Effect of the nature of noise on the spectrum of a random matrix

57 Views Asked by At

Consider the following two equations

$X = M + \eta_1$

$Y = M + \eta_2$

where, $X\in\mathrm{R}^{n\times n}$, ia a real random matrix with mean $M\in\mathrm{R}^{n\times n}$. $\eta_1$ is Gaussian white noise with mean $0$ and covariance $\sigma^2I_{n\times n}$. while $Y\in\mathrm{C}^{n\times n}$, ia a complex random matrix with the same mean $M$. $\eta_2$ is complex Gaussian white noise with same mean $0$ and covariance $\sigma^2I_{n\times n}$. As such, $X$ and $Y$ have same mean but they differ in the nature of noise. Will the eigenvalues of the two matrix $X$ and $Y$ have same mean or will they differ?

1

There are 1 best solutions below

2
On

The sum of the eigenvalues (counted by multiplicity) is the trace, and that is linear, so the means of the sums of the eigenvalues of $X$ and $Y$ are both equal to the sum of the eigenvalues of $M$. Apart from that, I don't understand the question. Each matrix has $n$ eigenvalues, usually distinct and complex. There is no canonical way to single out a particular eigenvalue of $X$ or $Y$, so how can you ask whether they have the same mean?