0% found this document useful (0 votes)
14 views9 pages

Problem Session 4: Solving Instances vs. Solving Problems How To Market Your P NP Solution

1) While testing programs on sample inputs can verify their functionality on those specific instances, it does not prove the programs will always solve the general problem correctly. A program that sorts lists may fail on a future, untested list. 2) Solving instances of a problem does not necessarily mean solving the problem itself. A program that answers whether strings are in a language L based on the string alone does not prove the language acceptance of L. 3) Having a polynomial-time algorithm that solves tested instances of an NP-complete problem does not mean the problem can be solved in polynomial time for all instances, or that the algorithm could be kept secret to only solve instances for that person. The P=NP

Uploaded by

Lalalalalala
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
14 views9 pages

Problem Session 4: Solving Instances vs. Solving Problems How To Market Your P NP Solution

1) While testing programs on sample inputs can verify their functionality on those specific instances, it does not prove the programs will always solve the general problem correctly. A program that sorts lists may fail on a future, untested list. 2) Solving instances of a problem does not necessarily mean solving the problem itself. A program that answers whether strings are in a language L based on the string alone does not prove the language acceptance of L. 3) Having a polynomial-time algorithm that solves tested instances of an NP-complete problem does not mean the problem can be solved in polynomial time for all instances, or that the algorithm could be kept secret to only solve instances for that person. The P=NP

Uploaded by

Lalalalalala
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 9

1

Problem Session 4
Solving Instances Vs. Solving
Problems
How to Market Your P=NP
Solution
2
Telling What a Program Does
Suppose I write a program to sort
integers.
Cant I feed it a list of integers and see
whether they come out sorted?
Or a million lists and check them all?
Yes, but maybe your program will fail
to sort the 1,000,001
th
list you try.
3
Solving Instances Vs. Solving
Problems
Instances of a problem are not problems.
Suppose TM M accepts language L, and
w is a string.
One TM M
yes
ignores its input and accepts.
Another TM M
no
ignores its input and rejects.
One answers the question is w in L?
But I cant tell which.
4
Polytime Algorithms for Part of
an NP-complete problem
Suppose I have a polytime algorithm
that works correctly on all tested
instances of an NP-complete problem.
Could I sell that solution?
Could I keep it secret so only I could solve
NP-complete problems in polytime?
5
Zero-Knowledge Proofs
Developed by Shafi Goldwasser, Silvio
Micali, and Charles Rackoff in 1985.
An early idea in cryptographic protocols.
Lets you prove that you know
something without revealing how you
know it.
Hypothetical polytime algorithm for an NP-
complete problem was key motivation.
6
Polytime Algorithm for
Instances of a Problem
It is possible to find polytime
algorithms that solve instances of NP-
complete problems.
Or even a million instances.
But as for the analogous matter of
decidability, it doesnt help us.
7
Testing Polytime Solutions
What about a polytime solution to
SAT that we test on 1,000,000 inputs
and it gives the correct answer each
time.
How do you know it is correct?
If a satisfying assignment exists, we
could expect the tester to show us one.
Which we could then check.
8
Testing Polytime Solutions (2)
But what if the algorithm says no?
We cant even check that it is correct in less
than exponential time.
But if the expression has many satisfying
assignments, we could try a few at random.
Alas, expressions with one satisfying
assignment exist.
Consider Cooks construction applied to a DTM.
9
Thanks
Thanks for being part of this course.
Good luck on the final exam.

You might also like