« first day (27 days earlier)      last day (1556 days later) » 

05:53
Hi @Fargle @Daminark
howdy
Whatcha working on?
nothing at this exact moment
You're a third year student right?
Junior = third year?
How're you finding GP?
05:58
I'm a, uh, fifth year student
>_>
and G-P is pretty dang dank but I'm struggling with it
@Fargle Oh right :P. I just read your profile
It said 'going into junior', so I guessed
oh yeah I haven't updated that in forever
Gotcha
I too am a fifth year student I suppose
dank
what kinds of math do you gravitate toward
@Daminark How much of Aguilar have you read so far btw?
@Fargle Lie theory and AG I guess mostly
Trying to learn some algebraic number theory and some more category theory though
06:02
Lie stuff looks neato but I don't know if I have the background for it
I think Humphreys linear algebraic groups book is pretty good, if you know some AG
Get all your classical Lie groups in terms of the zariski topology
I'm unfortunately not well-versed in AG
Haven't gotten too far, I mostly used that book for Dold-Thom/quasifibrations
Do you have courses still as a 5th year student?
Only fairly recently (procrastinating for my final in complex analysis last month) did I actually start to learn some stuff like H groups and all that
06:07
@Daminark Gotcha, that stuff is pretty cool
I am just working through it now for the sake of appreciating fibrations and cofibrations properly
That makes sense. At this point my understanding of AT is a bunch of very vague stuff kinda piecing together, a lot of it without proof
Are you taking any (more?) official AT courses?
I can't remember if it was just the REU
Like, I sorta get how fibrations work ish and that's kinda the only way I know how to compute any homotopy groups
This fall I'm gonna hopefully finally take my first actual AT course
I sat in on an REU last year but was actually doing an analysis camp. Then I had a reading course but... If I can learn properly everything that happened by the time I finish grad school I'll be happy tbh
06:47
@Alex yeah I'm just bad at life
07:03
There's no such thing as good at life
At best there is a popular notion of such, which as a notion is itself a societal ill
18 hours ago, by Leaky Nun
If I let A be the ring of functions R->R differentiable at zero, then is m/m^2 a finite R-module?
07:25
@Fargle Don't most 5th and 6th year students (first two years of PhD normally I guess) have to do courses?
I haven't finished undergrad
What courses are you doing next (now?)
?
@BalarkaSen Have you started reading Katz my lad
07:51
I'm not sure, I don't totally have that worked out
depends on what the dept is offering
 
1 hour later…
09:12
@MikeMiller Ah yes
(I thought you'd acknowledge me sadface)
(and then I could acknowledge the book back)
@Alex I haven't, I was burnt out yesterday. I am better and more energetic today
@BalarkaSen Ahhhhhh
I had 2100% forgotten you told me this before
I remember now
Only because you told me to read that book!
I was just reading the book
RIP me
09:50
@BalarkaSen I just did the (really easy) first section exercises (for completeness), and read a few pages of section 2 so far.
Had to go elsewhere to recall the definition of intersection multiplicity properly.
10:20
@Alex OK, I'm here now. Let's see what the book does.
 
