Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

How to Become a Good Theoretical Physicist

staff.science.uu.nl

Post reply on HN
Print view View original

31–40 of 53 posts

Re: How to Become a Good Theoretical Physicist

#31

Post by Swizec »

brendyn wrote: post #28
Stasis5001 wrote: post #14

For reference, generally people learn all this over four years of undergrad and probably the first few years of grad school (I have a BS in physics, and a PhD in a different field, so not 100% sure on the grad school work). That's 6 years of more or less full-time work, surrounded by excellent peers and mentors, where every week you read 2-4 chapters and do 10 problems per chapter. If you're motivated and talented, y…

Scott Young self studied a 4 year MIT course in 1 year, although not to such a high level as you discussed: https://www.scotthyoung.com/blog/myprojects/mit-challenge-2/

That's super impressive, but there's a niggle. From his site:

> Did you grade the work yourself? Yes.

From my experience of doing CS in college, I think that's a problem.

The difference between what I believe and think I can prove to be correct, and what the professor says is correct, can be painfully stark sometimes. Especially where it comes to more advanced topics where it's easy to make your explanation of a concept sound correct, but contain crucial mistakes that invalidate your answer. Mistakes that only somebody with more intimate knowledge than yourself can spot.

We're talking fundamentals like writing out a proof that some algorithm is O(n^3) and classmates agreeing the proof does look correct, then the professor looking at it, saying "LoL, no. Watch this" and proving that it's O(n).

Re: How to Become a Good Theoretical Physicist

#32

Post by Mehdi2277 »

Swizec wrote: post #31
brendyn wrote: post #28

Earlier quoted context omitted.

Scott Young self studied a 4 year MIT course in 1 year, although not to such a high level as you discussed: https://www.scotthyoung.com/blog/myprojects/mit-challenge-2/

That's super impressive, but there's a niggle. From his site: > Did you grade the work yourself? Yes. From my experience of doing CS in college, I think that's a problem. The difference between what I believe and think I can prove to be correct, and what the professor says is correct, can be painfully stark sometimes. Especially where it comes to more advanced topics where it's easy to make your explanation of a conc…

Not the best example. An algorithm that is O(n) is also O(n^3). If you want a tight bound you should use big theta.

I also think while it might be a bit of an issue for undergrad, by grad school at latest for math at least you should be able to tell the difference between a correct proof and incorrect proof most of the time. Usually I'd expect it after your first/second proof heavy math class (analysis/algebra/etc). Usually when I take math tests I can tell pretty precisely what grade I'll get as I know when I'm writing a proper proof or if I'm just writing for the sake of having some progress. Admittingly, my math interests are biased towards proofs/logic and I've spent time writing proofs in coq.

Re: How to Become a Good Theoretical Physicist

#33

Post by sn9 »

paulpauper wrote: post #17
Stasis5001 wrote: post #14

For reference, generally people learn all this over four years of undergrad and probably the first few years of grad school (I have a BS in physics, and a PhD in a different field, so not 100% sure on the grad school work). That's 6 years of more or less full-time work, surrounded by excellent peers and mentors, where every week you read 2-4 chapters and do 10 problems per chapter. If you're motivated and talented, y…

I'm pretty sure the number of high-quality researchers in theoretical physics, or any major field, who are totally self-taught is really quite small. I think it's zero

I know of at least one professor in Canada (Toronto maybe?) who was self taught and got into a graduate program on the strength of a letter of recommendation from a physicist he had been corresponding with. Can't remember the name at the moment, but I think his background had been in art.

And then there's Ed Witten, who studied history and linguistics. It's not really clear when or how he studied physics (so very possibly self-taught), though he had a famous physicist father so he would have had a ready source of advice on textbooks, etc.

Re: How to Become a Good Theoretical Physicist

#35

Post by Swizec »

Mehdi2277 wrote: post #32
Swizec wrote: post #31

Earlier quoted context omitted.

