Professional Documents
Culture Documents
A Generalized Matrix Multiplication Scheme Based On The Concept of Embedding Dimension and Associated Spacer Matrices
A Generalized Matrix Multiplication Scheme Based On The Concept of Embedding Dimension and Associated Spacer Matrices
ISSN No:-2456-2165
Abstract:- The article presents a Generalized Matrix max(a, b) denotes the maximum of the two inputs a
Multiplication scheme that allows Non-compatible
and b , a, b N
matrices to be multiplied using Spacer matrices, which are
unique for every possible pair of nonconforming a b denotes the absolute value of the difference of the
dimensions. The article presents an approach to define
positive integral powers of strictly rectangular complex two inputs a and b , a, b N
matrices and thereby define subspaces of the Complex The Generalized Matrix Multiplication operation is
Matrix space using these matrix powers to constitute the denoted by the symbol ‘ ’
pertinent spanning sets. The mathematical formalism is AT denotes the transpose of the Matrix A
presented and Illustrated with numerical examples.
I. INTRODUCTION
Keywords:- Matrix Multiplication, Generalized Matrix
Multiplication, Embedding Dimension, Spacer Matrix, The present article addresses the issue of Matrix
Components of the Spacer Matrix Multiplication pertaining to multiplication of Non-compatible
matrices by introducing a Multiplication scheme that
compensates for the non-compatibility by invoking the
Notations concept of an encapsulating higher dimension termed as
“Embedding Dimension” and defining “Spacer matrices”
M mn (C ) denotes the Complex Matrix space of order associated with each such possible embedding dimension. The
m by n spacer matrices are so defined that they are independent of the
matrix elements constituting the multiplication chain, and is
The Matrices Asm and Bnt are termed “ Non-
unique for given pair of nonconforming dimensions, also,
compatible ” w.r.t. ordinary matrix multiplication AB when there is compatibility these spacer matrices reduce to
when m n and for ordinary matrix multiplication BA Identity matrices of appropriate orders, therefore, this
when s t generalized multiplication scheme is reduced to ordinary
N denotes the set of all Natural numbers matrix multiplication scheme when there is compatibility
among all the matrices that constitute the multiplication chain.
r denotes the Embedding dimension
The presented scheme allows the embedding dimension to be
X mn denotes the Spacer Matrix of order m by n the same for a given pair of input dimensions regardless of
Pmr , Qrn are the component Matrices associated with their ordering in the ordered pair, and it also allows for
defining the powers (positive and integral) of a strictly
the Spacer Matrix X mn rectangular complex matrix Amn , thereby allowing formation
1 of Subspaces of the Complex matrix space M mn (C ) , using
1
the unit matrix of order m by n , the matrix Amn and its
m . , m 1 1 . . 1 1m defined powers as the corresponding spanning sets.
.
1
m1
u max(m, n) m n , v max(t , s) t s
The Spacer Matrix associated with dimension pair (m, n) is denoted by X mn , which is defined as follows:
X mn Pmu Qun , where the space Matrix components have the following expressions:
P I mm , when m n
1 0 . . 0 1m 1m . . 1
m
0 1 . . 0 1m 1m . . 1
1 m
P I mm ( ) m um . . . . . . . . . . , when m n
m mu
. . . . . . . . . .
0 0 . . 1 m 1m
1 . . 1
m
Q I nn , when m n
1 0 . . 0
0 1 . . 0
. . . . .
. . . . .
I nn 0
0 . . 1
Q 1 , when m n
( ) u n n 1n 1
n . . 1
n
n un 1n 1
1 . .
n n
. . . . .
. . . . .
