A ‘converse’ to the Constraint Lemma
E. Kolpakov
Supported by the Russian Foundation
for Basic Research grant 19-01-00169
We begin the article with the statement of the main result.
For the history and motivation see Remark 1; the current paper gives a partial answer to [Sk18, Problem 4.4.c].
Direct proofs of implications between classical theorems are interesting.
The main result is a direct proof of the implication (LVKFk,3)⇒(LTT3k−1,3) below.
Consider the following statements:
(LVKF1,3) From any 11 points in R3 one can choose 3 pairwise disjoint triples whose convex hulls have a common point.
(LVKFk,3) From any 6k+5 points in R3k one can choose 3 pairwise disjoint sets each containing 2k+1 points and whose convex hulls have a common point.
(LTT2,3) Any 7 points in R2 can be decomposed into 3 subsets whose convex hulls have a common point.
(LTTd,3) Any 2d+3 points in Rd can be decomposed into 3 subsets whose convex hulls have a common point.
These statements are correct and known, but the purpose of this paper is the direct deduction of one statement from another.
Remark 1.
Consider the following piecewise linear versions (or, equivalently, topological versions) of the above statements.
Denote by ΔN the N-dimensional simplex.
(VKFk,r) (the r-fold van Kampen-Flores conjecture) For any general position PL (piecewise linear) map Δ(kr+2)(r−1)→Rkr there are r pairwise disjoint k(r−1)-faces whose images have a common point.
(TTd,r) (the topological Tverberg conjecture) For any PL map Δ(d+1)(r−1)→Rd there are r pairwise disjoint faces whose images have a common point.
These conjectures are true for r a prime power, and are false for r not a prime power, see the surveys [BZ], [BBZ], [Sk18].
The implication (TTkr+1,r)⇒(VKFk,r) is called the Constraint Lemma [Gr10, BFZ14, Fr15], for a history see [Sk18, Remark 1.11].
This is one of the steps (the simplest) in the disproof of the topological Tverberg conjecture, see the surveys [BZ], [BBZ], [Sk18].
Our result (LVKFk,3)⇒(LTT3k−1,3) can be considered as a ‘converse’ to the Constraint Lemma.
However, (LTTd,3) is correct for every d, so our direct proof does not give any new results.
In addition to the above, there are the following implications for r=2, see the survey [Sk17]:
TTd,2⇒TTd−1,2;
TTd,2⇒VKFd−1,2 for even d;
VKFd,2⇒VKFd−1,2;
VKFd−1,2⇒VKFd,2 for odd d;
For direct proofs of other similar implications see [Sk17] and references therein.
The most novel part of our proof a simple proof of lemma 1, which is a weaker version of [Tv66, Lemma 3].
Lemmas 2,3,4 below are simple and so could be known.
Denote the convex hull of a set P by ⟨P⟩.
A set of points in the Rn is called in general position if it contains no k points in (k−2)-dimensional subspace for any k≤n+1.
Lemma 1. Let
∙* A be a set of 6k+1 points of general position in R3k−1×0⊂R3k;*
∙* A1 be a vertex of the convex hull ⟨A⟩;*
∙* A1×0,M1,M2,M3,M4 be points in the line A1×R⊂R3k−1×R=R3k in order of increasing the last coordinate;*
∙* Δ1,Δ2,Δ3⊂A∪{M1,M2,M3,M4} be pairwise disjoint subsets such that M2∈/⟨Δ1⟩∩⟨Δ2⟩∩⟨Δ3⟩=∅.*
Then there exist pairwise disjoint subsets Δ1′,Δ2′,Δ3′⊂A∪{M1,M2,M3,M4} such that
-
⟨Δ1′⟩∩⟨Δ2′⟩∩⟨Δ3′⟩=∅;
2. 2.
some two the points of A1,M1,M2,M3,M4 do not belong to
Δ1′∪Δ2′∪Δ3′;
3. 3.
(Δ1′∪Δ2′∪Δ3′)∩A1×R⊂(Δ1∪Δ2∪Δ3)∩A1×R.
Proof.
The point X is called higher than Y if last coordinate of X is greater than last coordinate of Y.
Consider all triples of pairwise disjoint subsets δ1,δ2,δ3⊂A∪{M1,M2,M3,M4} such that (δ1∪δ2∪δ3)∩A1×R⊂(Δ1∪Δ2∪Δ3)∩A1×R . The union of the intersections ⟨δ1⟩∩⟨δ2⟩∩⟨δ3⟩ is non-empty, closed and bounded. Denote by Z′ the point with the smallest last coordinate from this union.
