« first day (2744 days earlier)      last day (2572 days later) » 

00:01
what about Morse theory?
@MatheinBoulomenos But Riemannian geometry is also fairly useless for 99% of physicists. The stuff you need for GR is not quite what you'd see in a Riemannian geometry class
You still need curvature and torsion notions ... for a semi-Riemannian metric.
@MatheinBoulomenos Witten used Morse theory for something, but I don't know how exactly it ties in
Gromov-Witten classes are based on a generalization of Morse theory.
I've used Morse theory plenty, but I'm only a physicist in algebraist's eyes
00:03
Morse theory like what Witten does is different than classical morse theory (but not really)
@MatheinBoulomenos mathein :D got time?
@KasmirKhaan yeah
can I inviteyou to our room? :D
that example you solved
ok thanks :D
00:03
@MatheinBoulomenos I would say Riemannian geometry
It has a higher chance of being useful than the other stuff
I should note that the differential geometry course is for people who haven't seen manifolds before
Oh.
This gets back to my pet peeve. A basic course in differentiable manifolds is NOT differential geometry.
I agree with Ted
Connections on vector bundles, Riemannian connections, parallel translation, holonomy etc. — that's differential geometry. But people keep saying differential geometry any time a manifold lurks in the corners.
So 40%-50% of the course will be used to establish some basics on smooth manifolds and bundles, then they're doing Riemannian metrics
00:05
@MatheinBoulomenos flip a coin then
@MatheinBoulomenos He should go for whatever will least bore him
@Mathein: If he knows differential forms, I can send my notes from the geometry course I taught a few years ago ... but they're handwritten and not elegant.
How about picking whichever class is using the book you like better etc?
Detailed syllabus (and probability that the professor will actually cover what he thinks he will) becomes relevant.
As an adviser, I also operated on the basis of "amazing teacher" versus "adequate teacher" versus "horrible teacher."
@0celo7 that would probably be the differential topology course, since he knows smooth manifolds, bundles and forms already from the first differential topology course
also the topology prof is known to be a good lecturer
It's bad curriculum when there's so much overlap.
00:09
sup nerds
There should be a standard course on manifolds. Then you teach second courses doing differential topology (transversality, cobordism, etc.) and, separately, differential geometry
The thing is, we really only have one topology lecturer, so differential topology is offered irregularly, but differential geometry is offered every year
Heya nerd Eric.
So the diff top should have had the first diff geo course as a prereq, Mathein.
00:10
@EricSilva did you go to the talk?
no i had lab
@TedShifrin yes, that would be reasonable
i can maybe go to the one next week tho
is lab unskippable?
cause it's on a weird day
00:10
ok
yeah it's instant fail if you skip
Even if you email the professor/TA?
yeah there's 0 way out
unless youre ill or physically incapacitated
Oh, Mat would have helped you with the incapaciation
you could have watched in a wheelchair
l m a o
00:11
HAha
@TedShifrin maybe a classical diff geo course squeezed in there or smth
I'm just complaining about a half-course overlap on basics of differentiable manifolds — sorta like the same problem you had, Eric, with UC.
Yeah I think it's not optimal
i had a diff top course that was transversality and things with no geometry
We have the same problem with PDE here
00:12
There should be a basic course on point-set topology and some smooth manifolds
PDE 2 second half is the first half of Elliptic PDE
that's a prerequisite for alg top, diff top and diff geo
I want the basic course to cover manifolds, basics of bundles, vector fields & flows, Frobenius, and Stokes's Theorem.
Then do more advanced courses on whatever.
Point set topology is a prereq ... but you don't need all that much.
S;t'o]k\e,s.
Thanks for the Demonark effect.
00:14
@TedShifrin wha?
See Eric's previous line.
André apparently is gonna try to do a baby do carmo undergrad course next year
i think that would be nice to have here
@EricSilva We had that last year
@TedShifrin there's even more overlap. We don't have a point-set course, so the first 1.5 weeks of both alg top and diff top were on point-set (the diff geo course just assumed that you could teach yourself what you need)
I'm not fond of doCarmo's book, but whatever. Yes, there should be such a course.