1 n 1
n . . 1
n
The Spacer Matrix associated with dimension pair (t , s ) is denoted by Yt s , which is defined as follows:
Yts RtvWvs , where the space Matrix components have the following expressions:
R Itt , when t s
1
R Itt ( ) t v t , when t s
t tv
I s s
W 1 , when t s
( ) v s s
s v s
The Multiplication chain and the special case of chain of Identical Matrix units
General Scheme:
( A1 ) m1n1 ( A2 ) m2 n2 .... ( As 1 )ms1ns1 ( As )ms ns A1 ( X 1 )n1m2 A2 ( X 2 )n2 m3 ..... As 1 ( X s 1 )ns1ms As
Numerical Examples:
1 0 0
1 0
A1 (1)11 , A2 , A3 0 1 0 , Then, we have the following:
0 1 22 0
0 1 33
7 1 2
3 3 6 6 3
X1 , X2 1 7 2 , Therefore we have:
2 2 12
6 6 3 23
1 0 0
1 0
(1)11 0 1 0 2 2 2 13
0 1 22 0 0 1
33
r max(m, n) m n max(n, m) n m , m, n N , Therefore, if the Spacer Matrix associated with dimension
pair (m, n) be X mn , X mn Pmr Qrn
The Spacer Matrix associated with dimension pair (n, m) be Ynm , Ynm Q
T
nr P T r m
1 0 0 2
1 0
Hence, we have: 0 1 0
(1)11 2
0 0 1 0 1 22 2
33 31
Special case: Powers of a Strictly Rectangular Matrix
Therefore, the chain constituted by ‘s’ such Matrix units Amn , Amn M mn (C ) , has the following Expression:
Amn Amn .... Amn Amn X nm Amn .... Amn X nm Amn , Where X nm Pnr Qrm
Amn Amn Amn X nm Amn , Amn Amn Amn Amn X nm Amn X nm Amn ,
Amn Amn Amn Amn Amn Amn X nm Amn X nm Amn X nm Amn X nm Amn ,
Amn Amn Amn Amn Amn Amn Amn X nm Amn X nm Amn X nm Amn X nm Amn X nm Amn
And so on.
Numerical Examples:
1 0 0
1) A , we have : m 2, n 3, r 4
0 1 0
1 1 0
7 1
1 0 0
3 0
6 6
1
0
1
1 7
P34 1 0 , Q42 , X 32
3 1 1 6 6
0 1 2 2 2 2
0 1 1
3 1 3 3
2 2
7 1 25 7
0 18 0
6 6 18
A A , A A A ,
1 7
0 7 25
0
6 6 23 18 18 23
1.685185 0.685185 0
A A A A ... (Up to 6 decimal places)
0.685185 1.685185 0 23
2.080247 1.080247 0
A A A A A ... (Up to 6 decimal places)
1.080247 2.080247 0 23
1 1 1
B , we have : m 2, n 3, r 4
1
2)
1 1
4 4 4 16 16 16 64 64 64
B B
4 16 16 64 64
, B B B , B B B B ,
4 4 16 64
( s 1)
We have: B B .... B(' s ' times) 4 B , where s 1, 40 1
i 0 0
3) C , we have : m 2, n 3, r 4
0 i 0
7 1 4i
6 6
0 3 0 0
C C , C C C ,
1
7
0 0
4i
0
6 6 3
1.555556 0.222222 0
C C C C .... (Up to 6 decimal places)
0.222222 1.555556 0
A M mn (C ), A 0mn , m n
V (3) span( m n , Amn , A A, A A A) span( m n , Amn , Amn X nm Amn , Amn X nm Amn X nm Amn ) and so
on.
Numerical Example:
1 0 0
A , we have the following associated results:
0 1 0
1 1 1
V (0) span( ), dim.(V (0)) 1
1 1 1
1 1 1 1 0 0
V (1) span( ), dim.(V (1)) 2
1 0 0
,
1 1 1
7 1
0
1 1 1 1 0 0 6 6
V (2) span( , ), dim.(V (2)) 3
1 0 0
,
1 1 1 1 7
0
6 6
7 1 25 7
0 0
1 1 1 1 0 0 6 6 18 18
V (3) span( , , ), dim.(V (3)) 3
1 0 0
,
1 1 1 1 7
0
7 25
0
6 6 18 18
s max(m, n) m n max(n, m) n m
Invariance under the Generalized Multiplication operation for Matrix powers in the case of nonconforming dimension pairs
1
Emn ( ) m n , where r max(m, n) m n , m n
r
Then, we have the following: