Inner product of two vector series

54 Views Asked by At

In a proof to a problem in optimization, there's this part I don't quite understand.

$(\Sigma_{i=1}^m \alpha_i\textbf{d}_i)^TQ(\Sigma_{i=1}^m \alpha_i\textbf{d}_i) = \Sigma_{i=1}^m \alpha_i^2\textbf{d}_i^TQ\textbf{d}_i$

where $\boldsymbol{\alpha} = (\alpha_1, \alpha_2,\cdots, \alpha_m)^T \in \mathbb{R}^m$; and $\textbf{d}_1, \textbf{d}_2, \cdots, \textbf{d}_m \in \mathbb{R}^n$ are $Q$-orthogonal, $\textbf{d}_i^TQ\textbf{d}_j = 0$ whenever $i \neq j$.

I tried to expand the left hand side by inserting the $Q$ to the right like this

$(\Sigma_{i=1}^m \alpha_i\textbf{d}_i)^T(\Sigma_{i=1}^m \alpha_iQ\textbf{d}_i) $

I believe I need to use some rule involving inner products, but I'm not exactly sure what to do next.

Thanks in advance.

1

There are 1 best solutions below

0
On

You are almost there, you just need to develop the equation one step further. You have that: $$ \left(\sum_{i=1}^m\alpha_i\mathbf{d}_i\right)^\top Q\left(\sum_{i=1}^m\alpha_i\mathbf{d}_i\right) = \left(\sum_{i=1}^m\alpha_i\mathbf{d}_i\right)^\top\left(\sum_{i=1}^m\alpha_iQ\mathbf{d}_i\right) = \sum_{i=1}^m\sum_{j=1}^m\alpha_i\alpha_j\mathbf{d}_i^\top Q\mathbf{d}_j. $$ Now, since the $\mathbf{d}_i$ vectors are $Q$-orthogonal, all the terms such that $i\neq j$ are equal to zero, which leaves you with: $$ \left(\sum_{i=1}^m\alpha_i\mathbf{d}_i\right)^\top Q\left(\sum_{i=1}^m\alpha_i\mathbf{d}_i\right) = \sum_{i=1}^m\alpha_i^2\mathbf{d}_i^\top Q\mathbf{d}_i $$