Therefore Z′∈⟨Δ1′⟩∩⟨Δ2′⟩∩⟨Δ3′⟩ for some pairwise disjoint subsets Δ1′,Δ2′,Δ3′⊂A∪{M1,M2,M3,M4}. From the sets Δ1′,Δ2′,Δ3′ there is at least one, which not contain the points M3,M4. This set is contained in A∪{M1,M2}, and its convex hull besides the point M2 is below than the point M2. Hence the point Z′ is below than the point M2. Since the point A1 is a vertex of ⟨A⟩, at least one of the sets ⟨Δ1′⟩,⟨Δ2′⟩,⟨Δ3′⟩ intersects the line A1×R in no more than one point. Since Z′=M2, Z′ does not belong to line A1×R.
Each of polyhedrons ⟨Δ1′⟩,⟨Δ2′⟩,⟨Δ3′⟩ can be replaced by a simplex from its triangulation containing the point Z′. Therefore without loss of generality we consider that the polyhedrons ⟨Δ1′⟩,⟨Δ2′⟩,⟨Δ3′⟩ are simplexes. Replace each simplexes ⟨Δ1′⟩,⟨Δ2′⟩,⟨Δ3′⟩ by its face such that the point Z′ became is unique point of the intersection ⟨Δ1′⟩∩⟨Δ2′⟩∩⟨Δ3′⟩, and does not belong the faces of the simplexes ⟨Δ1′⟩,⟨Δ2′⟩,⟨Δ3′⟩. Since dim⟨Δ1′⟩+dim⟨Δ2′⟩+dim⟨Δ3′⟩≤6k, we have ∣Δ1′∣+∣Δ2′∣+∣Δ3′∣≤6k+3.
If any point from A−{A1} belong to some set from Δ1′,Δ2′,Δ3′ then from the points A1,M1,M2,M3,M4 at least two do not belong in any of the sets Δ1′,Δ2′,Δ3′ as required.
If Z′ belong to plane R3k−1×0 then for any i=1,2,3 we have Z′∈R3k−1×0∩⟨Δi′⟩=⟨α∩Δi′⟩. Therefore the sets R3k−1×0∩Δ1′,R3k−1×0∩Δ2′,R3k−1×0∩Δ3′ are required.
Next, we consider the remaining case where exists a point Y from the set A−{A1} that does not belong to any of the sets Δ1′,Δ2′,Δ3′, and the point Z′ does not lie on the plane R3k−1×0. Then Z′ have positive last coordinate. Consider a convex polyhedron M:=⟨Y∪Δ1′⟩∩⟨Y∪Δ2′⟩∩⟨Y∪Δ3′⟩, and proove that Z′ is a his vertex. Suppose that Z′ is not a vertex of polyhedron M, then there is a segment I⊂M, whose middle point is Z′. For any i=1,2,3 we have: I⊂⟨Y∪Δi′⟩, Z′ does not belong to face of simplex ⟨Δi′⟩. By Lemma 2 for any i=1,2,3 there is a segment Ji⊂I such that Ji⊂⟨Δi′⟩ and Z′is a middle Ji. Then for J=J1∩J2∩J3 we have J⊂⟨Δ1′⟩∩⟨Δ2′⟩∩⟨Δ3′⟩ but Z′ is a unique point of the intersection ⟨Δ1′⟩∩⟨Δ2′⟩∩⟨Δ3′⟩. This contradiction means that Z′ is the vertex of the polyhedron M.
For any i=1,2,3 denote τi=Δi′∩(A−{A1}). Denote by Z′′ orthogonal projection of point Z′ onto R3k−1×0. Then Z′′∈⟨A1∪τ1⟩∩⟨A1∪τ2⟩∩⟨A1∪τ3⟩ and Z′′=A1. Consider the ray L with vertex in A1 which contains a point Z′′. This ray intersect a simplexes ⟨τ1⟩,⟨τ2⟩,⟨τ3⟩. For any i=1,2,3 denote the point of intersection (possibly non-single) of the ray L and simplex ⟨τi⟩ by Vi. The segments [A1;V1],[A1;V2],[A1;V3] are ordered by inclusion. Without less of generality let [A1;V3]⊂[A1;V2]⊂[A1;V1].
