Introduction
Prerequisites
It is assumed that those reading this have a basic understanding of what a matrix is and how to add them, and multiply them by scalars, i.e. plain old numbers like 3, or -5. A secondary school algebra course would probably give one more than enough background, but is surely not required by any means.
If you need some background Go here
Matrix Multiplication Basics
In order to multiply 2 matrices given one must have the same amount of rows that the other has columns. In other words two matrices can be multiplied only if one is of dimension m×n and the other is of dimension n×p where m, n, and p are natural numbers {m,n,p }. The resulting matrix will be of dimension m×p.
Example: 2x4 Multiplied by 4x2
Take a matrix 2x4 call it ; another of dimension 4x2 and call it . Give them some arbitrary values and let's do some multiplication.
Our result matrix is going to be 2×2 as you will see as we go step by step. First we multiply and sum the first row with the first column: 2(3) + 3(2) + (-1)(-1) + 0(2). Then we do the same for the first row and second column: 2(4) + 3(1) + (-1)(2)+ 0(7), etc.
To make a long story short, our matrix would be:
As this implies, multiplication is non-commutative in general for matrices, i.e. since in this case if we reversed the order the resulting matrix would be 4×4 instead of 2×2.
More General Approach
Let's visualize A and B as m×n and n×p matrices respectively.
We are going to be adding and multiplying like before, but generally.
General Algorithms
So now that we have a general idea of what a matrix is, and how to multiply them in general, we can derive some pseudocode around it. We could break down the steps as follows.
- Check the sizes of two matrices (m×n) and (t×u): if n = t then we can multiply them otherwise no (in that order )
- If they can be multiplied, then create a new matrix of size m by u
- For each row in A and each column in multiply and sum the elements and the place the results in the rows and columns of the result matrix
Pseudocode
Here is some pseudocode treating matrices like if they have a m element and an n element, so the dimension of a matrix object is m×n.
multiplyMatrix(matrix1, matrix2)
-- Multiplies rows and columns and sums them
multiplyRowAndColumn(row, column) returns number
var
total: number
begin
for each rval in row and cval in column
begin
total += rval*cval
end
return total
end
begin
-- If the rows don't match up then the function fails
if matrix1:n != matrix2:m return failure;
dim = matrix1:n -- Could also be matrix2:m
newmat = new squarematrix(dim) -- Create a new dim x dim matrix
for each r in matrix1:rows and c in matrix2:columns
begin
end
end
Implementations
C Sharp
// Program in C# to multiply two matrices using Rectangular arrays.
using System;
class.ToString
{
int[,] a;
int[,] b;
int[,] c;
public void ReadMatrix()
{
Console.WriteLine("\n Size of Matrix 1:");
Console.Write("\n Enter the number of rows in Matrix 1 :");
int m=int.Parse(Console.ReadLine());
Console.Write("\n Enter the number of columns in Matrix 1 :");
int n=int.Parse(Console.ReadLine());
a=new int[m,n];
Console.WriteLine("\to.doublepenetration the elements of Matrix 1:");
for(int i=0;i<a.GetLength(0);i++)
{
for(int j=0;j<a.GetLength(1);j++)
{
a[i,j]=int.Parse(Console.ReadLine());
}
}
Console.WriteLine("\n Size of Matrix 2 :");
Console.Write("\n Enter the number of rows in Matrix 2 :");
m=int.Parse(Console.ReadLine());
Console.Write("\n Enter the number of columns in Matrix 2 :");
n=int.Parse(Console.ReadLine());
b=new int[m,n];
Console.WriteLine("\n Enter the elements of Matrix 2:");
for(int i=0;i<b.GetLength(0);i++)
{
for(int j=0;j<b.GetLength(1);j++)
{
b[i,j]=int.Parse(Console.ReadLine());
}
}
}
public void PrintMatrix()
{
Console.WriteLine("\n Matrix 1:");
for(int i=0;i<a.GetLength(0);i++)
{
for(int j=0;j<a.GetLength(1);j++)
{
Console.Write("\t"+a[i,j]);
}
Console.WriteLine();
}
Console.WriteLine("\n Matrix 2:");
for(int i=0;i<b.GetLength(0);i++)
{
for(int j=0;j<b.GetLength(1);j++)
{
Console.Write("\t"+b[i,j]);
}
Console.WriteLine();
}
Console.WriteLine("\n Resultant Matrix after multiplying Matrix 1 & Matrix 2:");
for(int i=0;i<c.GetLength(0);i++)
{
for(int j=0;j<c.GetLength(1);j++)
{
Console.Write("\t"+c[i,j]);
}
Console.WriteLine();
}
}
public void MultiplyMatrix()
{
if(a.GetLength(1)==b.GetLength(0))
{
c=new int[a.GetLength(0),b.GetLength(1)];
for(int i=0;i<c.GetLength(0);i++)
{
for(int j=0;j<c.GetLength(1);j++)
{
c[i,j]=0;
for(int k=0;k<a.GetLength(1);k++) // OR k<b.GetLength(0)
c[i,j]=c[i,j]+a[i,k]*b[k,j];
}
}
}
else
{
Console.WriteLine("\n Number of columns in Matrix1 is not equal to Number of rows in Matrix2.");
Console.WriteLine("\n Therefore Multiplication of Matrix1 with Matrix2 is not possible");
Environment.Exit(-1);
}
}
}
class Matrices
{
public static void Main()
{
MatrixMultiplication MM=new MatrixMultiplication();
MM.ReadMatrix();
MM.MultiplyMatrix();
MM.PrintMatrix();
}
}
VB.NET
Public Module MatrixMultiplication
' Generic stand-alone method
' Uses LINQ.
