In my finite element method book, there is a notation which is confusing me. Given $v:R^2\rightarrow R^2$, I'm supposed to evaluate

$\sigma\cdot \nabla v^T$
where $\sigma$ is a smooth tensor valued function. What is confusing me is the notation $\nabla v^T$. How do I interpret the gradient of a vector? Is it a matrix or a vector? Also, should I interpret the equation as $\nabla (v^T)$ or $\nabla^T v$?


The gradient of a vector field is

$$ (\nabla v)_{ij} = \frac{\partial v_i}{\partial x_j} $$

and so

$$ \sigma\cdot\nabla v^T=\sigma\cdot(\nabla v)^T=\sum_{i,j=1}^3\sigma_{ij}\frac{\partial v_i}{\partial x_j} $$

(given that $A\cdot B=\sum_{ij}A_{ij}B_{ji}$ and so $A\cdot B^T=\sum_{ij}A_{ij}B_{ij}$)

  • $\begingroup$ So the gradient of a vector is a matrix then, right? $\endgroup$ – Paul Sep 15 '12 at 16:50
  • $\begingroup$ Yes, it is a matrix. The gradient raises by one the order of the object to which it is applied: a gradient of a scalar field is a vector field; the gradient of a vector field is a double tensor field, and so on. $\endgroup$ – enzotib Sep 15 '12 at 16:51
  • $\begingroup$ According to the book, the final result $\sigma\cdot\nabla v^T$ is supposed to be a scalar quantity. Your final result seems to be a vector, if I'm not mistaken... $\endgroup$ – Paul Sep 15 '12 at 16:53
  • 1
    $\begingroup$ No, you sum on both indexes, so the resulting dot product is a scalar. $\endgroup$ – enzotib Sep 15 '12 at 16:55
  • $\begingroup$ Awesome! Thank you, @enzotib! :) $\endgroup$ – Paul Sep 15 '12 at 17:01

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.