I want to prove the theorem ∃x(A(x)→∀x A(x)). Because it is a theorem, it has no premises. The hint in the textbook says to only use the basic rules of TFL (truth-functional logic) and basic quantifier rules. It also says that it requires the use of IP (indirect proof). Thus, I know I probably shouldn't use identity rules but I don't see another way how. I have tried the following proof:
1 -∃x(P(x) → ∀xP(x)):Assumption
2 P(a):Assumption
3 -P(b):Assumption
4 a=b:Assumption
5 a=a:=Intro
6 b=a:=Elim lines 4,5
7 a=b->b=a:-> Intro lines 4-6
8 -a=b:Assumption
9 I'm stuck here
10 !?:Negation Elim, I plan on using
11 a=b:IP lines 8-10
12 P(b):= Elim lines 2,11
13 !?:Negation Elim lines 3,12
14 P(b):IP lines 3-13
15 AxP(x):Universal Intro line 14
16 P(a)->AxP(x):->Intro lines 2-15
17 ∃x(P(x) → ∀xP(x)):Existential Intro line 16
18 !?:Negation Elim lines 1,17
19 ∃x(P(x) → ∀xP(x)): IP lines 1-18
I'm stuck on line 9 but I don't think I should even be using identity rules. So I think most of my proof is wrong anyways. Thanks for any guidance on this proof.