« first day (2411 days earlier)      last day (2907 days later) » 

20:02
@Ted Can I just do a couple of the (later) exercises in 1.4? Most of them look not too hard, but I've found a couple interesting ones
@Zach: In general, you should be doing more interesting ones, just so long as you make sure you have the computational skills down. (You seem to have skipped my favorite geometric ones at the end of section 2.)
There are some cool geometric things in section 5, too, even though you know the content there.
geometric? Like, for instance, that law of cosine one?
Those looked pretty interesting, let me check em out :P
No, that's boring. The ones about intersections of angle bisectors, medians, perpendicular bisectors, etc.
But my favorite result about a triangle inscribed in a circle with one side a diameter is cool with vectors.
@Danu: You convinced yet?
Hi chat
@TedShifrin Now I'm curious. Lemme try to construct the proof of that.
20:05
Salut @Astyx
The fact that it makes a right angle?
@Astyx, @LeGrandDODOM, @Hippa, and anyone else in Paris or nearby: I shall in fact be in Paris June 5-11.
@Semiclassic: Don't give it away.
Lol
I'll figure it out myself, then.
@MeowMix Yes.
@TedShifrin I'm a bit confused about the difference between Euler characteristic of the total space (which is what it seems I'm computing if I compute e.g. $\chi$ of a cone or cylinder) as opposed to the Euler number of a bundle
20:07
@Danu: We're doing $\chi$ of the space, right?
Oh, I see it. Very cute.
With a little luck I won't have oral exams at that time @Ted
@Danu There's no bundle in sight.
Well, there's an almost bundle, @Mike :P
20:08
@MikeMiller The argument from my notes actually uses the Euler class of this $U(1)$ bundle over the non-singular part
I was suggesting thinking about collapsing one fiber in that bundle, @MikeM.
But what @Mike said was about $\chi$ of the total space I guess
@Astyx: J'espère que non.
I suggest calculating the Euler characteristic using its additivy.
@Danu: But we're not doing characteristic classes here. We're doing classical Euler characteristic of a triangulable space.
20:09
@TedShifrin my notes did do characteristic classes
That's why I'm thinking about those
@TedShifrin The one regarding $\vec{x} + \vec{y}$ bisecting the angle between $\vec{x}$ and $\vec{y}$ when they have equal lengths wasn't that hard
@MikeM: I suggested thinking about comparing a cylinder and a cone.
but maybe I shouldn't be
$\chi (X \cup Y) = \chi(X) + \chi(Y) -\chi(X \cap Y)$.
Additivity under what, @Mike?
20:09
@Zach: Not hard, but did you do it the "efficient" way? And same question for part b. You can bludgeon things or you can be thoughtful.
Okay, seems fair enough
hello
So you say not to use the Euler class of this circle bundle?
The choice of origin seems important as well for vector proofs.
Alternatively the beautiful theorem (why it's beautiful is a perhaps nontrivial exercise) that if $X$ has an $S^1$ action, $\chi(X)=\chi(X^{S^1})$, the latter being the fixed point set
20:11
No, @Danu. I said it again. Think in terms of a triangulation adapted to the geometry we're discussing.
Mike is saying essentially the same thing.
I mean, the thing I suggested earlier wouldn't have worked nearly as nicely had I taken one of the vertices as my origin.
(it still would've worked, of course.)
@Semiclassic: Yes, of course.
This reminds me a bit of elastic collisions in lab vs. COM frame.
@TedShifrin Yeah, don't worry. I understand your cylinder/cone thing. No problem. I'm just wondering why Kotschick used the characteristic class then.
Sure, Semiclassic.
20:13
I mean, not surprising since conservation of linear momentum is a vector equation. But I always liked that.
Well, there are results about characteristic classes with group actions. Maybe he's trying to get you thinking about those.
Hi @heather
@Ted I just did $\cos \theta = \frac{\vec{x} \cdot (\vec{x} + \vec{y})}{||x||||\vec{x+y}||}$ and for the other angle $\cos \theta = \frac{\vec{y} \cdot (\vec{x} + \vec{y})}{||y||||\vec{x+y}||}$ then i substituted $||x||$ for $||y||$ in the denominator, and distributed the dot products in the numerator
@TedShifrin, how are you?