00:15
No I meant more like, I always get the apostrophes right in the theorem of Stoke
it runs irregularly, when there's enough honors students who want it
@Mathein: Grad students should know basic point-set, and these are graduate courses [in my book].
i like do Carmo's booooooksssssss
@EricSilva nothing like a little nationalism in your book preferences
I don't like having to spend 100 pages belaboring 2-manifold shit to do surfaces, Eric.
00:16
@TedShifrin these are all undergrad courses here, except diff top, but that's a joke since diff top is actually easier than alg top here
I heard the reason Bourbaki started writing those books was to get to Stokes' theorem.
Not sure if they ever reached it.
i skipped that part of do carmo cause i knew what was happening lol
But it makes the course clumsy, @EricS.
@Bennett seems like a very inefficient way to get to Stokes theorem
that's fair i guess
i like his riemannian book more
00:17
Obviously, if I were happy with doCarmo's book for typical undergrads, I wouldn't have written mine. So, it follows that I wasn't.
cause it's very unfancy
Yeah, I just wish he didn't avoid moving frames entirely.
I haven't lectured out of it, so I don't have an educated opinion.
But I did hold on to it.
I still find it strange that he's apparently the type who doesn't use forms and yet he wrote a book on them
there's a lot of stuff he doesnt do that people want, but it is a good size and nice style i think
Yeah, it's strange, Demonark. But remember that he and I had the same adviser.
I'm less interested in Riemannian geometry, so I do other topics he doesn't care about (Grassmannians and characteristic classes, e.g.).
00:19
@EricSilva chapter 2 of CM is a mess
I also do more with vector bundles than the standard geometry books — my complex geometry influence.
I <3 Riemannian geo
CM vacillates between making me happy and making me furious very frequently
@EricSilva it's easier to read the original papers
same issue with everything Peter Li writes
that's what ive started doing tbh
Simons paper is actually quite ok
00:20
so what does he do that's so general?
im p sure he does the elliptic PDE for A in more than just $\mathbb{R}^{n}$
that's like the point of the paper
I'm curious, what would you say draws you guys to your preferred area of geometry over others?
other than that it's got an a decent exposition for the basics of minimal boiz
@Daminark least amount of algebra
@s.harp That looks amazingly written and fun!
00:22
well, and general relativity
@Daminark the whim of the cosmos tbh
Elon Musk is an absolute madman
@0celo7 somehow this doesn't surprise me in the least
@Daminark what does that mean?
I was joking
00:23
Eric is more analytically slanted than I am, for sure. And I liked complex algebraic geometry, so my work was more on complex differential geometry.
LOL, I figured DogAteMy would find a pic to post.
@AkivaWeinberger That's insane! I mean...
tbf im still but a lad and have yet to settle into something
@0celo7 didn't you also say stuff like "complex geometry isn't worth the amount of algebraic crap you need"
im jsut diong a lot of very analysis heavy things because they happen to excite me and i have a really good source for learning it
@MatheinBoulomenos That, and I don't find it a priori interesting
00:25
It's always wonderful when people who don't know much think they know everything.
Ted loves being passive aggressive to me
but i also feel a deep pull to learn things that are different from that
Not sure why
I found stuff on flows and vector fields unintuitive and confusing in diff geo, but there were nice theorems
i dont wanna just bury into a niche when im excited by a lot of different math (but i reeeeeeeaaaaally do like geometric analysis a looooot)
00:26
Eric, geometric analysis is so broad in its scope — you have no shortage of excuses to learn almost every type of math for a while.
> Thus Landau, who is dead, might well have objected to our use of the word “king” for his concept.
i certainly want to learn more bryant type stuff
Is this the same Landau whose elementary number theory text I've used I wonder? I remember there were like at least two...
@Bennett Since it was a test flight, they could put anything they wanted on it other than an actual satellite.
00:28
@TedShifrin (by the geom analysis comment i mean im willing to do things that arent that, not just other kinds of geometric analysis lol)
(Nobody wants to put their satellite on a rocket which has a high chance of exploding.)
I meant that as well, Eric.
Usually, the test payload is something like concrete to simulate the mass.
@Bennett there were probably 20 landaus, this one concerned himself with chickens
@Ted is the little book by Krantz that you recommended me also considered geometrical analysis?
00:30
Nah, not really.
The geometric integration theory one?
It's the interplay of differential geometry with complex analysis (which then gets very serious with several complex variables).
