# Integration problem in matrix calculus

Let $\mathbf{A}=\begin{bmatrix} f(x_1,x_1), & \ldots,& f(x_1,x_n)\\ \vdots&\ddots& \vdots \\f(x_n,x_1),&\ldots, &f(x_n,x_n) \end{bmatrix}$, where $f:\mathbb{R}\times\mathbb{R}\rightarrow \mathbb{R}$. I want to calculate $\int \mathbf{A}\mathrm{d}\mathbf{x}$, where $\mathbf{x}=\begin{bmatrix}x_1\\ \vdots\\x_n \end{bmatrix}$.

It would be nice if you can show some references.

Thanks :)

-

Consider the change

$${\rm d} {\boldsymbol y} = {\boldsymbol A}\, {\rm d} {\boldsymbol x}$$ $$= \left[ \begin{matrix} f(x_1,x_1)\,{\rm d}x_1 + f(x_1,x_2)\,{\rm d}x_2 + \ldots \\ f(x_2,x_1)\,{\rm d}x_1 + f(x_2,x_2)\,{\rm d}x_2 + \ldots \\ \vdots \end{matrix} \right]$$

and its integral

$${\boldsymbol y} = \int {\boldsymbol A}\, {\rm d} {\boldsymbol x}$$

$$= \left[ \begin{matrix} \int f(x_1,x_1)\,{\rm d}x_1 + \int f(x_1,x_2)\,{\rm d}x_2 + \ldots \\ \int f(x_2,x_1)\,{\rm d}x_1 + \int f(x_2,x_2)\,{\rm d}x_2 + \ldots \\ \vdots \end{matrix} \right]$$

-
thx, this looks reasonable to me. – han Jul 13 '12 at 14:51

Well, it seems the result should be the matrix whose $(i,j)$ coefficient is $$\int_{\mathbb R^n} f(x_i,x_j)\mathrm dx_1\cdots\mathrm dx_n.$$ Be mindful though that the diagonal coefficients are undefined as soon as $n\geqslant1$ and the off-diagonal coefficients are undefined as soon as $n\geqslant2$, hence the whole notion seems rather dubious.

-