3 hours later…
13:32
How do you feel about it @BalarkaSen? It seems to leave out many details
Page 30 and 31 is cool to read though
13:53
You're powering through that book m8. I haven't gotten that far. This is my side reading!!
Hahaha, nah I just skipped a little ahead for motivation
I'm free now so let's read something togather
The word stack appears three times
I think it's supposed to leave out details. It's a general overview
I guess after reading this, one wants to read that 2000 page mirror symmetry writeup :P
13:59
Starting Chap 2
Have you (seen) proved Bezouts theorem?
yesterday, by Leaky Nun
If I let A be the ring of functions R->R differentiable at zero, then is m/m^2 a finite R-module?
I think they don't form a local ring unless I take the germ at zero.
So let's take the germ at zero.
@Alex Yep. Cup product my dude
@BalarkaSen Proof by cup product qed
I haven't seen such a proof :P
Is that the proper intersection theory way?
Here's the point. Let's say $C_1, C_2 \subset \Bbb{CP}^2$ are two projective algebraic curves of degree $m$ and $n$ respectively.
14:01
Yep
@Alex (5000th message in the garbological chat)
Lemma: If $C \subset \Bbb{CP}^2$ is a projective algebraic curve of degree $d$, it's fundamental class $[C] \in H_2(\Bbb{CP}^2)$ is precisely $d[\Bbb{CP}^1]$.
@MatheinBoulomenos hi
hast du ahnung fur meine Frage?
From the lemma (which I'll prove in a bit), the theorem is easy enough. Since $[C_1] = m[\Bbb{CP}^1]$ and $[C_2] = n[\Bbb{CP}^1]$, $PD[C_1] = m\alpha$ and $PD[C_2] = n\alpha$ where $\alpha$ is the generator of $H^*(\Bbb{CP}^2) = \Bbb Z[\alpha]/(\alpha^3)$ and $PD$ = Poincare Dual
$PD[C_1] \smile PD[C_2] = mn\alpha^2 \in H^2(\Bbb{CP}^2)$.
14:07
That's some hot fire
By intersection theory, if $C_1$ and $C_2$ are in general position in $\Bbb{CP}^2$ - which you can always do, as any two submanifolds of a manifold can be made transverse by a $\delta$-small isotopy - then $C_1 \pitchfork C_2 = \text{PD}(PD[C_1] \smile PD[C_2]) = mn \in H_0(\Bbb{CP}^2)$.
So the transverse intersection of $C_1$ and $C_2$ consists of $mn$ points.
If the intersection is NOT transverse, then you have to deal with multiplicity, which is strictly an issue of the algebraic category.
it is very nice to know that the proof is counting how many points $d$ lines intersect with $e$ lines :p
@loch i'm back
What is this $\delta$-small isotopy thingo?
@LeakyNun I don't really want to think about it unless you convince me that this ring is interesting
14:09
ok
@Alex If $M, N \subset W$ are two submanifolds of a manifold $W$, then there is a $\delta$-small homotopy $h_t$ of the inclusion $i : N \hookrightarrow W$ (so $h_0 = i$) through embeddings such that $h_1(N)$ is transverse to $M$.
In particular, $h_1(N) \cap M$ is a submanifold of $W$.
That's by topological intersection theory
Ok
@loch That's basically the same proof as mine, unwrapped, actually. See the lemma above.
yes of course
$d[\Bbb{CP}^1]$ is homologous to $d$ lines in $\Bbb{CP}^2$
14:13
@BalarkaSen Is there notational abuse occurring in the last math mode?
@Alex $\text{PD} = {PD}^{-1}$ :3
Ok
@BalarkaSen in one definition of the tangent space of a manifold, it is defined as the equivalence class of C^1 curves.
my question is whether we can replace "C^1" with "differentiable"
Good question, I have never thought of that. Ask Ted.
I saw the proof for Bezout just by computing resultants
14:15
Where did you find this beast proof?
but I didn't delve into the multiplicities stuff, that was just an elementary seminar
so we left it as an inequality
Yeah I've seen that bounding by resultants before
Haven't seen this chad proof though
I dunno, it's a folklore proof. I realized it when I encountered Bezout first (I'm a bum so I learnt cohomology before Bezout's theorem)
Ofc you did :P
There is merit to the multiplicity deal though. This topological proof doesn't resolve it.
14:18
Gotta see a physics proof
