Correction of paper published in J. Combinatorial Theory 21, 1976: On the Existence of Hadamard Matrices
Jonas Lindstr{\o}m Jensen

TL;DR
This paper revisits a 1976 study on Hadamard matrices, clarifying the actual bounds for their existence and correcting an error in the original proof.
Contribution
It identifies and corrects an error in the original proof regarding the bounds for constructing Hadamard matrices.
Findings
The original proof's bound was not valid.
A corrected bound for Hadamard matrix construction is provided.
The paper clarifies the conditions for the existence of Hadamard matrices.
Abstract
In the paper On the Existence of Hadamard Matrices in J. Combinatorial Theory 21, 1976, it is shown that for a natural number q > 3, we can construct an Hadamard Matrix of order 2^s q for s \geq t where t = [2 log_2(q-3)]. I will show that this bound is not a consequence of the proof given in the paper and explain the error in the argumentation.
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
Topicsgraph theory and CDMA systems Β· Graph Labeling and Dimension Problems Β· Mathematics and Applications
Correction of paper published in J. Combinatorial Theory 21, 1976: On the Existence of Hadamard Matrices
Jonas LindstrΓΈm Jensen
(Department of Mathematics, University of Aarhus, Denmark)
1 Introduction
An Hadamard Matrix is a square matrix with entries , eg. a -matrix is an Hadamard Matrix if
[TABLE]
It is shown that for , it is a necessary condition that , and it is conjectured that it is also a sufficient condition. For a natural number it is proved in [1] that there exists Hadamard Matrices of order for , where . In this paper I will show that the bound on is wrong. It is true that there is some to make the claim true, but the bound on is wrong.
2 Structure of the proof
In [1] a corollary to a two dimensional version of Frobenius Coin Problem is used. The theorem states that for relatively prime integers , any integer can be written in the form for some nonnegative integers . The corollary states that given and where is odd, there exist nonnegative integers such that
[TABLE]
for some . The proof of Corollary 7 goes like this (I have filled in some details that is left out in [1]).
Let , then , and hence for some . Let
[TABLE]
and let be the smallest power of 2 greater than N. By the theorem we have nonnegative such that
[TABLE]
and since we have
[TABLE]
where .
In Lemma 9 this result is used to show that there exist nonnegative such that to construct an Hadamard Matrix of order . This only holds for being a prime , and in that case in the proof of the corollary and hence . A similar proof is done for where and . Since the Kronecker product of two Hadamard Matrices, the prime factorization of can be used to construct an Hadamard Matrix of order where for a sufficiently large .
3 The error
The error comes when trying to estimate how big has to be. In [1] it is stated that is enough for each prime factor, and since
[TABLE]
that bound is preserved under multiplication, and hence when used on all prime factors.
But this is not the case, since for has and choose such that
[TABLE]
which implies
[TABLE]
So if we choose , we can ensure (2) to be true. But as we stated earlier, this gives us relation for , and Lemma 9 uses this relation to give us an Hadamard Matrix of order , which is double the bound stated in [1]. With the new bound , it is not possible to make an estimate like (1), and the estimate cannot be extended from the prime factors to the product .
4 A numerical example
To make things a bit clearer, I provide a numerical example. Let . It is claimed that Lemma 9 gives an Hadamard Matrix of order where . So in our case . In the proof of Corollary 7, we choose such that is greater than
[TABLE]
hence . The theorem gives us a relation
[TABLE]
and by multiplication with 2 we get
[TABLE]
Lemma 9 in [1] uses this to provide us with an Hadamard Matrix of order , but the bound on the exponent given in [1] is .
The reference list from the paper itself. Each links out to its DOI / PubMed record.
- 1[1] J. Seberry Wallis (1976), On the Existence of Hadamard Matrices, J. Comb. Theory (A) 21 , 188-195
