Interior of a family of sets

43 Views Asked by At

I've the following result: Given any family of set $\{S_k\}_{k\in K}$, then $$\tag{*} \operatorname{cl}\left(\prod_{k\in K}S_k\right) =\prod_{k\in K}\operatorname{cl}(S_k), $$ and I'm looking for $$\tag{a} \operatorname{int}\left(\prod_{k\in K}S_k\right) =\prod_{k\in K}\operatorname{int}(S_k). $$

Further I use, with some freedom in the notation, $$\tag{b} \left(\prod_{k\in K} S_k\right)^C =\bigcup_{k\in K} \left(S_k{}^C\times\prod_{j\in K-k} X_j\right) $$ and $$\tag{c} \bigcup_{k\in K}\operatorname{cl}(S_k)\subseteq \operatorname{cl}\left(\bigcup_{k\in K}S_k\right), $$ and $$\tag{d} \operatorname{int}(S)=\big(\operatorname{cl}(S^C)\big)^C $$ Then $$ \begin{align} \operatorname{int}\left(\prod_{k\in K}S_k\right) &=^{(e)} \left(\operatorname{cl}\left(\left(\prod_{k\in K}S_k\right)^C\right)\right)^C \\ &=^{(f)} \left(\operatorname{cl}\left(\bigcup\left(S_k^C\times\prod_{j\in K-k}X_j\right)\right)\right)^C \\ &=^{(g)} \left(\bigcup\operatorname{cl}\left(S_k^C\times\prod_{j\in K-k}X_j\right)\right)^C \\ &=^{(h)} \left(\bigcup\left(\operatorname{cl}(S_k^C)\times\prod_{j\in K-k}X_j\right)\right)^C \\ &=^{(i)} \bigcap\left(\operatorname{cl}(S_k^C)\times\prod_{j\in K-k}X_j\right)^C \\ &=^{(j)} \bigcap\left(\left(\operatorname{cl}(S_k^C)^C\times\prod_{j\in K-k}X_j\right)\cup \left(\operatorname{cl}(S_k^C)\times X_j^C\times \prod_{\ell\in K-k-j}X_\ell\right)\right) \\ &=^{(k)} \bigcap\left(\operatorname{cl}(S_k^C)^C\times\prod_{j\in K-k}X_j\right) \\ &=^{(l)} \prod_{k\in K}\operatorname{cl}(S_k^C)^C\\ &=^{(m)} \prod_{k\in K}\operatorname{int}(S_k), \end{align} $$ where in (e) we applied (d); in (f) we applied (b); in (g) we applied (c) modified with the equality; in (h) we applied (*); in (i) we applied De Morgan; in (j) we applied (b) again; in (k) we applied $X_j^C=\varnothing$; in (l) we applied a not given formula; and in (m) we applied (d) again.

Question: Is the claim (a) correct? Are the used result (b) and (c) modified with the equality (I need the equality) correct?

1

There are 1 best solutions below

2
On BEST ANSWER

Claim a) is wrong as soon as you do not have finitely many sets, because a product of open subsets is then not open in the product set with the product topology. There are, however, other topologies one can use on a product space, such as the box topology. https://en.m.wikipedia.org/wiki/Box_topology

The issue in your proof is at step (g), because closure does not commute with reunion in general — only with finite reunion.