Public Function Multiply(matrix1 As IEnumerable(Of IEnumerable(Of Double)), _
matrix2 As IEnumerable(Of IEnumerable(Of Double))
As IEnumerable(Of IEnumerable(Of Double))
If matrix1.Count = matrix2(0).Count Then
Dim ret(matrix1(0).Count, matrix2.Count) As Double
For i As Integer = 0 To ret.Count - 1
For j As Integer = 0 To ret(0).Count - 1
ret(i, j) = 0
For k As Integer = 0 To matrix1(0).Count - 1
ret(i, j) = ret(i, j) + a(i, k) + b(k, j)
Next
Next
Next
Else
New RankException("Number of columns in matrix1 is not equal to the number of rows in matrix2.")
End If
End Function
' This method can be run in a console app.
Public Sub Main()
Dim m_1 As Integer, n_1 As Integer, m_2 As Integer, n_2 As Integer
Try
Console.WriteLine("Size of matrix 1:")
Console.Write("Number of rows: ")
m_1 = Integer.Parse(Console.ReadLine())
Console.Write("Number of columns: ")
n_1 = Integer.Parse(Console.ReadLine())
Dim matrix1(m_1, n_1) As Double
Console.WriteLine("Enter the elements of matrix 1:")
For i As Integer = 0 To m_1 - 1
For j As Integer = 0 To n_1 - 1
Console.Write("Element at " & i & ", " & j)
matrix1(i, j) = Integer.Parse(Console.ReadLine())
Next
Next
Console.WriteLine("Size of matrix 2:")
Console.Write("Number of rows: ")
m_2 = Integer.Parse(Console.ReadLine())
Console.Write("Number of columns: ")
n_2 = Integer.Parse(Console.ReadLine())
Console.WriteLine("Enter the elements of matrix 2:")
For i As Integer = 0 To m_2 - 1
For j As Integer = 0 To n_2 - 1
Console.Write("Element at " & i & ", " & j)
matrix2(i, j) = Integer.Parse(Console.ReadLine())
Next
Next
Dim product(,) As Double = Multiply(matrix1, matrix2)
Console.WriteLine("The product of the two matrices is:")
For i As Integer = 0 To product.Count - 1
For j As Integer = 0 To product(0).Count - 1
Console.Write(product(i, j) & " ")
Next
Console.WriteLine()
Next
Catch ex As Exception
Console.WriteLine("An error occured. The message associated with the error was:")
Console.WriteLine(ex.Message)
End Try
Console.WriteLine()
Console.WriteLine("Press any key to exit.")
Console.ReadKey()
End Sub
End Module
Python
def mat(a,b):
m=len(a)
n=len(a[0])
k=len(b)
res=[]
if len(b) != 1:
p=len(b[0])-1
else:
p=0
if len(b)==0:
print('bad size')
elif n!=k:
print('bad size')
else:
#print('good size')
n=k
#print('k=',k,'n=',n)
for q in range(m):
res.append([0])
#print('res=',res)
for q in range(m):
for w in range(p):
res[q].append(0)
for i in range(m):
for j in range(p+1):
for r in range(n):
#print('ijr:res',i,j,r,res)
res[i][j] =a[i][r]*b[r][j]+res[i][j]
return res
+----------------+
| AN ALTERNATIVE |
+----------------+
...
s = 0
for i in range(self.rows):
for c in range(other.cols):
s = 0
for j in range(other.rows):
s += self.mat[i][j] * other.mat[j][c]
sum.mat[i][c] = s
...
