Wiki
Wiki

Research notes on every problem, and a library of the papers behind them. Built from the open erdos repository.

Updated


Source. M. Grinsztajn, A 63-dimensional counterexample to Borsuk's conjecture, unpublished note, May 2026, as described on the source card. Section 4 starts on p. 3; Lemma 3 is on p. 3 and its proof ends on p. 4.

Statement

With B1,B2,B3,CB_1,B_2,B_3,C as in Lemma 1 and xvx_v as in Lemma 2, put Si=∑y∈BixyS_i=\sum_{y\in B_i}x_y for i=1,2,3i=1,2,3. Lemma 3 (p. 3) states that the points xcx_c, c∈Cc\in C, lie in a common 63-dimensional subspace of the standard R65\mathbb R^{65} representation.

The proof identifies the subspace as W=span⁡(S1,S2,S3)⊥W=\operatorname{span}(S_1,S_2,S_3)^{\perp} and records along the way that Si⋅Si=12288S_i\cdot S_i=12288, Si⋅Sj=−6144S_i\cdot S_j=-6144 for i≠ji\ne j, and dim⁡span⁡(S1,S2,S3)=2\dim\operatorname{span}(S_1,S_2,S_3)=2. The note identifies WW with R63\mathbb R^{63} from then on (p. 4).

Proof pointer

pp. 3--4: each c∈Cc\in C has 8 neighbors and 24 non-neighbors in each BiB_i, so xc⋅Si=0x_c\cdot S_i=0. The degree data inside and between the blocks give the Gram matrix of S1,S2,S3S_1,S_2,S_3, whose eigenvalues 00, 1843218432, 1843218432 show that the span is two-dimensional.

Dependencies and read depth

Depends on Lemma 1, items 3 and 4, and Lemma 2. Read depth: claims checked; the statement and proof were read on pp. 3--4.

Bears on. E0505: the reduction from dimension 65 to 63 in the note's claim (Theorem 1).