@Zach: So symmetry argument, in other words. BTW, please use different letters for the different angles. But, yes.
Doing just peachy, heather. How're you? :)
Hey @heather
20:14
This does remind me of a certain problem I had to give my first-semester students for a discussion problem.
If you do it by hand with algebra, it's a pain in the arse.
If you use vectors, though, it comes out slickly. (And it was a scattering problem, appropriately.)
Vectors are very powerful, @Semiclassic. Why not use the power?
time for number 22
Hi @Sha
Alas, students tend to fear vectors.
20:15
Hi Ted and chat!
pretty good =) I did that proof you gave me the other day - I figured out it didn't matter that there were parallel lines, but if it was an isosceles triangle, the opposite angles must be congruent, and if a scalene triangle, the opposite angles can't be; therefore in a square the hypotenuse bisects, and otherwise it doesn't. @TedShifrin
@MeowMix hello
@Zach: Did you see the elegant way to do part b of that angle bisector one?
hi @Sha
@TedShifrin let me see
Hi......... Meow?XD
:)
I saw the proof to 21 in one of your first few lectures
20:16
You had an identity shift? @MeowMix
Indeed
I find the vector-anxiety of my students to be rather frustrating, if I'm honest.
Yes, @Zach, #22 is one of my favorite geometry theorems ever. I still remember the proof I was shown in 10th grade. But the vector proof is cool, too.
hahah cool @MeowMix
0
A: Finding the changing rate

T L DavisIt's not entirely clear from the problem statement, but assume the astronaut's weight vs. altitude is $W=W_0\left( {c \over c+x}\right)^2$ where $W_0=150$ and $c=6400$. Then $$ {dW \over dt} = {dW \over dx}{dx \over dt} = vW_0c^2{d \over dx}(c-x)^{-2} = {2vW_0c^2 \over (c-x)^3} $$ For $v=6$ and $...

20:16
@Ted That theorem actually appears quite a bit in olympiad stuff :P
@Semiclassic: It's hard to do serious physics (or math) with vector anxiety.
I mean, I can understand it: Vectors take getting used to, and we don't really give them a chance to do so.
check it out if anyone would like to help
It really is.
Right, @Zach. I made up problems for our high school math competition using it most years.
20:17
@Danu I'm still not sure what bundle he's taking the Euler class of.
oh god who starred that video i posted
I hope someone has time for this simple algebra question;
Let $n\in\mathbb Z_{>0}$ and $a\in\mathbb Z$.
1. If $n$ is odd, then $\overline a=\overline{-a}\iff\overline a=\overline0.$
2. If $n$ is even, then $\overline a=\overline{-a}\iff\overline a=\overline0$ or $\overline a=\overline{n/2}.$
It's easy to show the "$\impliedby$" direction for both statements. I'm having difficulties with the other direction though. Consider 1.
The index bundle? I forget.
(> . >)
We have $a=q_1n+r$ and $-a=q_2n+r$, for $q_1,q_2,r\in\mathbb Z$ and $0\leq r<n$. We need to show that $0<r<n$ is not possible. Assume $r\in\{1,\dots,n-1\}$. We know that $a=q_1n+r=-q_2n-r$. So we can write $r=-(\frac{1}{2}q_1+\frac{1}{2}q_2)n$, since $n$ is odd. We need that $q_1$ and $q_2$ are even; is this somehow impossible? I need a contradiction.
20:17
You can kinda-sorta get through intro physics without really appreciating vectors, but
@heather: It should just be vertical angles when you cut parallel lines with a transversal :)
@MikeMiller The $U(1)$ bundle over the non-singular part.
0
A: Finding the changing rate

T L DavisIt's not entirely clear from the problem statement, but assume the astronaut's weight vs. altitude is $W=W_0\left( {c \over c+x}\right)^2$ where $W_0=150$ and $c=6400$. Then $$ {dW \over dt} = {dW \over dx}{dx \over dt} = vW_0c^2{d \over dx}(c-x)^{-2} = {2vW_0c^2 \over (c-x)^3} $$ For $v=6$ and $...

Don't spam.
@Sha: You're working in $\Bbb Z_n$?
20:18
@TedShifrin, well, that works too I suppose =) But the way I ended up doing it makes sense to me, so.