Ruby
require 'matrix'
def read_matrix
puts "Enter number of column for matrix: "
_cols = gets.chomp!.to_i
puts "Enter number of rows for matrix: "
_rows = gets.chomp!.to_i
raise "Invalid sizes" unless _rows.is_a? Fixnum && _cols.is_a? Fixnum
puts "Enter the elements of the matrix (one per line)"
a = Matrix.build _rows, _cols, do |m|
m = gets.chomp!.to_i
end
raise "An element is not a number" if a.any? { |eL| !eL.is_a? Fixnum }
a
end
puts "reading first matrix..."
mat1 = read_matrix
puts "reading second matrix..."
mat2 = read_matrix
begin
puts mat1 * mat2
rescue ExceptionForMatrix::ErrDimensionMismatch => e
puts e.message
end
# or if we don't care about user input
# following the C++ example ..to show you how concise Ruby is
Matrix.build(4) { rand } * Matrix.build(4) { rand }
Java
public class SimpleMatrix {
private Double data[][];
private Boolean transposed = false;
public SimpleMatrix(int i, int j) {
data = new Double[i][j];
}
public SimpleMatrix product(SimpleMatrix m) {
if(this.getColumns() != m.getRows()) return null;
SimpleMatrix outcome = new SimpleMatrix(this.getRows(),m.getColumns());
Double sumProduct = 0.0;
for(int i = 0 ; i < getRows(); i++)
{
for(int j = 0; j < m.getColumns(); j++)
{
sumProduct=0.0;
for(int k=0;k<getColumns();k++)
{
sumProduct+=this.get(i, k)*m.get(k,j);
}//end k
outcome.set(i, j, sumProduct);
}//end j
}//end i
return outcome;
}
@Override
protected Object clone() throws CloneNotSupportedException {
SimpleMatrix m = new SimpleMatrix(this.getRows(),this.getColumns());
m.transposed = this.transposed;
m.vectorSpace = this.vectorSpace;
for(int i = 0 ; i < getRows(); i++)
{
for(int j = 0; j < getColumns(); j++)
{
m.set(i, j, this.get(i, j));
}
}
return m;
}
public void transpose()
{
transposed = !transposed;
}
public Boolean isTransposed()
{
return transposed;
}
public int getRows() {
return transposed?data[0].length : data.length;
}
public int getColumns() {
return transposed? data.length : data[0].length;
}
public void set(int i, int j, Double t) {
data[i][j] = t;
}
public Double get(int i, int j) {
return isTransposed()?data[j][i]:data[i][j];
}
@Override
public String toString() {
StringBuffer outcome = new StringBuffer();
outcome.append("{[");
for(int i = 0 ; i < getRows(); i++)
{
for(int j = 0; j < getColumns(); j++)
{
outcome.append(get(i,j)).append(",");
}
outcome.deleteCharAt(outcome.length()-1).append("]\n[");
}
outcome.deleteCharAt(outcome.length()-1).deleteCharAt(outcome.length()-1).append("}");
return outcome.toString();
}
public static void main(String[] args) throws CloneNotSupportedException{
SimpleMatrix m = new SimpleMatrix(2,3);
m.set(0,0,1.0);
m.set(0,1,2.0);
m.set(0,2,3.0);
m.set(1,0,4.0);
m.set(1,1,5.0);
m.set(1,2,6.0);
System.out.println("Original:\n"+m);
SimpleMatrix m2 = (SimpleMatrix) m.clone();
m2.transpose();
System.out.println("Transpose:\n"+m2);
Matrix m3 = m.product(m2);
System.out.println("Product: \n"+m3);
}
}
C and C++
int main(void)
{
const int ROW=4, COL=4, INNER=4;
int A[ROW][INNER], int B[INNER][COL], int C[ROW][COL];
for (int row = 0; row != ROW; ++row)
{
for (int col = 0; col != COL; ++col)
{
int sum = 0;
for (int inner = 0; inner != INNER; ++inner)
{
sum += A[row][inner] * B[inner][col];
}
C[row][col] = sum;
}
}
return 0;
}
Clojure
(defn mulmat [a b transposer fuser]
(let [b-transposed (transposer b)]
(for [a-row a]
(fuser a-row b-transposed))))
(defn transpose [a]
(apply map list a))
(defn fuse-vector-vector [a b]
(reduce + (map * a b)))
(defn fuse-vector-matrix [v matrix]
(map (partial fuse-vector-vector v) matrix))
(println
(mulmat [[1 2]
[3 4]]
[[5 6]
[7 8]]
transpose fuse-vector-matrix))