@s.harp and he dead too.
Krantz also has the book "Geometric Function Theory: Explorations in Complex Analysis", which I consider getting, since I liked his first one on diffgeo/complex analysis
@s.harp I think you more or less said what I was trying to say (re: Deligne). I couldn't follow after that - you think the issue you're having is still distinct?
00:31
@AkivaWeinberger So why a car? Talk about a lack of imagination!
I haven't seen that book, Mathein. But Krantz is a good writer.
I saw a headline the other day that Musk was moving to Mars.
It seems like a more advanced version of the same idea as the other book
Maybe he's gonna drive there.
Should be interesting refueling along the way, Bennett.
00:35
Maybe he's filled up Russell's teapot with petrol.
@Bennett Compared to a block of concrete?
No, Tesla is all electric, Bennett.
I am disappointed he did not send a teapot.
That may very well be true @s.harp - I am sleepy and haven't thought about this as much as you have
I will be very interested in what your professor says though
Oh, electric.
00:37
@EricSilva This is probably silly, but are all n-Lebesgue measurable sets/functions Hausdorff measurable for dimensions \le n?
Has to be
It all sounds like a Douglas Adams story.
well they're all Borel measures, duh
but the sets of measure zero might throw things off
Do we say a set is measurable if its measure is infinite? I don't like talking about the a bounded open subset of the plane as being $\mathscr H^1$-measurable.
@TedShifrin You don't like R^n being Lebesgue measurable?
It's in the sigma algebra, so it should be measurable
Hard to see there, but there's a sign saying "don't panic"
Also, there's a towel in the glovebox
00:42
I guess I don't like thinking of $2$-dimensional things as having $1$-dimensional Hausdorff measure, although of course it can make sense as infinity.
And a copy of Hitchhikers Guide to the Galaxy
LOL, DogAteMy ... You're quite amused.
@0celo7 I would not believe this
@AkivaWeinberger I love that!
@EricSilva Well, then Federer doesn't make too much sense
00:47
Oh I think it's the other way around I don't believe
Now I'm losted.
I would need convincing to believe hausdorff and lebesgue measurable sets are the same (less than top dim obvs)
It's true for Hausdorff $n$-dimensional subsets of $\Bbb R^n$.
But otherwise measure 0 encapsulates too much craziness, I bet.
Yeah this is what bothers me
@EricSilva You need Lebesgue to be Hausdorff for 3.2.22 to make sense
00:52
The measure 0 sets of lebesgue can be wild from hausdorff perspective
yeah but the statements of the coarea formulas don't make sense then
I was worried about something of mine being measurable and now I'm in the rabbit hole
hello ,
hi @F.Tasos
i want to ask a question on the local maximum principle for harmonic functions
can anyone give me any help ?
00:53
You might find someone here who knows more than a little about that.
Just ask.
ugh
the proof of the coarea formula is bad
dunno why I bother trying to read Federer
ok , here in blake's proof math.stackexchange.com/questions/1486994/… , i can't quite make sense of the last paragraph
specifically this part : "that way after showing vv is constant on one disk, it attains the maximum at the center of the subsequent disk"
Maybe try Francesco maggis book
@EricSilva it’s open on my desk.
I’m gonna eat and then check it out
@F.Tasos: Once you have the function with a maximum at any interior point, it must be constant on any disk containing that point. But choosing another disk whose center is contained in that first disk, it follows that the function is constant on the union. Can you cover the whole region in this way?
01:01
@TedShifrin what i dont get why in the subsequent disk i.e D(x1,d1) for some d1>0
it follows that u(x1)<=u(x) for all x in D(x1,d1)
Should also just be able to do an easier connectedness argument.
If $a$ is your interior max pt, you should be able to show that the set of points $x$ with $f(x)=f(a)$ is both open and closed.
identity theorem?
We're doing harmonic, though, not holomorphic.
it sounds close enough though, I mean the theorem you just stated
Where did that come from, @F.Tasos? We started by assuming we had started with an interior maximum point in our first disk.
If $a$ is a local maximum and $a$ is an interior point, then the MVP tells you that $f$ is constant on any disk centered at $a$ contained in the region.
I think the connectedness proof is easiest.
