I came across the following expression when solving a maximisation problem. I have the following ingredients:

  • Matrices $\Omega, P \in \mathbb{R}^{n \times n}$
  • Vector $t \in \mathbb{R}^n$
  • Also let $\mathrm{diag}(x), x \in \mathbb{R}^n$ denote the diagonal $n \times n$ matrix whose diagonal entries are the components of $x$, $\mathrm{tr}$ be the trace of a square matrix and $\cdot$ denote usual matrix-algebra multiplication

Now, I have the following expression: $$X = \mathrm{tr} \left( \Omega \cdot \mathrm{diag}(t) \cdot P \cdot \mathrm{diag}(t) \right) $$

So it seems to me that the expression is quadratic in vector $t$, linear in $P$ and $\Omega$ - so I would expect (hope?) that it can be written as $$ X = t^T \cdot f(\Omega, P) \cdot t $$ for some function $$f : \mathbb{R}^{n \times n} \times \mathbb{R}^{n \times n} \rightarrow \mathbb{R}^{n \times n}$$ which is also hopefully is expressible as something simple (linear). Is this right? If so, how can I get there? Or perhaps there are some other simplifications to this effect?

Any help, including pointers, is highly appreciated.


The $ik$ element of the thing inside the trace is

$$ \omega_{ij} t_j P_{js} t_s $$ with a sum over $j$ and $s$. The trace is the sum of the $ii$ elements, hence

$$ \omega_{ij} t_j P_{ji} t_i $$ summed over $i$ and $j$, which is $$ \sum_i \sum_j t_j \omega_{ij}P_{ji} t_i $$ That has the form you required, with $U = f(\Omega, P)$ being a matrix whose $ij$ element is $$ u_{ij} = \omega_{ji}P_{ij}. $$

| cite | improve this answer | |
  • $\begingroup$ Awesome - thank you so much! $\endgroup$ – Magic May 15 '15 at 11:34

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.