That's super impressive, but there's a niggle. From his site: > Did you grade the work yourself? Yes. From my experience of doing CS in college, I think that's a problem. The difference between what I believe and think I can prove to be correct, and what the professor says is correct, can be painfully stark sometimes. Especially where it comes to more advanced topics where it's easy to make your explanation of a conc…

Not the best example. An algorithm that is O(n) is also O(n^3). If you want a tight bound you should use big theta. I also think while it might be a bit of an issue for undergrad, by grad school at latest for math at least you should be able to tell the difference between a correct proof and incorrect proof most of the time. Usually I'd expect it after your first/second proof heavy math class (analysis/algebra/etc).…

You're a better mathematician than I am. My experience passing mathematics exams involves a lot of "Ugh I know how this works in principle! Why doesn't it work when I apply it!? #@$%$#%!" I remember one time I was studying Newton's Method and got 5 different results for the same problem. It looked like I was applying the algorithm correctly in all of them, but the algo is very sensitive to small errors in arithmetic.

Hell, even for knowledge based tests, especially oral, I had this problem.

"How does CPU pipeline work?", prof "blahblahblah", swiz "Lol you have no idea what you're talking about" "Argh but that's what your book says!" "Nu-uh. Look, here" "Ugghhh I changed one little word!" "Yeah but that changes the meaning and now your explanation is wrong" "#@$@#%%!@#"

You know, little details I'd never realize on my own are wrong because I was 80% correct and that sounds correct enough, you can look up details when you need them. But prof is looking for 99.999% correct.

I did not graduate, as you can imagine. But I did get straight A's in coding-based classes. :)

Re: How to Become a Good Theoretical Physicist

#36

Post by vecter »

sn9 wrote: post #33
paulpauper wrote: post #17

Earlier quoted context omitted.

I'm pretty sure the number of high-quality researchers in theoretical physics, or any major field, who are totally self-taught is really quite small. I think it's zero

I know of at least one professor in Canada (Toronto maybe?) who was self taught and got into a graduate program on the strength of a letter of recommendation from a physicist he had been corresponding with. Can't remember the name at the moment, but I think his background had been in art. And then there's Ed Witten, who studied history and linguistics. It's not really clear when or how he studied physics (so very pos…

Ed Witten was very good at math when he was young. He scored exceptionally well in the American Mathematics Competition. He didn't pick physics up out of nowhere, he already had a quantitively talented mind.

Re: How to Become a Good Theoretical Physicist

#39

Post by Mehdi2277 »

Swizec wrote: post #35
Mehdi2277 wrote: post #32

Earlier quoted context omitted.

Not the best example. An algorithm that is O(n) is also O(n^3). If you want a tight bound you should use big theta. I also think while it might be a bit of an issue for undergrad, by grad school at latest for math at least you should be able to tell the difference between a correct proof and incorrect proof most of the time. Usually I'd expect it after your first/second proof heavy math class (analysis/algebra/etc).…

You're a better mathematician than I am. My experience passing mathematics exams involves a lot of "Ugh I know how this works in principle! Why doesn't it work when I apply it!? #@$%$#%!" I remember one time I was studying Newton's Method and got 5 different results for the same problem. It looked like I was applying the algorithm correctly in all of them, but the algo is very sensitive to small errors in arithmetic.…

Math is one place where I think it tends to be nice to be pedantic as you are starting to learn a topic. Generally, when I'm unsure of something proof wise it tells me I should re-read all the relevant definitions from that section/chapter. I also generally make it a goal that most important theorems are things I should be able to prove from scratch. Partly because if I can follow the proof well, then I'm less likely to misapply it. Lastly, I like how math has a heavy focus on building upon prior math so I try to review topics from previously taken classes often. I feel that cs classes are much less connected.

Re: How to Become a Good Theoretical Physicist

#40

Post by kayoone »

vecter wrote: post #36
sn9 wrote: post #33

Earlier quoted context omitted.

I know of at least one professor in Canada (Toronto maybe?) who was self taught and got into a graduate program on the strength of a letter of recommendation from a physicist he had been corresponding with. Can't remember the name at the moment, but I think his background had been in art. And then there's Ed Witten, who studied history and linguistics. It's not really clear when or how he studied physics (so very pos…

Ed Witten was very good at math when he was young. He scored exceptionally well in the American Mathematics Competition. He didn't pick physics up out of nowhere, he already had a quantitively talented mind.

That might be true but op was just claiming that the number of self taught people in theoretical physics is non zero.
Post reply on HN