01:08
@TedShifrin Ok , lets take things from the beggining we i have that my $ f $ has a local maximum at $ a $ then by using the mean value property it follows that $ f $ is constant on some disc containing $ a $

Then we try to prove that its constant so we connect $ a $ with an other point in our domain with a path . What i dont get is why in the subsequenct disc ( for example the second step ) we get that at the center of which our functions attains a maximum ( or a local max)
@0celo7 I think I'm spooked by questions involving the lebesgue sigma Algebra cause it has the potential to stray into foundations...
You make sure the center of your second disk is inside the first disk.
@EricSilva: And you're unfounded?
yes but i still can see why it has to be a local maximum at the second disc
Because your initial max point was the global maximum point !!
So the function can only get smaller; it can't get any bigger.
it was not global , we assumed it was local
01:11
No, no.
We assumed we had an interior global max point.
At least, that's how the proof should have gone.
The statement of the maximum principle is that the global max of a harmonic function must occur on the boundary.
The theorem states that if a harmonic function attains a local maximum inside in an open $ A $ then it must be constant
open and path connected *
OK, we have different statements.
I would start by choosing the local maximum point with largest value. And then my argument is fine.
what if they are not finite ?
We have a global maximum point by compactness.
If it's on the boundary, we're done. If it's not, then my argument is fine.
we started off with an open domain
01:16
We're stating different theorems. The theorem I'm talking about is precisely what's being discussed in the page you linked me to.
oh , im sorry
2
Q: Maximum principle for harmonic functions

user393664I know the following classical maximum principle for harmonic functions: If $\Omega \subset \mathbb{C}$ is open and connected and $u \in C^2(\Omega)$ is harmonic, then $u$ has maximum (or minimum) in $\Omega$ $\implies$ $u$ constant. How can I prove that the theorem is true if the h...

Right, so you need the identity principle, which I was avoiding. But Daniel made it clear in that post.
Once you have a real analytic function that's 0 on an open set, it's 0 everywhere on its domain.
the proof in my book goes (very poorly ) in the way blake went to prove his argument so i was trying to work something out this way
we haven't been taught this theorem neither the relation between harmonic and being the real part of a holomorphic function
So you want to have a local maximum point that is not a global maximum. So you have a point where the function is larger.
@0celo7 what if you look at a hyperplane sitting in euclidean space and take a crazy subset of that guy it's gonna be lebesgue measurable but I don't see why it would be hausdorff measurable
01:22
@TedShifrin i agree so far
The function is constant on the largest possible disk centered at the original point.
Take a boundary point of that disk that's an interior point of your region.
In particular, if I choose a path from $z_0$ to this point $z_1$ where the function is larger, I look where that path leaves my largest disk. Call that point $w$.
Can the function actually increase as I move from $w$ toward $z_1$ along the path?
i dont know that
Well, no it can't, by choosing a disk centered at $w$ and applying the mean value property. On "half" of that disk the function is constant and equal to $f(w)$. So larger values won't get averaged out to give me $f(w)$.
01:27
$ w $ is the point on the boundary right?
I don't see how to make a decent argument this way.
Boundary of that largest disk and on the path to $z_1$.
But I don't see how to finish.
I don't like this argument.
me neither it's not explicit enough tbh
No, it's not a good approach.
01:31
maybe there is no way getting around the argument daniel used
I would argue that a harmonic function that is constant on an open set is globally constant (assuming connectedness).
Yeah, I can't help. Sorry. I need to get going.
@TedShifrin thanks anyway
Suppose that $X \in M_n(\Bbb{C})$ is a PSD. Let $X_0$ denote the matrix formed by replacing the $(i,j)$ and $(j,i)$-th entries of $X$ with zero. Will $\det(X_0) \ge 0$?
If it makes things easier, take $i=1$ and $j=2$.
Damn...I think it's false: take the $3 \times 3$ matrix with $1$'s in every entry...
02:02
@Ted aligot ended in disaster
02:16
Ugh... grading is depressing. :(
People aren't doing too hot?
@Daminark No.
what are you grading xander
02:25
For example, can you identify the vertex of the parabola given by $y = 2(x+1)^2 + 3$?
oh, I see...
what course is it "technically"?
