« first day (2415 days earlier)      last day (2902 days later) » 

19:00
i have to do some linear algebra ted-cercises
@Semiclassical something happened?
Eh, I'm just grumpy. Don't worry about it.
robjohn seems is not around. Good days here are gone.
"Nothing gold can stay" and all of that.
I find the days here good enough
19:09
nights are better
Depends where you live :p
@MeowMix Dude, that's like 4 years ago:P
the joke is the names
And @Don'tdisturb ; Eh, thanks for the compliment? I must admit it’s refreshing to hear, because my boyfriend keeps joking(?) how ugly I am:P Anyhow, there’s no shortage of girls at our mathematical faculty, so I’m surprised there’s not many more of us online - though it is true that the ‘die-hard’ mathematicians seem to be boys in the end, at least from what I see at university and online communities.
I personally find the gender differences slightly fascinating where it concerns mathematics(/science)
I know, @MeowMix
but it's still OLD :P
19:25
Learning Mathematical Induction is slowly killing me X__X
what are you having a problem with?
Nothing in particular. I'm just trying to learn from practice right now. The Inductive step is giving me most of the trouble.
19:50
@ShaVuklia Old times here are gone for me. Once I liked to talk about math (here). Perhaps its best utility today is for some non-math chatting.
Or to talk about math like "Wow, today is the Pi day".
@Don'tdisturb why don't you post a math problem then? :P
I think "old times" are gone by definition
@Don'tdisturb I can imagine that someone big like you might not find a lot of satisfaction in the chat, but I've personally learned a lot in the chat and I've been helped a lot with homework exercises
@Astyx hehe true
@Astyx Old (with the sense of nice, good) times.
19:53
I don't think we really need this negativity here, if I may
@Astyx You can ignore the users you don't like, you have a button for doing that I think.
does anyone have a cool math problem?
I never said I disliked anyone
@Meow Do you know Sperner's lemma ?
no I'm not smart
what is it? :P
It states that if you have a triangle subdivided into smaller triangles
19:56
@Astyx or you can ignore the users that say things you don't like. I'm sure you got the point.
Hey guys, where did the $(k+1)(k+1)!$ come from? Shouldn't it be $i*i!$?
@Dragneel you take out the last term of the rsum
and just add it yourself
Then if you color the vertex with three colors such that the vertex of the bigger triangle have different colors, the vertex lying on its edges are the color of one of the extremities of this edge, and the rest how you like
Yea, so $i=k+1$
Here we take out $k+1$, whose term is $(k+1)(k+1)!$
Also, $i$ isn't real (no pun intended). By that, I mean, it's just a "temporary variable" in that we use it just to evaluate the sum.
19:58
Then there exists a small triangle of which the vertex are of all three colors
Salut @Astyx. Hi, Zach, @ShaVuklia.
@Ted Salut
Hey @Teddo
Bon soir! @TedShifrin
@Don'tdistrub I don't see why I would
19:58
Or jour*, if you live at the other side of the world:p
I'm on one of the other sides, yes.
It's one whole word @ShaVuklia Bonjour and Bonsoir :)
Waking up at noontime is weird...
@MeowMix I see now. Thanks Meow!
Oh right, thanks!@Astyx
19:59
@Dragneel No prob.
@ShaVuklia I'm a bit big, I need to lose some weight. Your intuition is good. Chat can be pleasant, especially with more girls around. :-) I'm sure you got the help you needed.
@MeowMix Did you read what I said ?
Hey @TedShifrin, I've got a question
@Astyx sorry let me read it
Don't be, I should have tagged you
I can only blame myself
20:00
Zach, btw, if you're looking for some interesting problems, check out the UGA high school math competition website. Most of the past years' competitions.
> color the vertex

what do you mean? In the picture you showed each vertex has 1 color, not 3
Yes, Mr. Danu?
So I have the following situation
@Ted Akiva's topology questions were cool even though I don't know any topology
Each vertice takes one color
20:01
And, speaking of Akiva, Hi @Akiva
@MeowMix And you can choose between a set of three colors (in the picture, green, red and blue)
Hi @Akiva
Hi, DogAteMy :)
You guys still snowed under?
@Astyx: Une vertice = vertex :)
Alright
Oh is vertice the plural ? Damn I mixed it up :p
@ShaVuklia did you celebrate the $\pi$ day? For instance, I ate some slices of apple pie ($\pi$).
20:02
Vertices is the plural.
Hey @Ted!
It's all Latin's fault.
Hi Demonark :)
Oh wrt vertices
Ok I'm completely lost now
Oh, that's pretty interesting
Does that have to do with graph theory?
20:03
@Astyx: One vertex, thirteen vertices.
