News for hackers. Stuff that matters.
Re:P = NP Proofs: Advice to claimers by neilv on Monday April 22, 2019 @05:03PM (#19722965)
Is all the good math institutionalized now (e.g., you can't do good math without all the benefit of learning background within the institution conventions)?

What's the intuition that an extremely important proof will come from someone who gets all the TeX tweaked to camera-ready perfection, goes around and gives the talks, and all the conventional things... rather than a genius out of nowhere who walks into a stranger's office, wordlessly drops on their desk a stack of handwritten pages wrapped in twine, and walks out, never to be seen again?

I was thinking of this because I once knew an autistic person, from some humble and isolated upbringing, who was getting some informal math instruction, and they told me they thought they knew a proof for P=NP. I considered it extremely unlikely, but not utterly impossible. (They reportedly tested "off the scale" high in some particular cognitive regard, and I'd noticed a couple minor superpowers.) If that person, to use them as an example, ever actually worked out the proof, and wanted to "turn it in" with their own judgment and access, I wonder where they'd go. And whether it would be somewhere that would take them seriously enough to look at it, and whether they could tell that a proof in unconventional terms had merit. If they get turned away from the first place, would they be discouraged and give up, or would they keep trying, there or elsewhere. If they persisted, would they appear to be a crank because of their persistence and frustration.

Re:P = NP Proofs: Advice to claimers by mkl on Monday April 22, 2019 @07:32PM (#19724141)
> What's the intuition that an extremely important proof will come from someone who gets all the TeX tweaked to camera-ready perfection, goes around and gives the talks, and all the conventional things

I think it's more that vanishingly few people with poor notation are able to get things completely right. Mathematicians acquire this intuition from marking thousands of undergraduate students' work and attempting to understand the research of (at least) hundreds of postgraduate students and professional researchers (i.e. everyone "in the system"). In these people's work the intuition bad notation => mistakes is almost universally true, so the odds against an outsider with poor notation accomplishing something mathematicians have been failing to solve but chipping away at for decades are astronomical.

It's not impossible, of course, but so unlikely that it's almost certainly not worth the time and effort to look into the details. It would be almost inevitably disappointing, and there's not enough time for the more trustworthy interesting stuff as it is.

Edit: a person like the specific one you're describing can earn professionals' time and effort (trust) with smaller results that are both insightful and short and correct enough to be easily checked, and then showing that they can learn from their mistakes.

Re:P = NP Proofs: Advice to claimers by jcranmer on Tuesday April 23, 2019 @01:43AM (#19725893)
> Is all the good math institutionalized now (e.g., you can't do good math without all the benefit of learning background within the institution conventions)?

Probably not; there is a surprising amount of unsolved theorems that are interesting that could be tackled with roughly an undergraduate level of math or thereabouts. In the less-well-traveled portions of math, there's definitely more work than workers to do it. But the problems that attract the cranks (things like P=NP, Riemann, Hailstone), these have been open for long enough and studied sufficiently thoroughly that you're likely to fall into an already-known trap if you haven't studied hard enough to know what's been down before.

I'll also point out that it's not so much that you need to study in the rigors of school to get this information as it is that you'll have studied enough samples to know that people are looking for particular formatting to get your work accepted. It's not a sign that you have the academic heritage; it's a sign that you've at least done the required reading.

Re:P = NP Proofs: Advice to claimers by andrepd on Monday April 22, 2019 @08:10PM (#19724399)
You are focused on one out of of 8 heuristics that he describes on that post. He speaks about other more important ones, technical ones, before he even mentions that. He also prefaces everything by saying "the author's reputation or institution must not be a factor, for ethical and practical reasons". And I stress the word: heuristics. It's a rough way of estimating how likely is this to be correct since it is impractical to carefully examine dozens of these claims per year.
Re:P = NP Proofs: Advice to claimers by azernik on Monday April 22, 2019 @05:12PM (#19723058)
I think the intuition is that the best mathematicians, autistic or not, tend to be in academia already. So your prior for someone from the general population should be that they're more likely to be a crank than a genius, ie non-academia has many more cranks than geniuses.
Re:P = NP Proofs: Advice to claimers by neilv on Monday April 22, 2019 @05:20PM (#19723113)
It's definitely intuitive to me that most all claims from outside the institution will be mistaken/naive. But, for really big wins that have thus far evaded the institution, what's the likelihood that the solution will come from within the institution rather than from outside? (I'm asking sincerely; I'm not a mathematician, and don't know.)
Re:P = NP Proofs: Advice to claimers by azernik on Monday April 22, 2019 @05:37PM (#19723243)
There's only one example I've heard of (some Russian guy), but generally they tend to involve innovative approaches within the establishment.