And, more importantly, do you need to quadratic formula to determine the vertex?
this is precalc
they ought to be just visualizing it as shifted, no?
exactly
and I would even give partial credit to the answer $(1,3)$, since signs are hard
is this an exam or hw
02:28
oh, another question from the same page: which is greater, $-\dfrac{1}{10}$ or $-\dfrac{1}{2}$?
this is their first midterm
the questions are organized into P, C, B, and A levels
60% of the points on the exam are the "P level" questions
both of those are from that section
like, basic basic stuff that you NEED to have a handle on if you hope to pass the class
there is not enough whiskey
right
what are the A level questions like?
one of them is "Hey, here are two piecewise defined functions $f$ and $g$. Figure out when $f(x) > g(x)$."
The other is "Some guys pave a given area in a given time. If you scaled the dimensions of the region being paved by a factor of three, how long would it take some more guys to pave the new area?"
gotcha
I sat in on the precalc course once at my undergrad school, it was mostly athletes. Is that the case for yours?
02:33
The first problem involves a quadratic term, just to make things fun
@Antonios-AlexandrosRobotis No. We have no real athletics program
gotcha, I don't know too much about UCR
We are required by the state to admit every student who applies as long as they were in the top 5% or 10% of their high school class (I don't remember the exact percentage)
and the region that we serve is rife with poverty
guys can anyone help me with the proof that if a harmonic function on a region attains a local max then it's constant
and a huge number of students are first-in-family to attend college
so they don't really have any idea what kind of study skills they need in order to be successfull
@F.Tasos You mean if a harmonic function attains its max on the interior of some region, then it is constant, yes?
Local max
02:37
same difference---if it attains a local max on the interior of some region, then it attains a max on a smaller region
the point is that the maximum is on the interior, and not on the boundary
yes?
yeap but how do we go globally ?
well this is a local question, right?
what do you mean @Antonios-AlexandrosRobotis
wp
patriotis
well, if a function attains some local max at a point $w$, then there is some disk $|z-w| < \delta$ such that $|f(z)| < |f(w)|$ for all $z$ in the disk
then by the maximum-modulus principle, it is constant on that disk
You're interested in a local max, so, it's sufficient to consider a small region.
02:39
@Antonios-AlexandrosRobotis no , i need the function to be constant on the whole region
though that might not be the easiest proof---it has been a few years since I thought about the nitty-gritty of complex analysis
μιλας ελληνικά
?
ναι
just wondering ;) best to keep in english in chat, I reckon.
@F.Tasos but if a harmonic function is constant on an open set, then it is everywhere constant, no?
that theorem has a name... the identity theorem, maybe?
02:40
@XanderHenderson seems to be true haven't been taught that.
Look
Yar... it is the identity theorem
okay, you haven't been taught that... what tools do you have?
Harmonic is not holomorphic nessecarily
First , the way the proof goes in my notes is something like this : ( we are talking about harmonic functions : $ R^n \to R $ $ \Delta f =0 $
first we prove its constant in a disc with the mean value property
then we connect the original point with another with a path
Then he gets a point on the intersection of path and the boundary of the disc
he claims that using the previus argument concludes that f is constant on a disc centered that point
which makes no sense to me at all
harmonic does not imply holomorphic, but if you have a harmonic function, you can use that to build a holomorphic function, then prove something like the identity theorem; but that is neither here nor there, because those aren't the tools that you have
yeap
and by "disk", I assume you mean "ball", since you are working on $\mathbb{R}^n$?
02:46
yeah
i have seen the notation $ D(x,\epsilon) $ for the open ball
in a normed space
yuck... $B(x,r)$ is better
or $B_{>}(x,r)$ or $B(x,r)^{\circ}$ if you really need to be sure
yeap
anywho, suppose that $x_0$ is the location of your local max
and that you have shown that $f$ is constant on $B(x_0,r_0)$
join $x_0$ to some other point with a path
then pick your boundary point
call that $x_1$
now, we want to show that $f$ is constant on $B(x_1, r)$ for some $r$, right?
02:51
yeap
but you can make a similar kind of argument as that which shows that $f$ is constant on the original ball
no?
on the original ball yes
no, on the new ball---$f$ is going to be constant on a big chunk of that ball, since the new ball intersects the old ball quite a lot
yes ok
but what about the rest of the ball
man, I really just want to hit it over the head with the identity theorem...

« first day (2744 days earlier)      last day (2572 days later) »