Suppose you have two functions $f:\mathbb{R}^2\mapsto\mathbb{R}$, $g:\mathbb{R}^2\times\mathcal{A}\mapsto\mathbb{R}$, where $\mathcal{A}$ is a convex set $\mathcal{A}\subseteq\mathbb{R}^2$, and $r\in\mathbb{R}^2$ is some arbitrary but known point in space. Does $$sup_{\boldsymbol{x}}\{f(\boldsymbol{x})-sup_{\boldsymbol{y}\in\mathcal{A}}\{g(\boldsymbol{x},\boldsymbol{y}) \} \} = sup_{\boldsymbol{x},\boldsymbol{y}\in \mathcal{A}}\{ f(\boldsymbol{x}) - g(\boldsymbol{x},\boldsymbol{y}) \}$$ hold? For example $$sup_{\boldsymbol{x}}\{\boldsymbol{x}^T\boldsymbol{r}-sup_{\boldsymbol{y}\in \mathcal{A}}\{ \boldsymbol{x}^T\boldsymbol{y}\} \} = sup_{\boldsymbol{x},\boldsymbol{y}\in \mathcal{A}}\{ \boldsymbol{x}^T\boldsymbol{r} - \boldsymbol{x}^T\boldsymbol{y}\}$$ ? So that if $\mathcal{A}=\{\boldsymbol{y}\in\mathbb{R}^2 | A\boldsymbol{y}\leq\boldsymbol{b}\}$ is a convex polyhedron, then $$sup_{\boldsymbol{x}}\{\boldsymbol{x}^T\boldsymbol{r}-sup_{\boldsymbol{y}}\{ \boldsymbol{x}^T\boldsymbol{y}:A\boldsymbol{y}\leq\boldsymbol{b}\} \} = sup_{x,y}\{ \boldsymbol{x}^T\boldsymbol{r} - \boldsymbol{x}^T\boldsymbol{y}:A\boldsymbol{y}\leq\boldsymbol{b}\}?$$ Can I do this? I think this question is somewhat related to this question but here, the problem involves not only one function.
EDIT: In this thread the rules for supremum and infimum used in LinAlgs answer are discussed. Also, the book "Lojasiewicz, S. (1988). An Introduction to the Theory of Real Functions. John Wiley & Sons, Inc." is referenced, where further rules/properties can be found in the introduction.