I think it's simpler if I write down where I wanted to use this symbol, rather than trying to explain it in an abstract way. I'll give a simple example and then explain the actual problem I was working on.
Simple example: Let $a \in \mathbb{N}$ and $b \in \mathbb{Z}$, then $a * b \in \mathbb{Z}$. The symbol I'm looking for (let's say it's $\{\cdot\}$) would be used like this: $\{\mathbb{N}\} * \{\mathbb{Z}\} = \{\mathbb{Z}\}$
The example I was working on was to prove that the Hessian matrix, $H$, of a likelihood function for a normal distribution is negative definite at the solutions to the likelihood equations. Calculating $z^{T}Hz$ where $z = (a, b)^T$ gives:
$$z^THz = \frac{-a^2n^2}{S_{xx}}+\frac{-b^2n^3}{2S_{xx}^2}$$
I know that
- $a, b \in \mathbb{R}$ and $(a, b) \ne (0, 0)$
- $n \in \mathbb{N^+}$
- $S_{xx} \in \mathbb{R_{\ge 0}}$
Then it can be shown that from the two elements that are added together only one can be zero under these conditions and the other one must be negative, so the Hessian is negative definite. Is there a symbol that I could use to show something like $-\{\mathbb{R}\}^2\{\mathbb{N}\}^2 = \{\mathbb{R^-}\}$, etc. deriving the result?
The question is not about how to calculate the definiteness but if there is a symbol that could be used here?
Thanks,
Norbert