Gaussian Function - Multi-dimensional Gaussian Function

Multi-dimensional Gaussian Function

In an -dimensional space a Gaussian function can be defined as

f(x) = \exp(-x^TAx) \;,

where is a column of coordinates, is a positive-definite matrix, and denotes transposition.

The integral of a Gaussian function over the whole -dimensional space is given as

\int_{\mathbb{R}^n}\exp(-x^TBx)dx = \sqrt{\frac{\pi^n}{\det{B}}} \;.

It can be easily calculated by diagonalizing the matrix and changing the integration variables to the eigenvectors of .

More generally a shifted Gaussian function is defined as

f(x) = \exp(-x^TAx+s^Tx) \;,

where is the shift vector and the matrix can be assumed to be symmetric, . The following integrals with this function can be calculated with the same technique,

\int d^nx e^{-x^TBx+v^Tx} = \sqrt{\frac{\pi^n}{\det{B}}} \exp(\frac{1}{4}v^TB^{-1}v)\equiv \mathcal{M}\;.
\int d^n x e^{- x^T B x + v^T x} \left( a^T x \right) = (a^T u) \cdot \mathcal{M}\;,\; {\rm where}\; u = \frac{1}{2} B^{- 1} v \;.
\int d^n x e^{- x^T B x + v^T x} \left( x^T D x \right) = \left( u^T D u + \frac{1}{2} {\rm tr} (D B^{- 1}) \right) \cdot \mathcal{M}\;.
\begin{align} & \int d^n x e^{- x^T A' x + s'^T x} \left( - \frac{\partial}{\partial x} \Lambda \frac{\partial}{\partial x} \right) e^{- x^T A x + s^T x} = \\ & = \left( 2 {\rm tr} (A' \Lambda A B^{- 1}) + 4 u^T A' \Lambda A u - 2 u^T (A' \Lambda s + A \Lambda s') + s'^T \Lambda s \right) \cdot \mathcal{M}\;, \\ & {\rm where} \; u = \frac{1}{2} B^{- 1} v, v = s + s', B = A + A' \;. \end{align}

Read more about this topic:  Gaussian Function

Famous quotes containing the word function:

    “Our father has an even more important function than modeling manhood for us. He is also the authority to let us relax the requirements of the masculine model: if our father accepts us, then that declares us masculine enough to join the company of men. We, in effect, have our diploma in masculinity and can go on to develop other skills.”
    —Frank Pittman (20th century)