Title: Proof%20Strategies
1Proof Strategies
- CS/APMA 202
- Rosen section 3.1
- Aaron Bloomfield
2Proof Strategies Whats Important
- Forward and backward reasoning
- Leveraging proof by cases
- Adapting existing proofs
- Conjecture and proof
- Conjecture and counterexamples
- Open problems 3x 1
- The halting problem
3Forward reasoning
- One example modus ponens / direct proof
- Given an antecedent p and a implication p?q
- Show that conclusion q is true
- Often hard to see where the proof should lead
- In that case, do a backward reasoning to
determine the forward reasoning
4Backward reasoning
- Start with the conclusion of the theorem
- Find an antecedent that yields the given
conclusion - In other words, given conclusion q and
implication p?q, find an antecedent p that allows
the conditional to be true - Will often only be one such antecedent
5Backward reasoning
- Is not the same as modus badus!
- Modus badus
- Given p?q and q, show p via the converse q?p
- Backward reasoning
- Given p?q and q, find a p that p?q is true
- Or show that no such p can exist
- It can be any p!
6Backward reasoning
- Is not the same as modus tonens!
- Modus tonens
- Given p?q and q, show p via the contrapositive
q?p - Backward reasoning
- Given p?q and q, find a p that p?q is true
- Or show that no such p can exist
- It can be any p!
7Backward reasoning
14. Suppose five ones and four zeros are
arranged around a circle. Between any two equal
bits you insert a 0 and between any two unequal
bits you insert a 1 to produce nine new bits.
Then you erase the nine original bits. Show that
when you iterate this procedure you can never get
nine zeros. (Hint work backwards.)
0 0
0
? 0
0
? 0
0 0 0
1 0 1
0 0 1 1
0 0
1 0 1
1 1 1
0 0 0
What to prove?
8Backward reasoning
- In order to create 9 0s, the previous step had
to have all 0s or all 1s - Consider two cases for the second-to-last step
- There were 9 0s
- The only previous step for this case is 9 0s
(this case) or 9 1s (other case) - There cant be any other ways to get 9 0s
- There were 9 1s
- Then every bit needed to be different than the
bit next to it - Not possible with an odd number of bits!
9Forwards vs. Backwards reasoning
- Example 1 (Rosen, p. 215) Prove thatwhen a?b,
agt0, and bgt0
10Forwards vs. Backwards reasoning
- Backwards reasoning
- Because (a-b)2gt0 when a?b, it follows that the
inequality is true - Forwards reasoning
- Given ,we can show that is
equivalent to (a-b)2gt0 - (a-b)2gt0 is always true when a?b
11Leveraging proof by cases
- When there is no obvious way to begin a proof,
but when extra information in each case helps
move the proof forward - Try to reduce the number of cases
- What to avoid
- Make sure you get ALL the cases
12Cases
32. Prove that ën/2û én/2ù ën2/4û for
all integer n.
Consider two cases Even(n) easy
Odd(n) trickier
Let n 2k1 ?(2k1)/2? ?(2k1)/2?
?(2k1)2/4? ?k½? ?k½? ?(4k24k1)/4? (k)(k
1) ?k2k¼? k2k ?k2k¼? k2k k2k
13Quick survey
- How are we doing so far?
- Very well
- With some review, Ill be good
- Not really
- Not at all
14An optical illusion
15Another optical illusion
16Definitions
- Theorem a statement that can be shown to be true
- Or something that is widely believed to be true
- Conjecture a statement whose truth value is
unknown
17Conjecture and counterexamples
- You can prove an existential quantifier by an
example - You can disprove an universal quantifier by
example - You cannot prove an universal quantifier by
example - You cannot disprove an existential quantifier by
example
18Proof by counterexample
- 15. Prove or disprove that n2-79n1601 is a
prime whenever n is a positive integer - How to disprove?
- Find a single example that shows this to be false
- Let n 1601
- Then n2 (1601)2-79(1601)1601 15231601
19End of lecture on 15 March 2005
20Proof by example
25. Prove or disprove that there are three
consecutive odd positive integers that are
primes, that is, odd primes of the form p, p2,
p4.
What to prove?
pÎZ (p gt 0 Ù prime(p) Ù prime(p2) Ù
prime(p4) ) Or pÎZ (prime(p) Ù
prime(p2) Ù prime(p4) )
Constructive existence proof 3, 5, 7
21Proof by example
- Prove that if you have an eight gallon jug of
water and two empty jugs with capacities of five
gallons and three gallons, then you can measure
four gallons by successively pouring some of or
all of the water in a jug into another jug. - Show that there exists a sequence of steps that
yields 4 gallons in one of the jugs
22Open problems
- Fermats last theorem
- xnynzn has no solutions in integers when xyz?0
and ngt2 - Finally proven!
- Goldbachs Conjecture
- Every integer n (for ngt2) is the sum of 2 primes
- Shown for all n up to 41014
- Thought to be true
- Twin prime conjecture
- There are an infinitely many number of primes
that differ by 2 - Biggest twin prime 318,032,3612107,0011
- That number has 32,220 digits!
- Thought to be true
23Open problems
- 3x1 conjecture
- Let f(x) return x/2 if x is even, and 3x1 if x
is odd - Shown for all numbersup to 5.61013
24The Halting problem
- Is it possible to write a function that will tell
if any program will halt with any given input? - Consider H(P,I)
- Returns halts or loops forever
- Note that P and I are both bit strings
- Now consider function K
- function K (s)
- if H (s, s) loops forever then halt
- else loop forever
- We then call K(K)
25The Halting problem
- Function K
- function K (s)
- if H (s, s) loops forever then halt
- else loop forever
- We call K(K)
- Note that this calls H(K,K), which in turn calls
K(K) - If H(K,K) returns halt, then K loops forever
- But H(K,K) stated that K halts!
- If H(K,K) returns loops forever, then K halts
- But H(K,K) stated that K loops forever!
- Both cases reached contradictions
26Adapting existing proofs
- Look at existing proofs
- To use as a template for your proofs
- To build upon for your proofs
27Reuse
- 49. Show that the problem of determining whether
a program with a given input ever prints the
digit 1 is unsolvable. - Leverage off the halting problem, which is known
to be unsolvable (undecidable). - Assume the print 1 statement occurs just before
a halt statement. - Since we cant know if program execution will
ever reach the halt statement, we cant know if
it will reach the print 1 statement.
28Quick survey
- I felt I understood the material in this slide
set - Very well
- With some review, Ill be good
- Not really
- Not at all
29Quick survey
- The pace of the lecture for this slide set was
- Fast
- About right
- A little slow
- Too slow
30Quick survey
- How interesting was the material in this slide
set? Be honest! - Wow! That was SOOOOOO cool!
- Somewhat interesting
- Rather borting
- Zzzzzzzzzzz