Then V3∈⟨A1∪τ1⟩∩⟨A1∪τ2⟩∩⟨τ3⟩ and ⟨A1∪τ1⟩∩⟨A1∪τ2⟩∩⟨τ3⟩=∅.
In the next two paragraphs prove that ∣Δ1′∣+∣Δ2′∣+∣Δ3′∣=6k+3.
If there is a simplex among ⟨Δ1′⟩,⟨Δ2′⟩,⟨Δ3′⟩ which dimension is 3k then it contains the point Z′ as internal.
Denote by (i1,i2,i3) a permutation of numbers (1,2,3) such that dim⟨Δi3′⟩=3k and i1<i2
Since the set A is in general position in R3k−1×0, we have dim⟨A1∪τi1⟩∩⟨A1∪τi2⟩=1+dim⟨A1∪τi1⟩∩⟨τi2⟩=∣τi1∣+∣τi2∣−(3k−1).
Since [A1;Z′′]⊂⟨A1∪τi1⟩∩⟨A1∪τi2⟩, we have dim⟨A1∪τi1⟩∩⟨A1∪τi2⟩≥1. Hence ∣τi1∣+∣τi2∣≥3k therefore
∣Δi1′∣+∣Δi2′∣+∣Δi3′∣≥(∣τi1∣+1)+(∣τi2∣+1)+(3k+1)≥6k+3 as required.
If dimensions of simplexes ⟨Δ1′⟩,⟨Δ2′⟩,⟨Δ3′⟩ not more than 3k−1 then dimensions of simplexes ⟨τ1⟩,⟨τ2⟩,⟨τ3⟩ not more than 3k−2
and this simplexes are faces of simplexes ⟨A1∪τ1⟩,⟨A1∪τ2⟩,⟨A1∪τ3⟩ respectively.
Since the points of the set {A1}∪τ1∪τ2 is in general position, we have dim⟨A1∪τ1⟩∩⟨A1∪τ2⟩=1+dim⟨τ1⟩∩⟨A1∪τ2⟩=∣τ1∣+∣τ2∣−(3k−1).
Since the points of the set {A1}∪τ1∪τ2 is in general position with the points of the set τ3, we have
[TABLE]
[TABLE]
[TABLE]
Since ⟨A1∪τ1⟩∩⟨A1∪τ2⟩∩⟨τ3⟩=∅, we have ∣τ1∣+∣τ2∣+∣τ3∣−(6k−1)=dim⟨A1∪τ1⟩∩⟨A1∪τ2⟩∩⟨τ3⟩≥0.
Since from the points A1,M1,M2,M3,M4 at least 4 belong to the union Δ1′∪Δ2′∪Δ3′, we have ∣Δ1′∣+∣Δ2′∣+∣Δ3′∣≥4+∣τ1∣+∣τ2∣+∣τ3∣=6k+3 as required.
Since ∣Δ1′∣+∣Δ2′∣+∣Δ3′∣=6k+3, we have ∣Y∪Δ1′∣+∣Δ2′∣+∣Δ3′∣=6k+4. Therefore
dim⟨Y∪Δ1′⟩+dim⟨Δ2′⟩+dim⟨Δ3′⟩≥6k+1.
Hence dim⟨Y∪Δ1′⟩∩⟨Δ2′⟩∩⟨Δ3′⟩≥1.
It is similarly proved that dim⟨Δ1′⟩∩⟨Y∪Δ2′⟩∩⟨Δ3′⟩≥1 and
dim⟨Δ1′⟩∩⟨Δ2′⟩∩⟨Y∪Δ3′⟩≥1.
The pairwise intersections of the sets ⟨Y∪Δi1′⟩∩⟨Δi2′⟩∩⟨Δi3′⟩ for all permutations {i1,i2,i3}={1,2,3} are an ⟨Δ1′⟩∩⟨Δ2′⟩∩⟨Δ3′⟩=Z′. Since ⟨Y∪Δi1′⟩∩⟨Y∪Δi2′⟩∩⟨Δi3′⟩ contains ⟨Y∪Δi1′⟩∩⟨Δi2′⟩∩⟨Δi3′⟩ and ⟨Δi1′⟩∩⟨Y∪Δi2′⟩∩⟨Δi3′⟩, we have dim⟨Y∪Δi1′⟩∩⟨Y∪Δi2′⟩∩⟨Δi3′⟩≥2 for {i1,i2,i3}={1,2,3}. By Lemma 3 any face of ⟨Y∪Δi1′⟩∩⟨Y∪Δi2′⟩∩⟨Δi3′⟩ which contains the point Z′, is contained in some of the sets ⟨Y∪Δi1′⟩∩⟨Δi2′⟩∩⟨Δi3′⟩ or ⟨Δi1′⟩∩⟨Y∪Δi2′⟩∩⟨Δi3′⟩.