(As said Russian guy proves, sometimes this heuristic fails; but your prior should be that it's likely correct.)

Re:P = NP Proofs: Advice to claimers by schuyler2d on Monday April 22, 2019 @06:47PM (#19723821)
If you are thinking of Perelman and the Poincare conjecture then I don't think his career really maps to the kind of "outsider" that is described in this thread: https://en.m.wikipedia.org/wiki/Grigori_Perelman#Early_life_...

He had a PhD in math with enough success to be offered faculty positions at Princeton and Stanford.

The fact that he recluded from academic life after that doesn't matter.

Re:P = NP Proofs: Advice to claimers by Someone on Tuesday April 23, 2019 @01:08PM (#19730624)
I was thinking of some Indian guy and (not as nice of an example) some patent officer in the Swiss patent office.

There also is the nice example of https://en.wikipedia.org/wiki/Marjorie_Rice

Re:P = NP Proofs: Advice to claimers by dagw on Wednesday April 24, 2019 @07:45AM (#19737410)
I was thinking of some Indian

Assuming you mean Ramanujan. Although he was self taught in mathematics and did much of his work outside the university setting, the vast majority of his significant discoveries and publications where made within the academic framework.

Re:P = NP Proofs: Advice to claimers by jlarocco on Monday April 22, 2019 @10:22PM (#19725076)
> I think the intuition is that the best mathematicians, autistic or not, tend to be in academia already.

Is that a safe assumption, though? Won't it limit the field to people who are happy making a low salary and don't mind teaching?

Re:P = NP Proofs: Advice to claimers by azernik on Tuesday April 23, 2019 @05:55PM (#19733335)
I'm not talking about a hard-and-fast policy; I'm talking about a Bayesian, "if a random person off the street sends a proof what should my priors be?"
Re:P = NP Proofs: Advice to claimers by tuxxy on Monday April 22, 2019 @11:02PM (#19725247)
This is an asinine assumption. What of the incredibly "genius" poor folks who do not have the opportunities to join academia?
Re:P = NP Proofs: Advice to claimers by throwawaymath on Tuesday April 23, 2019 @12:52AM (#19725695)
The point is that raw genius is insufficient for replicating Ramanujan-like accomplishments in the 21st century. It's almost irrelevant how brilliant someone is these days - if they haven't had the opportunity to learn a vast amount of theory, they can't really make significant progress towards theoretical problem.

The field simply has far less low hanging fruit than it did a century, or even a half-century ago. It's not reasonable to expect that a lone genius with little to no academic background in mathematics could resolve a famous open problem through sheer creativity. For literally generations the field's most intelligent researchers have spent their careers chipping away at this problem.

Re:P = NP Proofs: Advice to claimers by jbottoms on Tuesday April 23, 2019 @01:21AM (#19725803)
Just yeah, I didn't let Ramunujan into my school, much too messy, over promised, couldn't do proofs right. Not really academic material. He refused to learn LaTex.
Re:P = NP Proofs: Advice to claimers by viraptor on Monday April 22, 2019 @05:58PM (#19723433)
From the format point of view, it's also easier to write lots of maths in a consistent way in latex than word. Similar to: you can paint in excel, but it's uncommon enough that you'd become known for painting in excel. (Tatsuo Horiuchi) Or programming using Notepad. Ignoring what most people see as an improvement in tooling area can be easily used as one of indicators.
Re:P = NP Proofs: Advice to claimers by mojomark on Tuesday April 23, 2019 @06:31AM (#19727173)
> Ignoring what most people see as an improvement in tooling area can be easily used as one of indicators.