That's fine, @heather :)
You've already posted the question once. If people are going to respond, they will.
yes indeed @Ted sorry I forgot to add that
@Semiclassical Ok all good
Something about "it extends to the truncated moduli space" which is supposed to imply that its Euler number is zero. @MikeMiller
20:19
@Sha: So if $n$ is odd, how can $\overline{2x} = \bar 0$?
@TedShifrin, I may be posting two math videos to YouTube soon, which is kind of cool =)
@heather: Tell me when you do :)
What's the contradiction you get if its Euler number is nonzero?
Lol, went on the phone with my parents and now the chat has gotten pretty active
I kinda want to start making Mathematica animations for youtube.
20:19
@TedShifrin, will do =) I'm kind of excited.
@heather have you seen my projective geometry animations?
@ZachHauk, probably not
what's your Youtube channel?
Oh shit, those are on my Linux partition.
Let me see if they're in my discord messages
I've only seen what you put in here, Zach.
@TedShifrin Ahh, I see it! Thanks!
20:21
@MikeMiller The contradiction is that one the one hand it's supposed to be zero, but on the other hand it's not
@Danu I think he's basically making the same calculation we are.
Because if you mod out the gauge group from the configuration space, this Euler class is supposed to be a generator of the 2nd cohomology?
I'm confused... :P
So what I'm doing here is applying a linear transformation from one set of lines through a point, to another set of lines through some other points
It will always draw a conic section
Do you understand the proof I gave?
20:22
@Zach: You'll have to practice explaining this to people. (not quite linear transformation unless you explain how)
What you said is:
Since you mod out the free part, the result is still a nice closed, oriented manifold (no singularities) and therefore $\chi\equiv 0\mod 2$ but on the other hand $\chi=\pm 1$
@heather so, you know how if we have a linear transformation, we have that $T(c\vec{x}) = cT(\vec{x})$?
Is that right?
I'm not sure heather knows linear algebra ...
@ZachHauk, I believe so
20:24
I thought she said she did. Or was that someone ellse
My bad. I'll shaddup.
@TedShifrin I've read about it =) and learned a bit of it
@Danu What's the question? A space with a free S^1-action has Euler characteristic zero because Z/n is a subgroup of S^1 for all n; and free Z/n-action means $\chi$ is div by $n$ for all $n$ - only zero is possible. Now if you have a fixed point just remove it. $X - p$ has $\chi = 0$. Assume $X$ is a CW complex with a vertex $p$. Then $\chi(X) = \chi(X - p) + \chi(p) = 1$.
though i cannot profess to "know" it.
Oh, look. It's a @Balarka!
20:25
@heather Ok, so because of that
I hope what I said works
@BalarkaSen Too late to the party ;)
We'll take a set of vectors through the origin
Hi @Ted
Ted already spilled the beans
Thanks anyways
20:25
@Balarka: I was thinking about its making a circle bundle and using multiplicativity of Euler characteristic for fiber bundles.
That is, we'll take a vector $\vec{x}$ and consider all vectors of the form $c\vec{x}$ for $c$ is real not equal to 0
That would make a "line", right?
@TedShifrin Well, X - p is a circle bundle, you mean. Sure, works.
@Danu Ah, I see why he doesn't want to do this argument.
i believe so, yes
ah, yes, it would.
this is coming back to me =)
He almost certainly hasn't oriented the moduli spaces yet.
20:26
@heather Now, let's say when we take the output $T(\vec{x})$, it gives us some $\vec{y}$
okay
And so each $T(c\vec{x})$ gives $c\vec{y}$
right
so, now we have another line, right?
20:27
"Lines" through the origin map to other lines
@MikeMiller Yeah... It's a bit weird. We never proved orientability, but then at some point when it came to relaxing the orientability assumption he was like "yeah well it's oriented after all so we won't drop this assumption"
Or something like that
So what I did here
Was consider some "line" through a point, which we'll consider the origin
@Danu My argument is a little easier than Ted's, but yeah, I was busy.
then put it through some linear transformation, which will give out another line through that point, right?
right, yeah
20:28
But, what we're gonna do after that
is move that line we got outputted
to some other point
In this case, I moved it like 5 units to the left
hmm, okay
And so, what I do is rotate the input line, so that every single input is given to the transformation
and then, each output is a line, which is moved 5 to the right
Now, we consider the set of intersection points between these two lines
Which happens to always make a conic section
Because of some cool linear algebra stuff
@Balarka: I suppose you're allowed to be busy aside from us.
why are parabolas in standard position referred to being y^2 = 4ax with focus(a,0) why isn't it written as y = 2(sqrt(ax))
@WDUK: Cuz that only gives you half.
