Infinitely often and almost always

From Machinelearning

Let A1,A2,A3,… be a sequence of events in some sample space Ω. Let ω∈Ω be an outcome.

In the following table, all statements in the "infinitely often" column are logically equivalent. Similarly, all statements in the "almost always" column are logically equivalent.

perspective infinitely often almost always
unions and intersections ω∈⋂N=1∞⋃n=N∞An ω∈⋃N=1∞⋂n=N∞An
first-order quantifiers ∀N≥1∃n≥N:ω∈An ∃N≥1∀n≥N:ω∈An
verbal expression ω∈An for infinitely many n≥1 ω∈An for almost all n≥1, i.e. ω∈An for all but finitely many n≥1, i.e. ω∉An for finitely many n≥1
lim sup/lim inf ω∈lim supn→∞An ω∈lim infn→∞An
limit of sup/inf ω∈limN→∞⋃n=N∞An ω∈limN→∞⋂n=N∞An

Analogy with sequences of real numbers

Let (an)n=1∞ be a sequence of real numbers, let ϵ>0 be a real number, and let x be a real number.

We say (an)n=1∞ is eventually ϵ-close to x iff there exists some N≥1 such that for all n≥N we have |an−x|≤ϵ.

We say that (an)n=1∞ is continually ϵ-adherent iff for every N≥1 there exists some n≥N such that |an−x|≤ϵ.

I think we can even define An:={x∈R:|an−x|≤ϵ}.

lim supn→∞|an−x|=infN≥1supn≥N|an−x|≤ϵ

lim infn→∞|an−x|=supN≥1infn≥N|an−x|≤ϵ -- I think this one is equivalent to infinitely often, which is confusing since now the quantifier order has seemingly switched.

but this makes sense in terms of strength of "infinitely often" vs "almost always". We have lim infn→∞|an−x|≤lim supn→∞|an−x|, so if lim supn→∞|an−x|≤ϵ (i.e. an is ϵ-close to x almost always) then lim infn→∞|an−x|≤ϵ (i.e. an is ϵ-close to x infinitely often).

perspective infinitely often almost always
(an)n=1∞ is continually ϵ-adherent (an)n=1∞ is eventually ϵ-close to x
|an−x|≤ϵ for infinitely many n |an−x|≤ϵ for all but finitely many n
lim infn→∞|an−x|≤ϵ lim supn→∞|an−x|≤ϵ