how to calculate rank of matrix

Understanding the rank of a matrix is a fundamental concept in linear algebra with wide-ranging applications in fields like engineering, computer science, statistics, and economics. It provides crucial insights into the properties of a matrix, including the solvability of systems of linear equations, the dimensionality of vector spaces, and the invertibility of square matrices.

This article will guide you through what matrix rank is, why it's important, and the primary methods for calculating it, focusing on the robust Gaussian elimination method.

What is the Rank of a Matrix?

The rank of a matrix can be defined in several equivalent ways:

• Row Rank: The maximum number of linearly independent row vectors in the matrix.
• Column Rank: The maximum number of linearly independent column vectors in the matrix.
• Dimension of the Column Space: The dimension of the vector space spanned by the columns of the matrix.
• Dimension of the Row Space: The dimension of the vector space spanned by the rows of the matrix.

Crucially, the row rank is always equal to the column rank for any given matrix. This common value is simply referred to as the "rank" of the matrix, often denoted as rank(A) or ρ(A).

Key Properties of Matrix Rank:

• The rank of an m x n matrix cannot exceed the number of rows (m) or the number of columns (n). So, rank(A) ≤ min(m, n).
• A matrix has full rank if its rank equals min(m, n).
• A square matrix n x n is invertible if and only if its rank is n (full rank).
• The rank of a zero matrix is zero.

Methods to Calculate Matrix Rank

1. Gaussian Elimination (Row Echelon Form)

This is the most general and robust method for finding the rank of any matrix. It involves transforming the matrix into its row echelon form (or reduced row echelon form) using elementary row operations. The rank is then simply the number of non-zero rows in the row echelon form.

Steps for Gaussian Elimination:

1. Choose a Pivot: Start with the first column. Find the first non-zero element from the top. This is your pivot element. If the element in the first row, first column is zero, swap that row with another row below it that has a non-zero element in the first column. If the entire column is zero, move to the next column.
2. Make Elements Below Pivot Zero: Use elementary row operations (multiplying a row by a scalar and adding/subtracting it from another row) to make all elements directly below the pivot zero.
3. Move to the Next Row/Column: Once the first column (or current pivot column) is processed, move to the next row and the next column to find the next pivot. Repeat the process, ensuring that each new pivot is to the right of the previous one.
4. Count Non-Zero Rows: Once the matrix is in row echelon form (all entries below the leading entry of each non-zero row are zero), count the number of non-zero rows. This count is the rank of the matrix. A non-zero row is any row that contains at least one non-zero element.

Example of Gaussian Elimination:

Let's find the rank of matrix A:

A = [ 1  2  3 ]
    [ 4  5  6 ]
    [ 7  8  9 ]

1. Step 1: First Pivot (1,1) is 1.

Make elements below it zero:

• R2 = R2 - 4*R1
• R3 = R3 - 7*R1

A ~ [ 1   2   3  ]
    [ 0  -3  -6  ]
    [ 0  -6  -12 ]
                            

2. Step 2: Second Pivot (2,2) is -3.

Make element below it zero:

• R3 = R3 - 2*R2

A ~ [ 1   2   3 ]
    [ 0  -3  -6 ]
    [ 0   0   0 ]
                            

3. Step 3: Count Non-Zero Rows.

The resulting matrix has two non-zero rows (the first and second rows). The third row is all zeros.

Therefore, rank(A) = 2.

2. Determinant Method (For Smaller Matrices)

The rank of a matrix can also be found using determinants, especially useful for smaller matrices. The rank of a matrix A is the size of the largest square submatrix of A that has a non-zero determinant.

Steps for Determinant Method:

1. Start with the largest possible square submatrix (min(m, n) x min(m, n)). Calculate its determinant.
2. If the determinant is non-zero, then the rank is min(m, n).
3. If the determinant is zero, examine all square submatrices of the next smaller size ((min(m, n)-1) x (min(m, n)-1)). If at least one of these has a non-zero determinant, the rank is min(m, n)-1.
4. Continue this process until you find a square submatrix with a non-zero determinant. The size of this submatrix is the rank. If all 1x1 submatrices (individual elements) are zero, the rank is 0 (for a zero matrix).

Example of Determinant Method:

Consider the same matrix A:

A = [ 1  2  3 ]
    [ 4  5  6 ]
    [ 7  8  9 ]

1. This is a 3x3 matrix. Calculate its determinant:

det(A) = 1(5*9 - 6*8) - 2(4*9 - 6*7) + 3(4*8 - 5*7)

det(A) = 1(45 - 48) - 2(36 - 42) + 3(32 - 35)

det(A) = 1(-3) - 2(-6) + 3(-3)

det(A) = -3 + 12 - 9 = 0

2. Since det(A) = 0, the rank is not 3. We look for a 2x2 submatrix with a non-zero determinant.
3. Consider the top-left 2x2 submatrix:

M = [ 1  2 ]
    [ 4  5 ]
                            

det(M) = (1*5) - (2*4) = 5 - 8 = -3

4. Since det(M) = -3 ≠ 0, the largest submatrix with a non-zero determinant is 2x2.

Therefore, rank(A) = 2.

While effective for small matrices, the determinant method becomes computationally intensive for larger matrices as the number of submatrices and determinant calculations grows rapidly.

Why is Matrix Rank Important?

• Systems of Linear Equations: The rank of a matrix is crucial for determining the number of solutions to a system of linear equations Ax = b.
  • If rank(A) = rank([A|b]) = n (number of variables), there is a unique solution.
  • If rank(A) = rank([A|b]) < n, there are infinitely many solutions.
  • If rank(A) < rank([A|b]), there are no solutions.
• Invertibility: A square matrix is invertible if and only if its rank is equal to its dimension (full rank).
• Vector Spaces: The rank of a matrix tells us the dimension of the column space (image) and the row space of the linear transformation represented by the matrix.
• Data Compression and PCA: In data science, techniques like Principal Component Analysis (PCA) use concepts related to rank to reduce the dimensionality of data while preserving most of its variance.
• Linear Transformations: The rank of a transformation matrix indicates the dimension of the output space (image) of the transformation.

Conclusion

The rank of a matrix is a powerful concept that quantifies the "effective" dimensionality of the linear transformation it represents. While the determinant method offers a quick check for smaller matrices, Gaussian elimination is the go-to method for its general applicability and computational efficiency, especially for larger matrices. Mastering matrix rank is essential for anyone delving deeper into linear algebra and its myriad applications.