20:31
ohh of course
Standard position is like diagonalizing a quadratic form
oops :P
@BalarkaSen The divisibility by $n$ thing is nice. You just view it as an $n$-sheeted covering I guess and then the argument is sort of the same as Ted's :P
out of curiosity, what did you use to make the animation? mathematica?
At least I think
20:31
@Danu Ya, Ted's is the bundle-version. Mine is the covering space version.
:-)
Thanks for the input
I used Processing @heather
No problem.
Euler characteristic is such a great invariant
totally man
20:32
Balarka's is more elementary, I grant. Proving Euler characteristic is multiplicative for a fiber bundle is more nontrivial.
@MeowMix, huh, I've never heard of that
Neither had I
@Ted I don't actually know how to do it other than a spectral sequence argument.
Also, yeah, I don't like being busy with stuff other than math and other garbage I like to spend time with but such is life
I don't know how to do it :\
I know that argument, too, but why can't be do a triangulation argument just like for covering spaces.?
20:33
@TedShifrin Oh this looks like it relates to a problem in the textbook. Where you said "why does a constant angle always make a circle?"
@TedShifrin Whaddaya mean?
Oh, you've changed your name back, Zach? ... I didn't say that; I asked what you got.
In the covering space case you can sort of literally think of $n$ copies of the base
@TedShifrin I think lifting triangulations is harder for bundles.
You get, like a prism above a triangle
@Balarka: Yeah, must be. I'll have to ponder that.
@Danu: That Hirzebruch/Serre/Chern paper we discussed was the analogous thing for signature.
20:35
@MeowMix it's a very nice animation, I must say
@TedShifrin Right...
Damn spectral sequences
Thanks... I wish I could share the source code, but I'd have to go plundering my hard drive (And by that I mean shut down and boot Linux instead of Win10)
ah. well, it's not that big a deal =)
@Danu: I actually had to use a spectral sequence argument in my thesis to get cohomology of flag manifolds.
I secretly want to learn to be proficient at spectral sequences but I haven't had a great motivation to do that yet. Maybe I'd want to look for a nearby 1st year grad workshop on homotopy theory which covers Serre's theorem.
20:39
@heather Also, here is an animation showing what I was discussing earlier
The fact that lines through the origin map to other lines through the origin with a linear transformation
The red vectors are the basis vectors of this transformation
Don't confuse it with a 3d model of the plane :P
Because it kind of looks like it
I'm off to eat lunch. Back for a bit later.
Adios @Ted
Have a nice lunch
Just use them when they're relevant.
Let $n\in\mathbb Z_{>0}$ and $a\in\mathbb Z$.
a) Prove $\overline a=\overline{-a}\iff \overline a=\overline0.$
b) Prove that
1. if $n$ is odd, then $\overline a=\overline{-a}\iff\overline a=\overline0,$
2. if $n$ is even, then $\overline a=\overline{-a}\iff\overline a=\overline0$ or $\overline a=\overline{n/2}.$
c) Prove that if $n>2$, then $\phi(n)$ is even.
I've shown a) and b). Here $\phi$ is the Euler function, defined by
$\phi(n)=\#\{a\in\mathbb Z\mid1\leq a\leq n, \gcd(a,n)=1\}$.
However, I have no clue how to use a) and b) to prove this. Could someone give me a hint?
20:44
Yeah, fair enough.
@Danu I suspect there's an argument we can use that doesn't involve the orientations. You only really need those when defining invariants.
@MeowMix, another good animation
Maybe I could try induction
Consider $\mathbb Nm\mathbb N$. Assume that for some $n\in\mathbb N$, it holds that $\phi(n)$ is even. Now consider $n+1$. We know that $\phi(n+1)=\phi(n)$ if $\gcd(n+1,m)\neq 1$, and $\phi(n+1)=\phi(n)+1$ if $\gcd(n+1,m)=1$. But I should be able to show that we can only add 2, and not 1. But I wouldn't know how to proceed:l
Hm, I'll post it in the forum, because I gotta leave now anyways
@MikeMiller What about doing everything mod 2?