@Alex if you want a proof for $\mathbb{P}^2$ only, done algebraically (divisors), it is in chapter 5 of hartshorne (main point = to construct the intersection product). Then there is a slightly more generalised version in vakil (intersecting divisors), and you can also find the proof of the chow ring of $\mathbb{P}^n$ in 3264 and all that (if you believe the moving lemma)
@loch 3264?
"3264 and all that " by harris and eisenbud
Uh
there's a cohomological proof over arbitrary algebraically closed base fields with étale cohomology (our étale prof mentioned that, but we didn't do it)
14:19
it's a book on intersection theory full of examples
I'd never heard of that :O
I don't know how it works
but you have Poincare duality for étale cohomology
This room is great
it's great! a LOT of examples (and i have to work through them too..)
I really need to keep a list of these books somewhere
14:21
Here's the algebraic proof. Say $C_1, C_2 \subset \Bbb{CP}^2$ are given by $f = 0$ and $g = 0$ respectively (these are homogeneous polynomials in $k[X, Y, Z]$). Let $g'$ be another homogeneous polynomial in $k[X, Y, Z]$ of the same degree as $g$. Then $g/g'$ is a rational function on $C_1$.
oh and hi @LeakyNun
hi
@loch ok our time differ by 12 hours
Since $C_1$ is a Riemann surface, extending over poles you get a map $g/g' : C_1 \to \Bbb{CP}^1$. Degree of $g/g'$ can be computed in two ways: preimage of $0 \in \Bbb{CP}^1$ and preimage of $\infty \in \Bbb{CP}^1$.
The first corresponds to the number of zeroes of $g$, the second corresponds to the number of zeroes of $g'$.
So $\#$ of zeroes of $g$ = $\#$ of zeroes of $g'$ in $C_1$.
$\#$ of zeroes of $g$ on $C_1$ is exactly $\#(C_1 \cap C_2)$.
14:24
@BalarkaSen that proof is pretty cool, although it also only works over $\Bbb C$
@MatheinBoulomenos Nah, I'm giving a proof that works for arbitrary char 0 fields, and is basically that proof.
You have to work with Chow groups instead of singular cohomology.
I see
I think characteristic p you need étale cohomology or K-theory or some crazy shit
So let's say $C_2' = \{g = 0\} \subset \Bbb{CP}^2$. Then $\#(C_1 \cap C_2) = \#(C_1 \cap C_2')$ by what I discussed.
We had no restriction on what $g'$ can be except of the same degree as $g$ (which is $n$). So let's say $g'(X, Y, Z) = (aX + bY + cZ)^n$.
Similarly replace $C_1$ by $C_1'$ by replacing $f$ by $f'(X, Y, Z) = (dX + eY + fZ)^m$.
Then you're left to count the intersection-multiplicity of $(aX + bY + cZ)^n = 0$ and $(dX + eY + fZ)^m = 0$ in $\Bbb{CP}^2$.
Those intersect at zero with multiplicity $mn$, by definition of what multiplicity of zeroes mean in polynomials (and in hol. functions)
@BalarkaSen Did you point out above how to show that an order n zero gives n zeroes when perturbed?
@MikeMiller Nope, do point it out!
14:29
@BalarkaSen $g$ should be $g'$ here I guess?
Yup, thanks
Wow nice
I'm showing that the intersection-multiplicity $\#(C_1 \cap C_2)$ only depends on the degree of $C_1$ and $C_2$.
The key point lies of course in showing that a meromorphic function on a Riemann surface has number of zeroes upto multiplicity equal to number of poles upto multiplicity
Where multiplicity has a defined meaning because every meromorphic function on a Riemann surface looks like $z^n$ locally for some integer $n \in \Bbb Z$
Yeah I think I saw that in Forster
Yup!
Taking the moment to pay respect to this song
14:38
See page 80 of Katz @BalarkaSen at the bottom
Yeah that's the de Rham way to do this
note that Balarka gave a proof of "Algebraic Bezout", which says that the intersection number defined algebraically of a deg d, e curve is de
Sure
Bezout itself says that the intersection number may be computed as # of intersection pts counted with multiplicity
One should prove that you can perturb an order n zero to n positive zeroes
(Mike is saying something important)
14:41
this uses that these are algebraic/holomorphic curves in an essential way
If I do a quote in italics, how do I format a quote inside a quote?
Total nonsequitur:
Chow theory works in arbitrary fields (no alg.closed assumption/characteristc assumption necessary) - and just from pure glancing i think moving lemma only needs alg closed (but im not 100% sure on this) :p