Then any edge of the polyhedron M with vertex in Z′ is contained in one of the sets ⟨Y∪Δ1′⟩∩⟨Δ2′⟩∩⟨Δ3′⟩, ⟨Δ1′⟩∩⟨Y∪Δ2′⟩∩⟨Δ3′⟩, ⟨Δ1′⟩∩⟨Δ2′⟩∩⟨Y∪Δ3′⟩. Since the point Z′ above than the point Y, Z′ is not the lowest point of a polyhedron M. Then Z′ is not the lowest vertex of a polyhedron M. Therefore there is edge [Z′;Z1] such that the point Z1 is an below Z′. Without loss of generality [Z′;Z1]⊂⟨Y∪Δ1′⟩∩⟨Δ2′⟩∩⟨Δ3′⟩. Since the point Z1 is an below Z′, we have contradiction with choice of the point Z′.
Lemma is proved.
Lemma 2. Take a simplex S in the space Rn.
Take a point Y in general position with vertexes of S.
Take a segment I⊂⟨Y∪S⟩.
Denote by Z the middle point of I.
Suppose that Z∈S but Z does not belong any face of S.
Then there is a non-degenerate segment J⊂I such that J⊂S and Z is a midpoint of J.
Proof.
If dimS=n, then Z is interior point of simplex S. There is closed ϵ-neighborhood of the point Z which belong S for some ϵ>0. Take as J the intersection of such a neighborhood with I. Then Z will be the middle of the segment J and J⊂S as required.
If dimS≤n−1, then S is a face of the simplex ⟨Y∪S⟩. Since the point Z is a middle of the segment I and Z∈S, we have that both ends of the segment I belong to S. Take J=I. Lemma is proved.
Lemma 3. Take convex polyhedra A1,...,An in Rd which have more than one common point.
Then for any point P in any face of A1∩...∩An there are faces Aj′ of Aj, j=1,2,…,n, such that
[TABLE]
Proof. It suffices to prove the proposition for n=2.
Suppose that there is a point P∈(A1∩A2)′ such that P∈/A1′ and P∈/A2′.
Denote by d1 and d2 the dimensions of the polyhedra A1 and A2 respectively. Take the disks D1 and D2 with dimensions d1 and d2 respectively such that P∈D1⊂A1 and P∈D2⊂A2. The intersection D1∩D2 is a disk with dimension d1+d2−d=dimA1∩A2. Since D1∩D2⊂A1∩A2, the point P does not belong to face of A1∩A2. A contradiction.
Denote the affine hull of a set P by \mboxaffP.
Lemma 4. Let P a finite general position subset of R3k−1 and ϵ>0 a number. Then there is δ∈(0,ϵ) with the following property. For any two subsets P1,P2 of P such that the intersection ⟨P1⟩∩⟨P2⟩ is one point
we have that ϵ-neighbourhood of this point ⟨P1⟩∩⟨P2⟩ contains the intersection of δ-neighbourhoods of
⟨P1⟩ and ⟨P2⟩.
Proof.
Since P is finite, there is only a finite number of pairs (P1,P2) of subsets of P such that ⟨P1⟩∩⟨P2⟩
is a point, but none of the subsets is a point.
Let α>0 be the smallest of any angles between ⟨P1⟩ and ⟨P2⟩.(The angle between two intersecting affine subspaces of R3k−1, none of which is a point, is the minimal
angle beetwen two rays originating from an intersection point of the subspaces,
the first ray contained in the first subspace and the second ray in the second subspaces.)
Let δ=ϵsin(α/2).
Take any point E from the intersection of δ-neighbourhoods of ⟨P1⟩ and ⟨P2⟩.
Denote by E1 and E2 the orthogonal projections of E on \mboxaffP1 and \mboxaffP2 respectively.
Denote by E′ the point
of intersection \mboxaffP1∩\mboxaffP2.
