Hadamard Matrices: Difference between revisions

From tehowiki
Jump to navigation Jump to search
m Gfis moved page Hadamard matrices to Hadamard Matrices: Misspelled title
No edit summary
Line 1: Line 1:
In Ian Stewart's book "Die letzten Rätsel der Mathematik" (ISBN 978-3499616945), at the end there is a section about Hadamard matrices. That inspired me to develop a 3D model for the first 12 matrices (of edge length 2, 4, 8, 12, ... 44).  
In the last chapter of Ian Stewart's book "The Great Mathematical Problems" (ISBN 978-1846683374), there is a section about [[wikipedia:Hadamard_matrix|Hadamard matrices]]. These are square arrays whose entries are either 1 or 0 (Hadamard's original definition used −1 of 0). Each row in the matrix, when compared with some different row, must have exactly one half of the entries matching and one half non-matching. Apart from the 2 x 2 matrix, the edge lengths of all Hadamard matrices must be divisible by 4.  


===Raw data===
<code>1 1        1 1 1 1</code>
 
<code>1 0        1 0 1 0</code>
 
<code>1 1 0 0</code>
 
<code>1 0 0 1</code> 
 
That inspired me to develop a 3D model for the 11 matrices with  4, 8, 12, ... 44 = 4*n for n = 1..11.
 
The model has the s
 
==== Technical details ====
The article https://arxiv.org/html/2411.18897v1 describes in detail how to compute the matrices with SageMath. Online you can run https://sagecell.sagemath.org/ with the following program:
The article https://arxiv.org/html/2411.18897v1 describes in detail how to compute the matrices with SageMath. Online you can run https://sagecell.sagemath.org/ with the following program:



Revision as of 07:40, 9 August 2025

In the last chapter of Ian Stewart's book "The Great Mathematical Problems" (ISBN 978-1846683374), there is a section about Hadamard matrices. These are square arrays whose entries are either 1 or 0 (Hadamard's original definition used −1 of 0). Each row in the matrix, when compared with some different row, must have exactly one half of the entries matching and one half non-matching. Apart from the 2 x 2 matrix, the edge lengths of all Hadamard matrices must be divisible by 4.

1 1 1 1 1 1

1 0 1 0 1 0

1 1 0 0

1 0 0 1

That inspired me to develop a 3D model for the 11 matrices with 4, 8, 12, ... 44 = 4*n for n = 1..11.

The model has the s

Technical details

The article https://arxiv.org/html/2411.18897v1 describes in detail how to compute the matrices with SageMath. Online you can run https://sagecell.sagemath.org/ with the following program:

from sage.combinat.matrices.hadamard_matrix import hadamard_matrix, skew_hadamard_matrix

for i in range(1,12):
  H = hadamard_matrix(4*i)
  print(i, "========")
  print(H.str())

The output can be reduced with:

perl -pe "s{\-1}{\.}g; s{[\[\]\|\+\-]}{}g; s{\A\s+\Z}{};"  infile > outfile

3D Design

Since the edge lengths increase by 4, solid blocks of bxwxh = 1x1x2 will be used to represent a one in a matrix plane, and empty space for a zero (resp. -1). The matrix planes will be stacked as a pyramid, with the 44x44 matrix at the bottom and the 4x4 at the top. Maybe the 2x2 matrix becomes the real top, maybe with blocks 1x1x1?