Is there a quick and easy way to show that a $n\times m$ Matrix (with $n<m$ ) has rank n ?

Is the determinant method in general the best way ? (I mean, if you find that the determinants of 2 $n\times n$ submatrizes, cannot be simultaneously 0.)

  • 1
    $\begingroup$ It normally is. Otherwise you could sometime (if lucky) spot some easy linear combination or a zero column or row. $\endgroup$ – user88595 Feb 5 '14 at 12:33
  • $\begingroup$ I'm not sure if I underastand the question correctly: Do you suppose your matrix to have full rank and want to show that it has rank $n$? Do you want to show that the matrix in general has at most rank $n$? Or do you want to test a given matrix on its rank? $\endgroup$ – Roland Feb 5 '14 at 12:48
  • $\begingroup$ It is easier to add columns one by one and see, whether the new column is in the span of previous columns. $\endgroup$ – user68061 Feb 5 '14 at 12:49
  • $\begingroup$ @Roland to show: $n\times m$-Matrix has rank min{n,m} $\endgroup$ – derivative Feb 5 '14 at 12:51
  • $\begingroup$ The most effective systematic way is the Gaussian elimination procedure. When the matrix is in reduced form you can compute the rank by counting pivots. $\endgroup$ – Giuseppe Negro Feb 5 '14 at 13:04

If $X$ is $n \times m$ with $n<m$ , then consider $Y = X X ^t$ which is $n \times n$. See that $Y$ is nonsingular iff $X$ has rank $n$. So, you only need to compute the determinant of $Y$.


You can use $${\rm rank\,}A = {\rm rank\,}A^t,$$

which means that the row rank of $A$ coincides with the row rank of $A^t$, i.e. the column rank of $A$. Thus, the rank of $A$ is the maximal number of linear independent rows of $A$, but also the maximal number of linear independent columns of $A$, which implies $${\rm rank\,}A \leq \min(n,m).$$

If $A$ is singular, equality will not be attained - consider the $n\times m$ matrix which is filled with zeros, which has rank $0$.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.