Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

I know quite a few identities about quadratic forms of random vectors, but I'm having difficulty coaxing something out of this quadratic form of random matrices. Suppose I know $\mathbb{E}[W W^{T}]$ and $\mathbb{E}[W] = 0$. Then can I deduce a closed for form $\mathbb{E}[W^{T} P W]$ for a non-random matrix $P$?

EDIT Changed given from $\mathbb{E}[W^{T} W]$ to $\mathbb{E}[W W^{T}]$, added $\mathbb{E}[W] = 0$.

share|cite|improve this question
Do you know the distribution of $W$ as well, or just the variance $\mathbf{E}[W^TW]$? – Srivatsan Nov 24 '11 at 2:27
All I know is that E[W] = 0 and E[W * W'] = Sigma_{w}, not the distribution itself. – duckworthd Nov 24 '11 at 5:56
If you're going to fundamentally change the question by some minute changes that are easily overlooked, it would make sense to point out the change. – joriki Nov 24 '11 at 6:09
@Sasha: Why did you delete your answer? It seemed to solve the problem. – joriki Nov 24 '11 at 7:27
I entered an edit comment stating my changes, but apparently they weren't picked up. – duckworthd Nov 24 '11 at 9:42
up vote 1 down vote accepted

Quite simply, you can't say much -- at least not in the sense that you can decompose it. Consider the $(i,j)$ index of $\mathbb{E}[W^{T} P W]$.

$$ \mathbb{E}[W^{T} P W](i,j) = \sum_{k1,k2}P(k1,k2) W(k1, i), W(k2, j) $$

While on the other hand,

$$ \mathbb{E}[W W^{T}](i,j) = \sum_{k} W(i,k) W(j,k) $$

To say anything about the former, what you really need is the terms of the latter. Were $W$ a vector, the covariance matrix $\mathbb{E}[W W^{T}]$ would have just that, but as it stands you can only retrieve linear combinations of those terms without a way to decompose them.

share|cite|improve this answer

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

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