Let $x_1, x_2, x_3, x_4, y_1, y_2, y_3$ and $y_4$ be fixed real numbers, not all of them equal to zero. Define a $4 \times 4$ matrix A by

$$\left( \begin{array}{ccc} x_1^2+y_1^2 & x_1x_2+y_1y_2& x_1x_3 +y_1y_3& x_1x_4 +y_1y_4\\ x_2x_1+y_2y_1 & x_2^2+y_2^2 & x_2x_3 +y_2y_3 & x_2x_4 +y_2y_4 \\ x_3x_1+y_3y_1 & x_3x_2 +y_3y_2 & x_3^2+y_3^2 & x_3x_4 +y_3y_4 \\ x_4x_1+y_4y_1 & x_4x_2 +y_4y_2 & x_4x_3 +y_4y_3 & x_4^2+y_4^2 \end{array} \right)$$. We need to calculate its rank.

In the worst case lets assume all except one variables are zero. Then exactly one entry of the matrix is non-zero, thus the rank is 1. Is this correct ?

  • $\begingroup$ If all except one are $0$, then you get a $0$-matrix with one non-zero diagonal entry, which will have deteminant $0$. $\endgroup$ – gt6989b May 7 '13 at 14:27

The rank of this matrix is at most two, since it is the sum of two rank 1 matrices:

$$ A = (x_1, x_2, x_3, x_4)^T (x_1, x_2, x_3, x_4) + (y_1, y_2, y_3, y_4)^T (y_1, y_2, y_3, y_4) $$

If the resulting sum is not the zero matrix, then its rank is at least 1. Sufficient conditions for the rank to be exactly 1 are for the matrix $A$ to be nonzero and for the vectors $(x_1, x_2, x_3, x_4)$ and $(y_1, y_2, y_3, y_4)$ to be linearly dependent.

Conversely, if the rank is at least 1, then $A$ is not the zero matrix. Also, for the rank of $A$ to be exactly 1, it is necessary for the vectors $(x_1, x_2, x_3, x_4)$ and $(y_1, y_2, y_3, y_4)$ to be linearly dependent.

Combining these observations gives us exact characterizations of when $A$ has rank respectively 0, 1, or 2. Among other things this confirms the OP's idea, that if exactly one variable is nonzero, then the rank of $A$ is 1.


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.