We have ∠E1E′E+∠E2E′E≥∠E1E′E2. So without loss of generality we can assume that ∠E1E′E≥21∠E1E′E2. Therefore EE′=sin∠EE′E1EE1<sin(α/2)δ=ϵ. QED
Proof that (LVKFk,3)⇒(LTT3k−1,3). Suppose to the contrary that the statement (LTT3k−1,3) is false for points A1,…,A6k+1 in R3k−1=R3k−1×0⊂R3k.
Without loss of generality assume that A1 is a vertex of ⟨A1,…,A6k+1⟩.
Construction of 6k+5 points to apply (LVKFk,3).
We can assume that the points A1,…,A6k+1 are in general position.
Consider all distances from any of these 6k+1 points to a simplex formed by some other of these 6k+1 points.
For any three pairwise disjoint sets P1,P2,P3 of these 6k+1 points consider the positive distance from ⟨P1⟩ to ⟨P2⟩∩⟨P3⟩.
Choose ϵ>0 such that 10ϵ is smaller than the smallest of all the considered numbers.
Denote by
[TABLE]
the orthogonal projection.
Denote by πK:R3k−{K}→R3k−1×0 the central projection from a point K∈/R3k−1×0.
Take δ∈(0,ϵ) given by Lemma 4 for the set {A1,…,A6k+1} and the number ϵ>0.
Consider points Mj:=(A1,mj), j=1,2,3,4, where the numbers mj are defined recursively.
Let m1:=1.
If m1,…,mj−1 are defined, then let mj be so big that
∣πMjX,πX∣<δ for every X∈M1,…,Mj−1,A1,…,A6k+1.
Let us prove that there exist pairwise disjoint sets
[TABLE]
such that ⟨Δ1⟩∩⟨Δ2⟩∩⟨Δ3⟩=∅ and {M1,M2,M3,M4,A1}⊂Δ1∪Δ2∪Δ3. Assume the contrary, then for any neighborhood of the point A1 in R3k−1 there exists a point A1′ from this neighborhood such that for any three pairwise disjoint subsets S1,S2,S3⊂{M1,M2,M3,M4,A1′,A2,...,A6k+1} such that ⟨S1⟩∩⟨S2⟩∩⟨S3⟩=∅ we obtain {M1,M2,M3,M4,A1′}⊂S1∪S2∪S3 and A1′∈⟨A1,…,A6k+1⟩.
Apply (LVKFk,3) to the set {M1,M2,M3,M4,A1′,A2,...,A6k+1}, we get pairwise disjoint sets S1,S2,S3 such that ∣S1∣=∣S2∣=∣S3∣=2k+1.
Since {M1,M2,M3,M4,A1′}⊂S1∪S2∪S3, one of the sets S1,S2,S3 contains a point A1′.
Without loss of generality, we assume that A1′∈S1.
Since ∣S1∣+∣S2∣+∣S3∣=6k+3, we get some point from A2,…,A6k+1, which not belong in any of the sets S1,S2,S3.
Without loss of generality, we assume that A6k+1∈/S1∪S2∪S3.
Then S1,S2,S3⊂{M1,M2,M3,M4,A1,A1′,A2,...,A6k} and A1∈/S1,S2,S3.
Therefore, among the sets S1∩(A1×R),S2∩(A1×R),S3∩(A1×R) there are at least 2 one-point ones.
Then ⟨S1⟩∩⟨S2⟩∩⟨S3⟩∩(A1×R)=∅.
Therefore M2∈/⟨S1⟩∩⟨S2⟩∩⟨S3⟩.
Apply Lemma 1 to {M1,M2,M3,M4,A1,A1′,A2,...,A6k}, we get the sets S1′,S2′,S3′.
Then ∣{A1,M1,M2,M3,M4}∩(S1′∪S2′∪S3′)∣≤3 и A1∈/S1′∪S2′∪S3′.
This contradicts the fact that for any three pairwise disjoint subsets S1,S2,S3⊂{M1,M2,M3,M4,A1′,A2,...,A6k+1} such that ⟨S1⟩∩⟨S2⟩∩⟨S3⟩=∅ we obtain {M1,M2,M3,M4,A1′}⊂S1∪S2∪S3.
We have proved that there exist pairwise disjoint sets
[TABLE]
such that ⟨Δ1⟩∩⟨Δ2⟩∩⟨Δ3⟩=∅ and {M1,M2,M3,M4,A1}⊂Δ1∪Δ2∪Δ3.
For this Δ1,Δ2,Δ3 we have M2∈/⟨Δ1⟩∩⟨Δ2⟩∩⟨Δ3⟩.