And elementary ! Try and prove it
It does, but you don't need to know graph theory to solve it
In the REU, a bunch of us made the mistake of talking about "a verticy" or "a matricy"
And got funny looks from Laci
I've got this space $E$ which fibers over $B$, $\pi:E\to B$ with fiber $\Bbb P^1$ and I want Pontryagin classes of the total space. I can probably get the classes of the base, and I can use that $TE\cong \pi^*(TB)\oplus T\pi$ where $T\pi$ is the vertical tangent bundle or whatever it's called. I know there is no torsion so I have $p(E)=p(B)p(T\pi)$. Now all I need to do is figure out $p(T\pi)$.
You'd get smacks from me, Demonark :P
morning
20:04
G'night, @MikeM.
@Ted thanks
I had a previous situation where my space happened to be a projectivized bundle and there I could use that I had a class which restricts to the hyperplane class on each fiber to understand the (in that case Chern) classes of the vertical bundle
Hi @MikeMiller
brings out Ted's multivariable math book
In this case, I don't really know how to proceed to determine $p(T\pi)$. Any thoughts?
20:05
this is not IRC :P
@Danu: I'm rusty, but $T\pi$ is still a complex bundle of rank $1$, so has only trivial Pontryagin classes.
@MeowMix Do you want a hint ? (Or do you even want to try and solve it now ? :p)
@TedShifrin Whatcha mean by trivial?
@Danu Zero.
20:07
Sorry @Astyx, I need to do some Linear Algebra. tell you what, I'll look at it later
What is $p_1$ of a complex vector bundle?
lololol
@Don'tdisturb Haha! I would have, but I'm not really impressed by the reason why they chose the date for $\pi$ day, because I don't like to approximate mathematical constants, so I skipped:d
my life
OK, besides that ridiculous oversight; how does this work in general?
(the other case I was working with had $\Bbb P^2$)
@MeowMix As you wish !
20:08
What is the general case you're working with?
Right, then the subtlety is that $T\pi$ isn't quite $TF$, because there's no projection to the fiber by which to pull back.
@Astyx: Tu vas mieux?
Un peu oui :)
J'en suis ravi :)
@MikeMiller Let's assume each fiber is some easy space like $\Bbb P^n$ or something else whose classes I all know
What I think I'm supposed to do is find the total space class which restricts to the Pontryagin class of each fiber
I see no reason to believe that will substantially help you.
20:10
@ShaVuklia what do you think of a negative pi day?
is that right?
OK, and you say there's no standard/easy way of doing that?
@ShaVuklia I wonder if it's fair to let other constants not celebrated like Euler's number, Euler-Mascheroni constant or Catalan's constant. I think you're right to some extent.
@Danu I certainly don't know how. But maybe that has more to do with me than the bundle.
20:11
Well, with a connection, we could put the curvature matrix in block form, I guess.
@Danu: Have you looked at Hirzebruch's book to see if it's in there?
@Don'tdisturb It might be harder to celebrate Euler-Mascheroni on the 0/57
@Astyx connections can always be found if one wants to celebrate anything
(it should be the last worry)
Hi pals
20:13
Hi @Semi (or rehi ?)
@TedShifrin Ehh, no
@skullpetrol hi
Hi @skull
@Don'tdisturb Yea, I don't know why $\pi$ is so popular. Maybe because it can be implemented in easy maths/science calculations at high school, so most people know about it? Euler's number comes much later in the curriculum. But that shouldn't be a measure for popularity/celebration, of course.
20:17
@MeowMix you asked for a cool problem. Prove that $$\int_{-k}^{\infty} \frac{a^x}{\Gamma(x+1)}\textrm{d}x+\int_0^{\infty} \frac{x^{k-1} e^{-a x}}{\pi^2+\log^2(x)}\left(\cos(\pi k)-\frac{1}{\pi}\sin(\pi k) \log(x)\right) \textrm{d}x=e^a, \ a>0, \ k\ge0$$
I believe he wanted linear algebra
@TedShifrin I haven't really.
@Astyx maybe he just changed his mind seeing the beauty in front of his eyes.
no thank you i suck at that
Who knows ?
20:19
Yes, Danu, I saw your answer. In general, I can only think of trying to give a difference SES involving the bundle. We might figure out transition functions to see how it's $TF$ twisting along the base. I'll have to ponder.
the first integral might not be too hard to evaluate
It seems he didn't :p
@TedShifrin So eh... Am I confused or is the Pontryagin class of a complex rank 1 bundle not always trivial (in fact, trivial iff the bundle itself is trivial)?
Hey I'm back, and yeah I imagine
@Ted
Welcome back
20:20
@Danu You are confused.
So $p_1(E) = c_2(E\otimes\Bbb C)$.
@TedShifrin $p_1(E)=-c_2(E_{\Bbb R}\otimes \Bbb C)=e^2(E)$?