The reference for this is intersection theory by Fulton - who (along with macpherson) developed a way of doing intersection theory without using moving lemma, but rather with vector bundles.

I've never seen etale cohomology in the intersection theory context - but on number theory I have seen things about Arakelov theory - which ive read is an arithmetic version of intersection theory
Enumerate the positive zeroes to $\tan x = x$ as $\lambda_n$. Prove that $$\sum_{n=1}^\infty \frac{1}{\lambda_n^2} = \frac{1}{10}$$
4
(The content of the proof I gave, if I recall correctly, is that I am really looking at the Chow group $\mathcal{A}^1(\Bbb{CP}^2)$ whose elements are divisors modulo proper equivalence? I forget what it's called. "Homologous" divisors are coming from a 1-parameter (algebraic) family of divisors interpolating them. There is a cup product $\mathcal{A}^*(X) \times \mathcal{A}^*(X) \to \mathcal{A}^*(X)$ given by algebraic intersection)
14:47
@loch "However, prior to the development of etale cohomology for curves over fields of characteristic p, one had to use algebraic methods to prove Bezout's theorem [...]"
that's int the handbook of K-theory
I shall sleep now. Thanks for this insight @BalarkaSen @loch @MikeMiller @MatheinBoulomenos
Yep. That is also the Picard group (well - some people define this as the group of line bundles - but we're find here because $\mathbb{P}^2$ is smooth).

Yeah - and here a 1-parameter family specifically means having a dominant map $f: X\rightarrow \mathbb{P}^1$ and saying that $f^{-1}(0) \sim f^{-1}(\infty)$ (rational equivalence)
@loch who is shimura?
maybe Chow theory is referred to as "algebraic methods", idk, but I was referring to cohomological proofs
14:48
Ah right rational equivalence, not proper equivalence.
hm yeah i mean i guess i know notinhg about etale cohomology to say more
I learnt a tiny bit about intersection theory from Chapter 1 Fulton
I really loved it
@LeakyNun number theorist
@loch what is a shimura variety?
idk
it's a question ive been wanting to know - but all i hear is it's pretty involved
lol
14:50
My officemate knows
I will not ask though
@loch do you need resolution of singularities for Chow theory?
@BalarkaSen yeah! fulton's book is quite hard in general though. my advisor says that if you think reading vakil makes reading hartshorne easier, then reading fulton makes you think hartshorne = vakil
@MatheinBoulomenos no
resln of singularities is only proved for char 0 too!
Like there are any fields other than C
Hironaka recently (a year ago) released a paper announcing it for arbitrary characteristics
Not sure if you know this
quote from Vakil "Alarming fact: This ring is
apparently not known to be commutative in general, because the argument requires resolution
of singularities. (It is known to be commutative in characteristic 0, and for smooth
things in positive characteristic, and a few more things.) I think it should be possible to
show that the ring is commutative in general using technology not available when this
theory was first developed, using Johan de Jong’s “alteration theorem” in positive characteristic.
@BalarkaSen I heard it, but is it verified?
14:52
Not sure of the state of the matter
ah - ok im assuming im only talking about smooth varieties then
I thought nobody believed
Apparently de Jong's theory is usually good enough to avoid resolution of singularities
I'm gonna sleep now, I will be interested if anyone likes my pinned problem / has ideas!
It is exercise 0.44(a) in s book by Aubin which title I have forgotten
in general if you have $X$ non-smooth - then $A^*(X)$ is defined differently and i don't know that well about it (something something ch20 of fulton)

(in the smooth case you can just define $A^i(X) = A_{n-i}(X)$, so everything is very geometric - and you have an intersection product which is what you think it should be )
by 20 i meant 17
I challenge u all to have proof when I wake
At which time I will have stopped being ChrisSis
15:07
"I am become Ramanujan" - Mike atm
 
2 hours later…
17:29
but actually, one doesn't need all the chow group machinery i was talking about if you're just looking for bezout for $\mathbb{P}^2$ (for any characteristic), or more generally to do intersection theory on surfaces - that is already developed in chapter 5 of hartshorne (with alg.closed assumption)!

