CS 511, Fall 2018, Lecture Slides 09

Do You Believe de Morgan’s Laws?

Assaf Kfoury

September 11, 2019

Assaf Kfoury, CS 511, Fall 2019, Lecture Slides 09 page 1 of 14

Do You Believe de Morgan’s Laws Are Tautologies?
I Of course you believe they are!
I But now, for each, choose a most efficient procedure to confirm it!

Assaf Kfoury, CS 511, Fall 2019, Lecture Slides 09 page 2 of 14

Do You Believe de Morgan’s Laws Are Tautologies?
I Of course you believe they are!
I But now, for each, choose a most efficient procedure to confirm it!
I de Morgan’s laws can be expressed as valid WFF’s/tautologies:
1. |= ¬(p ∧ q) → (¬p ∨ ¬q)
2. |= (¬p ∨ ¬q) → ¬(p ∧ q)
3. |= ¬(p ∨ q) → (¬p ∧ ¬q)
4. |= (¬p ∧ ¬q) → ¬(p ∨ q)
or, in the form of four formally deducible sequents:

1. ` ¬(p ∧ q) → (¬p ∨ ¬q)

2. ` (¬p ∨ ¬q) → ¬(p ∧ q)
3. ` ¬(p ∨ q) → (¬p ∧ ¬q)
4. ` (¬p ∧ ¬q) → ¬(p ∨ q)
Assaf Kfoury, CS 511, Fall 2019, Lecture Slides 09 page 3 of 14
Available methods

Already discussed:
I Truth-tables to establish |= ϕ?
I Natural-deduction formal proofs to establish ` ϕ?

Yet to be discussed:
I Analytic tableaux?
I Resolution?
I BDD, OBDD, or ROBDD?
I DP or DPLL or CDCL procedures?

In this set of slides we restrict the comparison to

truth-tables and natural-deduction proofs. We delay the
comparaison with the other methods to later handouts.

Assaf Kfoury, CS 511, Fall 2019, Lecture Slides 09 page 4 of 14

Natural-deduction proof of de Morgan’s law (1):

 ¬(p ∧ q) assume
 ¬(¬p ∨ ¬q) assume
 ¬p assume
 (¬p ∨ ¬q) ∨i 3
 ⊥ ¬e 2, 4
 ¬¬p ¬i 3-5
 ¬q assume
 ¬p ∨ ¬q ∨i 7
 ⊥ ¬e 2, 8
 ¬¬q ¬i 7-9
 p ¬¬e 6
 q ¬¬e 10
 p∧q ∧i 11, 12
 ⊥ ¬e 1, 13
 ¬¬(¬p ∨ ¬q) ¬i 2-14
 (¬p ∨ ¬q) ¬¬e 15
 ¬(p ∧ q) → (¬p ∨ ¬q) →i 1-16

Assaf Kfoury, CS 511, Fall 2019, Lecture Slides 09 page 5 of 14

Natural-deduction proof of de Morgan’s law (2):

 ¬p ∨ ¬q assume
 p∧q assume
 p ∧e1
 q ∧e2
 ¬p assume
 ¬q assume
 p assume
 ⊥ ¬e 4, 6
 ¬p ¬i 7-8
 ¬p ∨e 1, 5-5, 6-9
 ⊥ ¬e 3, 10
 ¬(p ∧ q) ¬i 2-11
 (¬p ∨ ¬q) → ¬(p ∧ q) →i 1-12

Assaf Kfoury, CS 511, Fall 2019, Lecture Slides 09 page 6 of 14

Natural-deduction proof of de Morgan’s law (3):

 ¬(p ∨ q) assume
 p assume
 p∨q ∨i 2
 ⊥ ¬e 1, 3
 ¬p ¬i 2-4
 q assume
 p∨q ∨i 6
 ⊥ ¬e 1, 7
 ¬q ¬i 6-8
 ¬p ∧ ¬q ∧i 5, 9
 ¬(p ∨ q) → (¬p ∧ ¬q) →i 1-10

Assaf Kfoury, CS 511, Fall 2019, Lecture Slides 09 page 7 of 14

Natural-deduction proof of de Morgan’s law (4):

 ¬p ∧ ¬q assume
 ¬p ∧e 1
 ¬q ∧e 1
 p∨q assume
 p assume
 q assume
 ¬p assume
 ⊥ ¬e 3, 6
 ¬¬p ¬i 7-8
 p ¬¬e 9
 p ∨e 4, 5-5, 6-10
 ⊥ ¬e 2, 11
 ¬(p ∨ q) ¬i 4-12
 (¬p ∧ ¬q) → ¬(p ∨ q) →i 1-13

Assaf Kfoury, CS 511, Fall 2019, Lecture Slides 09 page 8 of 14

Natural-deduction proof of de Morgan’s law (4), once more:
We organize the proof differently to make explicit how the rule “∨E” is used on line 10;
“∨E” has three antecedents, two of which are boxes (here: the first box has one line,
{line 5}, and the second box has five lines, {line 5, line 6, line 7, line 8, line 9}.