@MikeMiller Am I?
The first time someone said "verticy", Laci said "what", happened 3-4 times before he was like "Oh, vertex"
But yeah, final today was one of those times where I actually had fun in a test
A complex line bundle is real rank 2---I don't think those have trivial Pontryagin class
What am I doing wrong?
20:24
You need an oriented rank 4 bundle to get nontrivial $p_1$.
@MeowMix Oh, such a disappointment, my heart hurt.
So what's wrong with $E_{\Bbb R}\otimes \Bbb C=E\oplus \bar E$, hence $c_2(E_{\Bbb R}\otimes \Bbb C)=-c_1(E)^2$?
I have forgotten soooo much :( ... For a complex bundle, there's a formula like $p_1(E) = c_2(E)-c_1^2(E)$ or somefin.
Your last equality is not true. $E \otimes \Bbb C$ is isomorphic as a complex vector bundle to $E \oplus \bar E$, and thus has $c_2(E \oplus \bar E)$
It was more theorems from class, but since I didn't have the proofs memorized, figuring them out was a bit of a puzzle
20:25
one second
@MikeMiller yeah, which is $-e^2(E)$, no?
hm? I've messed up here too
idk maybe complex analysis for the first one?
wtf guys
You're still wrong. But I'm confused.
20:25
I'm so confused
LOL ... I am allowed to be confuzled. I'm retired.
I'm pretty sure Kotschick also wrote down exactly this in my lecture notes
Are y'all misleading me?!
@Don'tdisturb Thanks, read it!
Demonark: Sometimes a little bit of memorization helps you get out of the blocks faster, but one shouldn't do math merely by memorizing.
Nah, you're right.
How irritating.
20:26
Are you sure?!
Yes, $p_1(L) = c_1(L)^2$.
@MikeMiller Oh, you didn't need to spell it out that you were looking to put me down :P
Maybe with a minus sign.
Yes, with a minus sign.
First was basically a subset of the proof of the spectral theorem on Hilbert spaces, second was proving that the range of a compact operator was closed if and only if the rank was finite, third was proving that the spectrum of a self-adjoint operator was real, fourth was about the space of convergent sequences, and fifth was proving that the set of $4\times 5$ matrices of rank $2$ was a manifold, and what its dimension was
No,
$p_1=-c_2$
the signs cancel
20:27
Calculate $$\lim_{x \to 0^{+}} \sum_{n=1}^{\infty} \frac{x}{1+n^2 x^2}$$
No they don't. You were right the first time.
$p_1(E)=-c_2(E\oplus \bar E)=-(-c_1^2(E))=c_1^2(E)$.
$c(E \oplus \bar E) = c(E)c(\bar E) = (1+c_1(E))(1-c_1(E)) = -c_1(E)^2$
Oh, sign conventions. Fine.
yeah, and $p_1=-c_2$
Oh, that's not a standard convention? I wasn't aware.
Of course $p_k = e^2$. Sorry.
I just don't pay attention to sign conventions until I have to.
20:29
Unfortunately on the last one, I never ended up figuring out why those equations were independent, so once again, on this test, I had to use the flimsy and prob incorrect point from my homework
Anyways, I take it you guys have no further input :P
Also, a couple slip ups here and there
Demonark: We discussed the matrices of constant rank problem. I told you the right way to approach that :)
@ShaVuklia welcome
@MeowMix I look at this limit and know the answer. $$\lim_{x \to 0^{+}} \sum_{n=1}^{\infty} \frac{x}{1+n^2 x^2}$$ How I did it?
20:33
taking the $x$ out of that sum would probably make it easier
which would be ok because distributive
@TedShifrin I missed this. Thanks for that.
Now that I think about it, in many situations where I was able to calculate I indeed did have another SES
@Astyx I don't see you. Where you are?
Yeah, and the part that I was supposed to figure out on my own was one that I only figured out a while after the pset was due, and I didn't remember how that worked, only the nonsense argument about how changing one entry shouldn't affect the others
Namely some kind of (relative?) Euler sequence for these projectivized bundles and the tautological line bundle involved
Oh, so you never figured out the right way to see that $f(A,B,C,D) = D-CA^{-1}B$ has maximal rank. :) @Demonark
20:36
Yeah
The key thing, Demonark, is the first term. The second mess is irrelephant.
@Don'tdisturb What do you mean ?
I know at one point I saw this curves argument that I convinced myself was true
No, you get the derivative surjective just by applying to vectors in the $D$ direction. The other stuff doesn't matter.
@Astyx I meant you were not visible to me anymore. But ,look, you reappeared.
