# Difference between revisions of "What is matrix rank and how do i calculate it?"

Line 1: | Line 1: | ||

− | The rank of a matrix <math>A</math> is the number of independent columns of <math>A</math>. A square matrix is full rank if all of its columns are independent. That is, a full rank matrix has no column vector <math>v_i</math> of <math>A</math> that can be expressed as a linear combination of the other column vectors <math>v_j \neq \Sigma_{i = 0, i\neq j}^{n} a_i v_i</math>. For example, if one column of <math>A</math> is equal to twice another one, then those two columns are linearly dependent (with a scaling factor 2) and thus the matrix would not be full rank. | + | The rank of a matrix <math>A</math> is the number of independent columns of <math>A</math>. A square matrix is full rank if all of its columns are independent. That is, a square full rank matrix has no column vector <math>v_i</math> of <math>A</math> that can be expressed as a linear combination of the other column vectors <math>v_j \neq \Sigma_{i = 0, i\neq j}^{n} a_i v_i</math>. For example, if one column of <math>A</math> is equal to twice another one, then those two columns are linearly dependent (with a scaling factor 2) and thus the matrix would not be full rank. |

− | A simple test for determining if a matrix is full rank is to calculate its [http://mathworld.wolfram.com/Determinant.html determinant]. If the determinant is zero, there are linearly dependent columns and the matrix is not full rank. Prof. John Doyle also mentioned during lecture that one can perform the [http://mathworld.wolfram.com/SingularValueDecomposition.html singular value decomposition] of a matrix, and if the lowest singular value | + | A simple test for determining if a square matrix is full rank is to calculate its [http://mathworld.wolfram.com/Determinant.html determinant]. If the determinant is zero, there are linearly dependent columns and the matrix is not full rank. Prof. John Doyle also mentioned during lecture that one can perform the [http://mathworld.wolfram.com/SingularValueDecomposition.html singular value decomposition] of a matrix, and if the lowest singular value |

is near or equal to zero the matrix is likely to be not full rank ("singular"). | is near or equal to zero the matrix is likely to be not full rank ("singular"). | ||

+ | |||

+ | For Single-input-single-output (SISO) systems, which are the focus of this course, the reachability matrix will always be square; more inputs make it wider (because the width <math>B</math> is equal to the number of inputs). In the case of non-square matrices, full rank means that the number of independent vectors is as large as possible. | ||

--[[User:Fuller|Sawyer Fuller]] 16:22, 29 October 2007 (PDT) | --[[User:Fuller|Sawyer Fuller]] 16:22, 29 October 2007 (PDT) | ||

+ | |||

+ | modified by [[User:Fuller|Sawyer Fuller]] 18:12, 3 November 2007 (PDT) to be more specific about square and non-square matrices | ||

+ | |||

[[Category: CDS 101/110 FAQ - Lecture 5-1]] | [[Category: CDS 101/110 FAQ - Lecture 5-1]] | ||

[[Category: CDS 101/110 FAQ - Lecture 5-1, Fall 2007]] | [[Category: CDS 101/110 FAQ - Lecture 5-1, Fall 2007]] |

## Revision as of 01:13, 4 November 2007

The rank of a matrix is the number of independent columns of . A square matrix is full rank if all of its columns are independent. That is, a square full rank matrix has no column vector of that can be expressed as a linear combination of the other column vectors . For example, if one column of is equal to twice another one, then those two columns are linearly dependent (with a scaling factor 2) and thus the matrix would not be full rank.

A simple test for determining if a square matrix is full rank is to calculate its determinant. If the determinant is zero, there are linearly dependent columns and the matrix is not full rank. Prof. John Doyle also mentioned during lecture that one can perform the singular value decomposition of a matrix, and if the lowest singular value
is near or equal to zero the matrix is likely to be not full rank ("singular").

For Single-input-single-output (SISO) systems, which are the focus of this course, the reachability matrix will always be square; more inputs make it wider (because the width is equal to the number of inputs). In the case of non-square matrices, full rank means that the number of independent vectors is as large as possible.

--Sawyer Fuller 16:22, 29 October 2007 (PDT)

modified by Sawyer Fuller 18:12, 3 November 2007 (PDT) to be more specific about square and non-square matrices