He proves that for $X$ a projective surface, there is a unique pairing $Div X \times Div X \rightarrow \mathbb{Z}$ satisfying the properties you want for an intersection product - most importantly that if $C,D$ are two curves intersecting transversally, then $C\cdot D = \# (C\cap D)$ - and this
17:45
@loch I see
anyway - the only reason i remember that things work for finite characteristic because i remember for some enumerative problem the answer is different for characteristic $2$ lol
I wasn't claiming that you can't do it without étale, but just that there's no cohomological proof in characteristic p without étale
to be honest i dont really know what people refer to when they say 'cohomological methods' in general (without context) lol

but yes looking for an analogue of singular cohomology then probably requires etale cohomology - if by cohomological methods we're asking for 'cohomology' to show up in general
@loch tbh, I didn't work through the details, I was just going by remarks like the one in the K-theory book I quoted
i would be interested in knowing about how etale coh shows up in intersection theory!

but maybe only when i can speak etale cohomology..
17:53
I tried to take the class (it was offered by a prof who's an expert in etale homotopy), but I dropped it since it was too diffcult (I didn't even took alg geo before that)
I still managed to get the rough idea of the construction, but the details are beyond me
i remember reading someone's advice (probably matthew emerton) in comments on one of tao's posts on his blog who suggested something along the lines of there's no need to understand all the details of etale coh from the beginning - but just enough (probably for the general picture) so that one can understand its applications
The thing about etale cohomology I was amazed by was that it satisfies the Lefschetz hyperplane theorem
Hi @Ted!
Hi a @Balarka
the thing about etale cohomology to me that's amazing is that it exists and does what it does
the problem was that the applications require (among other things) some knowledge of the objects you want to apply it to, like say you need to know some stuff about projective curves if you want to compute the etale cohomology of that (in addition to Galois cohomology, interestingly)
Hi @Ted
17:57
hi @Mathein, @loch
@loch Meh I am too geometric to like it for what it is
hello @TedShifrin
But the point is it has some nontrivial geometric content to it already
You home safe and sound, a @Balarka?
Yup, back home!
17:59
the construction is just hardcore algebra and some categories. the only geometric input you need is the notion of an étale morphism. the whole theory of topoi was created as a by-product
Hey everyone
Is there any "key idea" underlying it's construction?
I guess I haven't actually encountered anything in math which is just 100% technicality to me
hi Demonark
How's it going?
the following probably contains some mistakes- but i think the idea of its construction is that you have something that's like the usual notion of topology (being a grothendieck topology)- enough for you to talk about sheaves

and then you can use things like constant sheaves / locally constant sheaves to define how you would for singular cohomology using sheaf cohomology

at least i would guess that's the construction.. (no idea on how you'd prove the properties though)
18:08
@BalarkaSen I think the key technical insight is that the only information that matters from a topological space for sheaves on that are the inclusion relations between open sets and the information when an open set is covered by some other open sets.
On a more geometric level, the problem with the Zariski topology is that it is too coarse, so we would like to introduce a "topology" with more open sets. One observation is that for any open immersion is an étale morphism, but étale morphism is generally weaker. If we basically construct a topology that forces the étale morphisms to be "local
I know the Grothendieck topology. But it's not clear to me why it is useful.
it's used because zariski topology is crap and people (i would guess grothendieck) realised that to talk about sheaves you don't need topology in the usual sense - only the data which defines a grothendieck topology
maybe the idea for the étale topology is that a holomorphic map between compact Riemann surfaces is a local isomorphism iff it is unramified, so maybe that's the inspiration for wanting a topology such that smooth unramified morphisms (= étale morphisms) become local isomorphisms
If you do the same construction in the topological category, do you get back something like the Cech cohomology of a topological space?
Or say something like the category whose objects are maps from simplices to $X$, and morphisms are commuting triangles where maps between simplices are the linear morphisms (composition of collapse/face maps)
So you'd declare singular simplices to be the open sets in your Grothendieck topology (sorry for the noobish writing; my understand of these things is mostly superficial)
Does running your abstract stuff give me back singular cohomology?
18:18
I haven't thought about that and I don't know the answer right now
Mmkay. Thanks
I'll think about it
given a topological space, if you define the category where its objects are open sets, and morphisms are open inclusions - then i think the answer is yes. After all the construction in etale coh is just sheaf cohomology

