Cross Validated is a question and answer site for people interested in statistics, machine learning, data analysis, data mining, and data visualization. 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

Whats a quick way to work the below (and problems similar) out?

For me to take this derivative it involves a lot of time and boring calculation, there has to be a better way.

This is taken from the solutions to Chris Bishop's Pattern Recognition and Machine learning:

from Pattern Recogition and Machine Learning

share|improve this question
up vote 13 down vote accepted

There is something called the Matrix Cookbook, which includes a lot of identities and matrix derivatives. So if we look at eq. (88) of the Matrix Cookbook,

$$\frac{\partial}{\partial A} (\mathbf{x} -\mathbf{A}\mathbf{s})^T\mathbf{W}(\mathbf{x} -\mathbf{A}\mathbf{s}) = -2\mathbf{W}(\mathbf{x}-\mathbf{A}\mathbf{s})\mathbf{s}^T$$

we see that this directly refers to your problem, if we assume $\Sigma^{-1}$ is a covariance matrix and therefore symmetric.

share|improve this answer
Thank you! That is very helpful :) – user3490171 Feb 25 at 10:02
I made a couple of formatting changes - you might like to know that you can start and terminate a piece of Latex with $$ ... $$ if you want an equation to be centre-aligned. – Silverfish Feb 25 at 11:25

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.