Take the 2-minute tour ×
Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. It's 100% free, no registration required.

Let $C$ be a three-dimensional tensor of dimensions $n\times n\times n$. Define: $$[C(x,y)]_k=\sum_{i,j}C_{ijk}x_iy_j,$$ i.e. $C(x,y)$ is a vector of dimension $n$.

Is there a way to bound the norm: $$||C(x,y)||_2$$ such that the bound depends on some norm of $C$ and some norms of $x$ and $y$? (something like a generalized Cauchy-Schwarz).

What about a similar bound on $||C(x,y)-C(x',y')||_2$ (something that will depend on $||C||$ and say $||x-x'||$ and $||y-y'||$)?

Thanks.

share|improve this question

1 Answer 1

up vote 0 down vote accepted

We have $$[C(x,y)]_k^2\leq \left(\sum_{i,j}C_{ijk}^2\right)\left(\sum_{i,j}x_i^2y_j^2\right)=\lVert x\rVert_2^2\cdot\lVert y\rVert_2^2\sum_{i,j}C_{ijk}^2$$ so putting $\lVert C\rVert_2^2:=\sum_{i,j,k}C_{ijk}^2$ we get $\lVert C(x,y)\rVert\leq\lVert C\rVert_2\cdot \lVert x\rVert_2\cdot \lVert y\rVert_2 $. For a fix $x$, the map $y\mapsto C(x,y)$ is linear hence \begin{align} \lVert C(x,y)-C(x',y')\rVert_2&\leq\lVert C(x,y-y')\rVert_2+\lVert C(x-x',y'\rVert_2\\ &\leq \lVert C\rVert_2\lVert x\rVert_2\lVert y-y'\rVert_2+\lVert C\rVert_2\lVert x-x'\rVert_2\lVert y'\rVert_2\\ &\leq \lVert C\rVert_2\max\{\lVert x\rVert_2,\lVert y'\rVert_2\} \left(\lVert x-x'\rVert_2+\lVert y-y'\rVert_2\right). \end{align}

share|improve this answer
    
thanks! exactly what I was looking for. –  kloop May 26 '12 at 20:41

Your Answer

 
discard

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.