I think the contradictions still work?
By the way @Semiclassical d'you see this: arxiv.org/abs/1703.02532
Non-orientable manifolds can have euler char one
20:50
mm
@ShaVuklia $\phi$ is multiplicative and any odd prime minus one is even.
That's kinda bizarre.
Pretty out there
@ShaVuklia If $\gcd(a,n) =1$ what do you know about $\gcd(n-a,n) =1$ ?
20:53
"Keywords: Helium-4, Helium-3, 4-Manifold, Elliptic Curve"
Not really sure what it's supposed to do.
huh, interesting
The first few paragraphs give some motivation
I find it hard to take seriously, if I'm honest.
I think most people agree on that.
21:00
but its atiyah
The ideas just seem too unmotivated
It also doesn't really suggest any obvious 'predictions'. For instance, suppose I construct the purported model of the hydrogen atom. Can I deduce the usual hydrogen spectrum in a sensible way?
Can I say something useful about the helium atom (the simplest two-electron atom) using it?
Now, those might be the wrong questions: they seem less interested in the electrons than in the nucleons.
If it says less than quantum mechanics already does, then I find it hard to believe I should be interested.
21:22
@MikeMiller All almost complex structures compatible with a given symplectic form induce isomorphic canonical bundles, right?
I think I learned that the space of compatible ACS's is contractible and I hope it should follow from that
Rehi @Ted
Rehi @Balarka
21:51
Balarka: Is H_3 still Omega_3?
I forget if that stops at 3 or 4.
Hi @Ted
And everyone else too of course
@MikeMiller Is that the de Rham chain group?
bordism
Oh.
It doesn't hold after 2
Hmm, wonder if I have a counterexample
But can you represent everything as a map out of a 3 manifold?
22:00
Yeah, that should be true. I think you can do that for any dimension less than 7?
It's getting an embedded nullhomology that's the real trouble
for reference eg Thom
I can't see how would one do it for a 3-manifold in a straightforward way. Issue is that K(Z, 3) is not a manifold so the standard arguments break
You probably also need a restriction on the dimension on you're ambient manifold. Probably at least 2*3 = 6 dimensional
Not if you allow disconnected 3-manifolds in the domain I guess
22:18
Just take a simplicial complex representing the homology class and surger out the singularities.
What I'm interested in now is whether you can represent every homology class as a map out of a simply connected thing
Ah, the singularities are all codimension 2
aka point in 3-manifolds
in dim > 2.
Right. Thanks.
1
Q: Finding the changing rate

Nick Pavini When a space shuttle is launched into space, an astronaut's body weight decreases until a state of weightlessness is achieved. The weight $W$ of a $150$-lb astronaut at an altitude of $x$ kilometers above sea level is given by $W=(\frac{6400}{6400+x})^2$. If the space shuttle is moving away fr...

having trouble if anyone would like to help
codimension 2 inside 3 is 1
aka graph
22:23
I am messing up my arithmetic again.
Yes, so now less obvious how to do that
This might be of interest.
Still by hand just less trivial
Oh, take a regular nbhd of the graph and replace it like S^1 x D^3 by S^2 x D^2
I think my problem for the question I asked is that I don't know how to do surgery on simplicial complexes.
@MikeMiller In H_3, you mean? It's false in H_2, as you know I guess
22:28
Yes
3 and higher
How about 2[CP^1]xS^1 in CP^2 x S^1? Can you represent that by an S^3? Doubt it
Not an $S^3$ but maybe e.g. a $\Sigma T^2$
Comment on this answer if you come up with a proof or counter example math.stackexchange.com/q/2181934/98602
Interesting
22:51
Balarka, how the hell do you know all this math at your age
I had a good idea but it failed :( Darn it
@Krijn I don't know how to handle these questions. I stick to it, maybe? I do not place importance over knowing as much as over being able to solve things
2
Even simple things like a good combinatorics problem, not just high-tech stuff
And I am bad at that. I have improved a bit. Trying to get better

« first day (2411 days earlier)      last day (2907 days later) »