Well said. I would make the distinction though that intentionally ignoring a well known toolset can be liberating in the exploration of the unknown, whereas being ignorant of a well known toolset by not investing research time is likely to lead one into well known pitfalls.

Re:P = NP Proofs: Advice to claimers by throwawaymath on Monday April 22, 2019 @05:28PM (#19723168)
> Is all the good math institutionalized now (e.g., you can't do good math without all the benefit of learning background within the institution conventions)?

The short answer is yes. We haven't had a Ramanujan in decades, and they were more or less always extraordinary. Insofar as you can generalize all of its subfields together, it's reasonable to say mathematics is an extremely mature discipline now.

What I mean by that is there isn't much low hanging fruit around. Even when Ramanujan was dazzling Hardy with his results, many of them were already known. That was over a century ago. There just isn't much fertile ground left where a single brilliant mind can make significant headway based on raw talent or intuition, without first working through significant education in prerequisites. It's more and more commonplace for authors to collaborate on their work, because compelling problems at the forefront of mathematics research are increasingly requiring cross-disciplinary knowledge to resolve.

To focus on Ramanujan a bit in particular - one of the reasons Ramanujan was so talented was because he not only came up with novel results in complete isolation; rather, he also came up with original perspectives and theorems. That means he was capable of posing interesting questions, which is usually much more interesting and inspiring for further research. That's very inspiring, but the reason I point it out is because Ramanujan wasn't solving open problems in the West left and right so much as he was exercising ingenuity to pose - and resolve - open problems in fertile ground he could reach. But a century later, the amount of existing mathematics has increased so much in both breadth and depth that a burst of creativity is almost certainly not going to get an undergrad to find something that e.g. Rudin hasn't written down somewhere already. But more importantly, it's vanishingly unlikely they will resolve a famous open problem which is the subject of intense research interest. All the stories of untrained amateurs doing that are from the mid 20th century or earlier.

To turn your question on its head a bit: consider how astonishingly brilliant an untrained individual would have to be find a nontrivial result that the rest of the mathematical community hasn't found. Either they're inventing their own definitions and discovering entirely new mathematics in isolation (wow!), or they've somehow found a way to use the tools accessible to them (university analysis and algebra, at best) in a way every other trained mathematician has not. It's rare even for senior math undergrads to publish nontrivial research on their own. It's significantly rarer for that research to improve progress towards an open problem. The only example I can think of off the top of my head where undergrads actually resolved a well known open problem is AKS, and even then it wasn't a solo author.

Re:P = NP Proofs: Advice to claimers by piyushahuja on Monday April 22, 2019 @10:45PM (#19725169)
But when we are talking about unsolved problems, we are in the land of vanishing small probabilities in any case. Could it not be that the very fact that someone has been trained in conventional mathematics a hindrance to them resolving an unsolved question, and someone thinking in isolation has an advantage?

Here's Grothendieck, for example:

“In fact, most of these comrades who I gauged to be more brilliant than I have gone on to become distinguished mathematicians. Still, from the perspective of thirty or thirty-five years, I can state that their imprint upon the mathematics of our time has not been very profound. They've all done things, often beautiful things, in a context that was already set out before them, which they had no inclination to disturb. Without being aware of it, they've remained prisoners of those invisible and despotic circles which delimit the universe of a certain milieu in a given era. To have broken these bounds they would have had to rediscover in themselves that capability which was their birthright, as it was mine: the capacity to be alone.”

I can certainly see this happening: someone away from the pressures of publishing, the drudgery of admin tasks, the frustration of applying for grants or department politics, or the fear of looking "bad" to their peers etc. devoting their time to mathematics out of pure joy, play and drive and coming up with novel definitions and assumptions that those in establishment mathematics, out of pure sociological and psychological reasons, have not even dared venture into.