20:38
I wasn't chatting, that might be why
@Astyx I think I have an idea to solve it
@Astyx nice to try $$\lim_{x \to 0^{+}} \sum_{n=1}^{\infty} \frac{x}{1+n^2 x^2}$$
(Just take tangent vectors of the form $(0,0,0,D')$.)
@TedShifrin hi, i need your help with a previous question of mine
$f(x) = x \ ^ t M x$ , where $M \in M_n$ is symmetric matrix
20:41
i need to prove $M$ has eigenvalue (not via linear algebra)
Damimark =?= Demonark @Daminark
I already said to you yesterday, @Liad, that you need to write down Lagrange multipliers.
You want to find extreme points of that function relative to the constraint $g(x)=\|x\|^2=1$.
ok , im trying to maximize $f$ under the constrain $|x| \ ^ 2 = 1$
so i defined the lagradiant :
snaps nice
$L(x,\lambda ) = f(x) + \lambda ( \sum_i x_i \ ^ 2 - 1)$
20:43
And yeah @skulldiesel
@TedShifrin you're with me so far ?
Yes, @Liad.
Hm, my book says that if we take an element $a\in\mathbb Z/n\mathbb Z$, then we can write $a=qn+r$ ($q\in\mathbb Z, 0\leq r<n$). However, I'm confused, because $\mathbb Z/n\mathbb Z$ doesn't consist of integers, yet its elements are entire sets of elements with the property that their remained by division of $n$ is the same?
Why does my book write this $a\in\mathbb Z/n\mathbb Z$ as an integer then?
Precisely correct, @ShaVuklia.
20:45
Because it's the same thing
@TedShifrin the differential of $f$ with respect to $x_i$ is something that i think is hard to work with
It doesn't matter how we represent the elements
They should NOT. They should write $\bar a\in\Bbb Z/n\Bbb Z$ and choose an $a$ in that equivalence class.
@Meow About Sperner's lemma ?
@Liad: So you need to compute that the gradient of $f(x)$ is precisely $2Mx$.
20:45
Ahh right @TedShifrin, thanks!
@ShaVuklia Do you understand why it's the same thing whether they write it as $a$ or $\overline{a}$? (Even though the latter is correct in the context of the group we're talking about)
@MeowMix It's not the same thing, see Ted's response!:)
No, it's really wrong to say $a$ in the quotient group and think of it as an integer. I made a huge deal about this teaching algebra.
I gotta go now, they're hushing me away from the library! Bye!
20:47
@Ted Why is it so wrong ?
Bye @ShaVuklia
@ShaVuklia see ya
I think there's a subtle mental shift between writing $\bar a=\bar b$ and $a\equiv b$, where in the former we have equality of sets of integers and in the latter we have equivalence of integers
Because $a$ is an integer and $\bar a$ is an infinite collection of integers (the equivalence class). One of my standard final exam proofs in ring theory was for then to prove that if $\langle a\rangle + I = R$ ($I$ a proper ideal), then $\bar a\in R/I$ was a unit. That problem tested understanding.
Not that it matters much
@Ted I get that, but I don't get why it's such a big deal as long as you know what you're doing, since the operations are compatible with the equivalence class (which I know you already know)
20:51
DogAteMy: In my algebra course, I started just with the latter, and only defined the ring $\Bbb Z_n$ after they were used to working with integers themselves.
@Astyx: Because students do not know what they're doing for the first semester!! :)
Makes sense.
And can be easily confused.
Oh, then sure, I agree with you :)
Also makes historical sense (which does not always equal or overlap with sense)
Often the Bourbaki-ing of mathematics (at the expense of history) sacrifices students' understanding.
20:53
Was it Euler that first introduced congruences ?
Oh, it must have been earlier.
I think Gauss?
Oh yeah Gauss
I suppose the $2^2\not\equiv2^{12}\pmod{10}$ thing might be one of the first examples where one realizes that care is needed
Or $2^0\not\equiv2^{10}$ for an easier-to-verify example, I guess
Students frequently get confused, yes, and try to use congruence of the exponents.
20:55
I'm out for some more research.
Bye @Don'tdisturb
You know how people generally treat numbers and such as urelements (i.e. not sets), in contrast with ZFC's framework?
That's how I normally think of the elements of $\Bbb Z_p$, I think
Not as sets, but not in $\Bbb Z$ either. Just their own weird thing
Lol, I love how Bourbaki is a verb now
20:57
Everything's a verb if you verb it.
I'm waiting for it to become an adverb
You can ad it to the list
(…he said Bourbaki-ly.)
And then a preposition
One "l" too many, DogAteMy?
I don't think its really necessary to look at how Leibniz or Newton thought about derivatives or how Heaviside thought about dirac deltas and (what would become) weak derivatives
These things are really cool and fun, but a direct presentation is more helpful when learning

« first day (2415 days earlier)      last day (2902 days later) »