed points, P(EX∣X,Y)=δ/(π−d). Thus the conditional comparison reduces to proving
∫Jp(t)dt≥δ.
If δ=0, this follows from p(t)≥0. Assume δ>0.
We estimate the probability that mt falls outside [t+a,t+b], on either its left or its right. The key observation is that the law of mt increases stochastically with t. Indeed, for 0<c≤d,
P{mt<c}=μ[t,c)+μ[c,t+π]μ[t,c).
As t increases, μ[t,c) decreases and μ[c,t+π] increases, so the displayed probability is nonincreasing. For c≤0 or c>d, the probability is constant.
Write
A(t)=P{mt<t+a},B(t)=P{mt>t+b}.
These are the probabilities of the two disjoint ways to miss the interval, so
p(t)=1−A(t)−B(t).
Since 0≤mt≤d and δ≤b−d,
A(t)B(t)=0(d−π≤t≤d−π+δ),=0(−δ≤t≤0).
Moreover, δ≤b−a and stochastic monotonicity give, for d−π≤t≤−δ,
A(t+δ)≤P{mt<t+δ+a}≤P{mt≤t+b}=1−B(t).
To integrate this comparison, first remove the intervals on which A and B vanish. For A, this gives
As b=π−a, this is (2.7). This argument also applies to non-smooth tangent-angle measures; choices at window endpoints affect only a Lebesgue-null set of starts.
The identity
sinr=∫0π/21[a,π−a](r)cosada,0≤r≤π,
and Tonelli’s theorem yield (2.8) by integrating (2.7) against cosada.
Proof of Theorem 1.1. As we outlined before Proposition 2.2,
Combining Propositions 2.1 and 2.2 with (2.11), we obtain
θ(K)≥6P.
It remains to exclude equality for a nondegenerate convex body. Suppose therefore that
θ(K)=6P.
Then both inequalities supplied by Propositions 2.2 and 2.1 must be equalities. Recall that we denote μ the tangent-angle measure of ∂K, of total mass P. We show that equality in (2.8) requires that no open semicircle contain three points of suppμ. Suppose otherwise. Choose small arcs U,V,W about three such points, lifted in this order to R, and a0>0 so that
v−x>a0,y−v>a0,y−x<π−a0(x∈U,v∈V,y∈W).
All three arcs have positive μ-measure. Fix 0<a<a0 and a pair x∈U, y∈W. Translate the angular coordinates so that x=0, y=d, and use mt, p(t), A(t), B(t) from the proof of Proposition 2.2. Here δ=π−d−a>0. The difference between the two sides of its integral comparison is
∫d−π0p(t)dt−δ=∫d−πd−π+a[1−B(t)−A(t+δ)]dt.
For t=d−π+s with 0<s<a, stochastic monotonicity gives
Indeed, the translated arc V lies in (a,d−a), hence in [s,d+s−a], and a third sample in V is the median. The excess is therefore at least aμ(V)/P for every pair in U×W. The pair disintegration in Proposition 2.2 now makes (10) strict for every 0<a<a0. Integration against cosada makes (11) strict, a contradiction. Now, we have shown that
∣suppμ∣≤4.
Indeed, if five points occurred in cyclic order and g1,…,g5 were the intervening angular gaps, then the absence of three points in an open semicircle would imply
gi+gi+1≥π(i(mod5)).
Summing gives 4π≥5π, since ∑igi=2π, which is impossible. Hence K is a polygon with at most four sides. We now show that the inequality of Proposition 2.1 is strict for every nondegenerate polygon. Choose an open interval J of angles t so that for every t∈J
ut=(−sint,cost)
is not normal to any edge of K. For t∈J the linear functional x↦⟨x,ut⟩ attains its minimum and maximum at unique vertices v and w, respectively. Those vertices split ∂K in two complementary polygonal chains. Denote their lengths by S1,S2, and let ν1,t,ν2,t be their normalized arclength measures projected onto ut.
The equality in the concavity step of Proposition 2.1 is equivalent to
ν1,t=ν2,t.
Let e1,e2 be the unit vectors along the edges of the two chains incident to v and directed away from v. For t∈J,
⟨ei,ut⟩>0.
Near the projection of v, the densities of ν1,t and ν2,t are respectively
S1⟨e1,ut⟩1,S2⟨e2,ut⟩1.
Therefore equality of the two projected laws would imply
⟨S1e1−S2e2,ut⟩=0.
As S1e1−S2e2=0, this can hold only for isolated values of t, whereas J is an interval. Hence the Gini concavity inequality is strict on a set of positive measure, and so Proposition 2.1 is strict for every nondegenerate polygon. Therefore,
θ(K)>6perK
Sharpness follows from the rectangles Kε=[0,1]×[0,ε]. As ε↓0, normalized boundary arclength concentrates on the two horizontal sides, and hence θ(Kε)→E∣U−V∣=1/3, where U,V∼Unif[0,1], while perKε→2.
References
[1]G. Bonnet, A. Gusakova, C. Thäle, and D. Zaporozhets, Sharp inequalities for the mean distance of random points in convex bodies, Adv. Math. 386 (2021), 107813.DOI
[2]A. S. Tokmachev, The Mean Distance Between Random Points on the Boundary of a Convex Figure, J. Math. Sci. 286 (2024), 798–806.DOI
[3]A. S. Tokmachev, Inequalities for convex functions of random points inside and on the boundary of convex bodies, arXiv:2607.08869, 2026.arxiv.org/abs/2607.08869
[4]S. Yitzhaki, Gini’s mean difference: a superior measure of variability for non-normal distributions, Metron 61 (2003), no. 2, 285–316.
[5]E. Shen, The Zaporozhets–Tarasov Conjecture on Mean Distances, arXiv:2608.06470v3, 2026.arxiv.org/abs/2608.06470