Re:P = NP Proofs: Advice to claimers by throwawaymath on Tuesday April 23, 2019 @01:05AM (#19725736)
I can see a case for a trained mathematician contributing a legitimately new perspective to the community by withdrawing and spending time looking at things in an unorthodox manner. That's closer to that Grothendieck was describing than what you're posing.

In the early to mid 20th century, it was feasible for someone to pick up a book on number theory and apply relative genius to an open problem to quickly solve it. The bottom line is that prerequisites for doing so were often just basic calculus and high school algebra. An understanding of sequences and series went a long way.

This isn't the case in the 21st century. We're firmly out of that territory. You can't offer a profound new perspective on a thing which you can't understand, and the barrier to understanding research mathematics continually rises. The forefront of modern mathematics is so far removed from even graduate level mathematics course material that it's not going to just be intuited through untrained brilliance. You have to actively learn it, which is (unfortunately) vanishingly unlikely outside of academia.

If the tools available to you are basic real analysis and linear algebra, P/NP is beyond your reach - full stop. At this point we actually have proofs that a valid proof of P/NP (and similar problems) cannot be achieved through large swathes of elementary techniques.

We don't live in a world like Good Will Hunting where an amateur can succeed by being a genius. That's useful in the long term but not enough on its own.

Re:P = NP Proofs: Advice to claimers by piyushahuja on Tuesday April 23, 2019 @02:44AM (#19726164)
What prerequisites, according to you, does a graduate from a good CS program need to understand P/NP enough to begin taking a stab at?
Re:P = NP Proofs: Advice to claimers by throwawaymath on Tuesday April 23, 2019 @09:55AM (#19728570)
For starters:

1. All math and CS prerequisites to an intro computational complexity course.

2. An intro computational complexity course.

3. All math and CS prerequisites to an advanced, graduate-level computational complexity course. Let’s say by this point you have worked through two linear algebra courses, three calculus courses, one or two discrete mathematics courses, some graph theory, some combinatorics, some logic, and a couple of algorithms courses.

4. The advanced, graduate-level computational complexity course. Hopefully you work on probabilistic computation and, in particular, quantum computation.

5. Now read and understand the relevant papers advancing the field (not just this one problem) for the last two decades. That one paper from 1990 about reducing the permanent to the determinant? You should know about that.

6. Finally, to reach the temple you must walk the path littered with the bodies of would-be explorers before you. Read the papers of failed proofs, starting with the easily refutable ones. The most sophisticated failed proofs have errors so subtle that they are useful for the research community in their own right as an exercise in peer review.

As a rule, a grad student near their PhD in this area should know of everything Aaronson mentioned here: https://www.scottaaronson.com/talks/pvsnp.ppt. None of that should be unfamiliar or unknown. A postdoc and beyond should actively have ideas percolating on how to chip away at some small aspect of a lesser problem featured therein.

Here’s the reality: at almost every stage of a researcher’s career, “taking a stab at” a famous outstanding problem is the wrong way forward. Usually a problem is still outstanding because it actually needs a new theory - this is the practical utility in solving theoretical problems in the first place. Therefore the best bet for solving this problem is actually to chip away at it over a long period of time.

Think hacking through a rainforest to reach a goldmine, not managing to somehow parachute to the goldmine directly when it’s hidden beneath the trees.

Re:P = NP Proofs: Advice to claimers by piyushahuja on Sunday April 28, 2019 @04:02PM (#19773350)
Ok. This looks like sound advice.

I am through 1, 2 and 3, I was thinking of learning/going straight for Ketan Mulmuley's geometric complexity theory. So you'd say that's a little misguided approach, right.

Re:P = NP Proofs: Advice to claimers by neilv on Monday April 22, 2019 @05:57PM (#19723417)
Thank you, that makes sense. Rather than this reality being discouraging to all the aspiring mathematicians who've been learning and building on their own, I think it should be encouragement -- to find ways to get closer to the community, to learn more of what they already know, and perhaps eventually find mentoring or collaboration.
Re:P = NP Proofs: Advice to claimers by Retra on Monday April 22, 2019 @07:30PM (#19724116)
If you're skilled and practiced enough to be a world-class talent in your field, it is highly unlikely that you don't work in that field as a professional.