Solution

ID: past-exam-of-the-year-1-of-the-computer-science-course-of-the-university-of-oxford/2019/digital-systems-linear-algebra-introduction-to-formal-proof/8/c/solution

Assume the existential statement and eliminate it using a fresh witness . Let abbreviate . Instantiating its universal quantification at the witness itself gives two incompatible conditional requirements.
1  | ∃x ∀y ((R(x,y) → ¬R(y,y)) ∧ (¬R(y,y) → R(y,y)))     assume
2  | | ∀y ((R(u,y) → ¬R(y,y)) ∧ (¬R(y,y) → R(y,y)))        fresh witness
3  | | (A → ¬A) ∧ (¬A → A)                                 ∀-elim, 2, y := u
4  | | A → ¬A                                               ∧-elim-L, 3
5  | | ¬A → A                                               ∧-elim-R, 3
6  | | | A                                                  assume
7  | | | ¬A                                                 →-elim, 6 and 4
8  | | | ⊥                                                  ¬-elim, 6 and 7
9  | | ¬A                                                   ¬-intro, discharge 6–8
10 | | A                                                    →-elim, 9 and 5
11 | | ⊥                                                    ¬-elim, 10 and 9
12 | ⊥                                                      ∃-elim, 1, subproof 2–11
13 ¬∃x ∀y ((R(x,y) → ¬R(y,y)) ∧ (¬R(y,y) → R(y,y)))        ¬-intro, discharge 1–12
The contradiction at line 12 contains no free witness variable, so the existential quantifier elimination rule meets its eigenvariable condition. This argument is already valid without classical first-order logic double-negation elimination.

New to topics? Read the docs here!