MHB Could there be an error in the proof of the Poincare conjecture?

AI Thread Summary
The discussion centers around the nature of certainty in mathematical proofs, specifically regarding Grisha Perelman's proof of the Poincaré conjecture. It posits that while Perelman may have been confident in his proof, absolute certainty is unattainable due to human limitations in peer review. Each review increases the probability of correctness, but never reaches 100%. The conversation touches on the implications of mathematical certainty, referencing concepts like the infinite decimal 0.99999... equating to 1, and the challenges of defining "almost certain" in a probabilistic framework. It emphasizes that while rigorous checks can enhance confidence, the potential for oversight remains, reflecting the inherently human aspect of mathematics and peer review. The dialogue suggests a philosophical perspective on certainty, proposing that it is more about sufficiency than absolute assurance.
alexmahone
Messages
303
Reaction score
0
When Grisha Perelman submitted his proof of the Poincare conjecture, he may have been reasonably sure that it contained no mistakes. But he could not have been 100% sure as he is, after all, human. Each time it was checked, say by the referee of an academic journal, the probability that it contains no mistakes increases. But does it ever reach 100%? After all, the referees and checkers are human as well and they could theoretically have overlooked some subtle flaw in the proof.

So, I claim that we can be almost certain that the Poincare conjecture has been proved, but in theory we can never be 100% sure.

Comments?
 
Physics news on Phys.org
Alexmahone said:
When Grisha Perelman submitted his proof of the Poincare conjecture, he may have been reasonably sure that it contained no mistakes. But he could not have been 100% sure as he is, after all, human. Each time it was checked, say by the referee of an academic journal, the probability that it contains no mistakes increases. But does it ever reach 100%? After all, the referees and checkers are human as well and they could theoretically have overlooked some subtle flaw in the proof.

So, I claim that we can be almost certain that the Poincare conjecture has been proved, but in theory we can never be 100% sure.

Comments?

1) 0.99999... = 1

2) "Conjecture" - Will its name change when it is proven? As a conjecture, it is, by definition, unproven.
 
tkhunny said:
1) 0.99999... = 1

The proof has been checked only by finite number of people, so we can't be 99.999...% sure it's right.
 
Alexmahone said:
The proof has been checked only by finite number of people, so we can't be 99.999...% sure it's right.
Except that the sum rule [math]S = \frac{a_0}{1 - r}[/math] for an infinite geometric series is a proven fact. So 0.99999... = 1 is a certainty.

-Dan
 
topsquark said:
Except that the sum rule [math]S = \frac{a_0}{1 - r}[/math] for an infinite geometric series is a proven fact. So 0.99999... = 1 is a certainty.

-Dan
I'm confused now.
 
Alexmahone said:
So, I claim that we can be almost certain that the Poincare conjecture has been proved, but in theory we can never be 100% sure.
Comments?

This is a difficult claim to substantiate mathematically. How would you model this probabilistically (sample space, event space, probability measure) such that terms such as "almost certain" and the "law of large number of reviewers" become precisely defined and formulated?
June29 said:
I'm confused now.

An infinite (i.e. limiting) decimal expansion is well-defined in terms of a geometric series.

On a more "soft" level:

Mathematics is a human activity, and so is peer review. However, the bigger the claim, the more numerous and scrutinous the reviewers will be. In general, errors will always remain possible, also because reviewing is a "ungrateful" task: If a reviewer checks a tedious argument, he may spend (possibly a lot of) time on it, while the author(s) reap the benefits - no matter the outcome of the check.
 
June29 said:
I'm confused now.
Here's a derivation. Please let us know if you have any difficulties with it.

[math]0.99999... = \frac{9}{10} + \frac{9}{100} + \frac{9}{1000} + \text{ ...}[/math]

[math]= \left ( \frac{9}{10} \right ) \left ( 1 + \frac{1}{10} + \frac{1}{100} + \frac{1}{1000} + \text{ ...} \right )[/math]

This is a geometric series with [math]a_0 = \frac{9}{10}[/math] and [math]r = \frac{1}{10}[/math]

So
[math]S =\frac{9}{10} ~ \sum_{n = 0}^{\infty} \left ( \frac{1}{10} \right ) ^n = \frac{\frac{9}{10}}{1 - \frac{1}{10}}[/math]

[math]= \frac{\frac{9}{10}}{\frac{9}{10}} = 1[/math]

-Dan
 
I know that $0.{\overline 9} =1.$ I was confused by how the conversation got to whether or not $0.{\overline 9} =1$ is proven/certainty! (Rofl)
 
June29 said:
I know that $0.{\overline 9} =1.$ I was confused by how the conversation got to whether or not $0.{\overline 9} =1$ is proven/certainty! (Rofl)

1) Set the level certainty that satisfies you.
2) Find enough reviewers.

"Certainty" in this mortal world is only an intellectual construct. Search more for "sufficiency".
 
Back
Top