Apply Lemma 1 to the set {M1,M2,M3,M4,A1,A2,...,A6k+1}.
We get the sets Δ1′,Δ2′,Δ3′ such that ⟨Δ1′⟩∩⟨Δ2′⟩∩⟨Δ3′⟩=∅ and there are 2 points from A1,M1,M2,M3,M4 such that do not belong to the sets Δ1′,Δ2′,Δ3′.
Denote by Z any point from ⟨Δ1′⟩∩⟨Δ2′⟩∩⟨Δ3′⟩.
Case 1: Δh′⊂{A1,…,A6k+1} for some h.
Then ⟨Δh′⟩⊂R3k−1×0.
Hence Z∈R3k−1×0.
Take i=1,2,3.
Since the points M1,M2,M3,M4 belong to the same half-space, ⟨Δi′⟩∩(R3k−1×0)=⟨Δi′∩(R3k−1×0)⟩.
Then the set Z∈⟨Δi′∩(R3k−1×0)⟩ for i=1,2,3.
Therefore sets Δi′∩(R3k−1×0) for i=1,2,3 constitute the required partition.
Case 2: none of Δ1′,Δ2′,Δ3′ is contained in {A1,…,A6k+1}. Hence each set Δi′ has only one vertex among the points {M1,M2,M3,M4} and A1∈/Δi′ for i=1,2,3. Take the highest vertex of each Δi′.
Denote these vertices by W1,W2,W3 in the increasing order along 0×R.
Without loss of generality Wi∈Δi′ for each i=1,2,3.
Denote by Δi− the image of Δi′−Wi under the central projection πWi from Wi to R3k−1×0. Then Δi−=Δi∩R3k−1×0. Therefore A1∈/Δi− for i=2,3. Since πΔ1′=A1∪(Δ1′∩R3k−1×0), it follows that pairwise intersections of πΔ1′,Δ2−,Δ3− are empty.
Since Z∈⟨Δi′⟩−Wi, it follows that πWiZ∈⟨Δi−⟩ for i=2,3.
Hence
[TABLE]
The last inequality is true by definition of Wi and Mj.
Then by Lemma 1 the point πZ belong to the ϵ-neighbourhood of the intersection ⟨Δ3−⟩∩⟨Δ2−⟩.
Since Z∈⟨Δ1′⟩, it follows that πZ∈⟨πΔ1′⟩.
Hence ∣⟨πΔ1′⟩,⟨Δ3−⟩∩⟨Δ2−⟩∣<ϵ.
Then by definition of ϵ a convex hulls ⟨πΔ1′⟩,⟨Δ2−⟩,⟨Δ3−⟩ have a common point.
Therefore sets πΔ1′,Δ2−,Δ3− form a required partition.
Список литературы
- [BBZ]
I. Barany, P. V. M. Blagojevic and G. M. Ziegler., Tverberg’s Theorem at 50: Extensions and Counterexamples, Notices of the AMS, 63:7 (2016), 732-739. http://www.ams.org/journals/notices/201607
- [BFZ14]
P. V. M. Blagojevic, F. Frick, and G. M. Ziegler, Tverberg plus constraints, Bull. Lond. Math. Soc. 46:5 (2014), 953-967, arXiv: 1401.0690.
- [BZ]
P. V. M. Blagojevic and G. M. Ziegler., Beyond the Borsuk-Ulam theorem: The topological Tverberg story, arXiv:1605.07321v2
- [Fr15]
F. Frick, Counterexamples to the topological Tverberg conjecture,
arXiv:1502.00947.
- [Gr10]
M. Gromov, Singularities, expanders, and topology of maps. Part 2: from combinatoricsto topology via algebraic isoperimetry., Geometric and Functional Analysis, 20:2 (2010), 445–446.
- [Sk17]
A. Skopenkov, On van Kampen-Flores, Conway-Gordon-Sachs and Radon theorems, published as §4
Russian Math. Surveys, 73:2 (2018), 323-353
arxiv:1704.00300
- [Sk18]
A. Skopenkov, A user’s guide to the topological Tverberg
conjecture, Russian Math. Surveys, 73:2(2018) 323-353,
arxiv:1605.05141.
- [Tv66]
H. Tverberg (1966), A Generalization of Radon’s Theorem. Journal of the London Mathematical Society, s1-41: 123-128. doi:10.1112/jlms/s1-41.1.123