 ¬p ∧ ¬q assume
 ¬p ∧e1 1
 ¬q ∧e2 1
 p∨q assume
 p assume q assume
 ¬p assume
 ⊥ ¬e 3, 5
 ¬¬p ¬i 6-7
 p ¬¬e 8
 p ∨e 4, 5-5, 5-9
 ⊥ ¬e 2, 10
 ¬(p ∨ q) ¬i 4-11
 (¬p ∧ ¬q) → ¬(p ∨ q) →i 1-12

Assaf Kfoury, CS 511, Fall 2019, Lecture Slides 09 page 9 of 14

Truth-table verification of de Morgan’s laws (1) and (3):

p q ¬p ¬q p∧q ¬p ∨ ¬q ¬(p ∧ q) ¬(p ∧ q) → (¬p ∨ ¬q)

T T F F T F F T
T F F T F T T T
F T T F F T T T
F F T T F T T T

p q ¬p ¬q p∨q ¬p ∧ ¬q ¬(p ∨ q) ¬(p ∨ q) → (¬p ∧ ¬q)

T T F F T F F T
T F F T T F F T
F T T F T F F T
F F T T F T T T

and similarly for de Morgan’s laws (2) and (4)

Assaf Kfoury, CS 511, Fall 2019, Lecture Slides 09 page 10 of 14

natural-deduction proofs versus truth-tables

I For the four de Morgan’s laws on slide 2, each with two propositional variables p
and q, truth-tables beat natural-deduction proofs – or do they?

Assaf Kfoury, CS 511, Fall 2019, Lecture Slides 09 page 11 of 14

natural-deduction proofs versus truth-tables

I For the four de Morgan’s laws on slide 2, each with two propositional variables p
and q, truth-tables beat natural-deduction proofs – or do they?
I Two of the four de Morgan’s laws are intuitionistically valid and two are not. The
truth tables do not show it, the natural-deduction proofs show it:

I the formal proofs for de Morgan’s (1) and (4) on slide 5 and slide 8
are not admissible/correct intuitionistically (they use rule “¬¬E”).
I the formal proofs for de Morgan’s (2) and (3) on slide 6 and slide 7
are admissible/correct intuitionistically (they do not use rule
“¬¬E” nor the two rules derived from it, LEM and PBC).

I but perhaps we did not try hard enough to avoid the rule “¬¬E” in the
formal proofs of (1) and (4)???
I it can be shown (not easy) that, however hard we may try, there are
no intuitionistically admissible/correct formal proofs of de Morgan’s
(1) and (4).

Assaf Kfoury, CS 511, Fall 2019, Lecture Slides 09 page 12 of 14

natural-deduction proofs versus truth-tables
Exercise
1. Write a natural-deduction proof of the following WFF:

ϕ1 , ¬(p ∧ q ∧ r) → (¬p ∨ ¬q ∨ ¬r)

This is a more general version of de Morgan’s law (1) on slide 5.

2. Write a natural-deduction proof of the most general de Morgan’s law (1):

ϕ2 , ¬(p1 ∧ · · · ∧ pn ) → (¬p1 ∨ · · · ∨ ¬pn )

where n > 2.
Hint: Use the natural-deduction on slide 5 to guide you.
3. Show there is a natural-deduction proof of the generalized de Morgan’s law
above ϕ2 whose length (the number of lines in the proof) is O(n).
4. Compare the complexity of a natural-deduction proof of ϕ2 and the complexity of
a truth-table verification of ϕ2 .

Assaf Kfoury, CS 511, Fall 2019, Lecture Slides 09 page 13 of 14

 ( THIS PAGE INTENTIONALLY LEFT BLANK )

Assaf Kfoury, CS 511, Fall 2019, Lecture Slides 09 page 14 of 14