Consider the following code:
\RequirePackage{fix-cm}
\documentclass[10pt,conference,compsoc,letterpaper]{IEEEtran}
\usepackage{amsfonts}
\usepackage{amsmath}
\newcommand{\Foo}{\mathsf{Foo}}
\newcommand{\Blob}{\mathsf{Blob}}
\newcommand{\powerset}{\mathfrak{P}}
\begin{document}
Let \(A\) \(=\) \(\Bigl\{g\colon \bigl(\Foo\mkern-1mu\times\mkern-1mu\powerset((\Blob{\times}\Foo)^2)\bigr)\to\Foo\to\mathbb{N}_{\ge 0}\,\allowbreak\Bigm|\,\allowbreak\forall\,a{\in}\Foo,\,{\leadsto}\mkern-1mu\subseteq\mkern-1mu(\Blob{\times}\Foo)^2\colon\mkern1mu \sum_{b\in\Foo}g(a,{\leadsto})(b)=\bigl\lvert\{t{<}n\mid l_t\mkern-1mu=\mkern-1mua \land {\leadsto}\mkern-1mu=\mkern-1mu{\rightarrow_t}\}\bigr\rvert\Bigr\}\).
\end{document}
It is ugly in the second line: the spacing around ⊆ is larger than the space after a∈Foo,. It should be the other way round: the space before ↝ should be larger than after it. Reason: the all-quantifier ∀ logically binds two variables: a and ↝, where a ranges over Foo, and ↝ ranges over subsets of (Blob × Foo)². Moreover, the spacing around ⊆ must remain larger than the spacing around × in (Blob×Foo)².
Of course, I could adjust the spacing manually via \mkern, \,, \, \hspace, \!, etc. in an ad-hoc way, but I've got tons of similar formulas. Therefore, I need a more general solution. It should also work when the column width is changed for the whole document. Ideas?


\subseteq, as in{\subseteq}, be close to what you had in mind? – Steven B. Segletes Jan 17 '17 at 12:17\forall a \in X, b \in Y: a < bas example. – pst Jan 18 '17 at 10:33