so i think it should be the same here too
i don't know about the category you described there though
I might seriously learn some sheaf theory
in relation to intersection theory stuff
One easier example that one can work it is that this gives a sheaf-theoretic description of group cohomology
so if $G$ is a discrete group, then consider the category of $G$-sets with equivariant maps as morphisms. Then we say a covering family is a surjective family of equivariant maps (surjective family meaning that the union of the images is the (common) codomain), then one can show that abelian sheaves on that correspond to $G$-modules and sheaf cohomology works out to be the same as group cohomology
18:30
i don't think you need a lot of "sheaf theory" for intersection theory - depending on what you mean by sheaf theory lol (and maybe also depending on what you mean by intersection theory)

if one is familiar with how sheaves work in AG (which is not too involved) - then one should be fine in reading e.g. 3264& all that / fulton's intersection theory (basically the only two references i know on intersection theory lol)
@loch I mean Goresky-MacPherson's perverse stuff
@MatheinBoulomenos that's pretty nice
oh
too many intersection theory out there
then yes that's more involved and idk anything about them :(
@MatheinBoulomenos Huh that's pretty cool
@loch Learn w me
18:35
something something intersection cohomology
not really on my list atm :p
i think my friend who does geom. rep theory has to learn these things (perverse stuff)
i dont really know how they show up though
I want to learn the perverse stuff eventually
but exams are not too far away and I have a couple of projects going on
Surely because GIT quotients are stratified
there's a book on proving the Weil conjectures using (among other things) perverse sheaves
18:38
Which book is that?
one of the authors is a prof here in Heidelberg and I'm on good terms with him and he likes to give long spontanous talks when he's asked something, so I could try to ask him stuff about that
That'd be pretty cool!
Give me some time to learn perverse sheaves though
I don't know the first thing about them yet
something something abelian category sitting inside the derived category of constructible sheaves..
18:41
okay, when you have a precise question, then you can tell me and I can pass it along
hopefully I can reproduce what he answers, lol
That'd be very cool, thanks a lot
oh and our topology prof works with singular spaces, he offered seminars on intersection homology before, I'm sure he can answer stuff as well
Wow cool
This is basically my reading list
RIP me
and a classmate of mine from the local CFT course I'm taking right now participated in that seminar on intersection homology, so I might ask him first if I don't want to bother the profs
that's a big list
18:46
I have been reading into the h principles stuff but not the stratified stuff
18:56
@loch might you have a reference for GCFT?
as in, the statement itself
the counterpart of LCFT
i think you should've tagged @MatheinBoulomenos instead

but one reference i know is andrew sutherland's notes on number theory I 18.785
@BalarkaSen maybe the section "Singular cohomology via the h'
-topology" starting on p.92 in this: http://home.mathematik.uni-freiburg.de/arithgeom/preprints/buch/springer-v5.pdf is a bit like what you have in mind
I don't really follow argument as I don't have any idea what a hypercover is, but the argument that this Grothendieck topology gives you singular cohomology as sheaf cohomology seems reasonably short
Thanks, I'll look into it later. I know you can set up an isomorphism of singular cohomology with the $S^\bullet$-valued Cech cohomology where $S^\bullet$ is the complex-valued presheaf sending $U \mapsto \bigoplus S^*(U)$ to the singular complex.
Something involving that perhaps
@LeakyNun chapter 6 in Neukrich
there are different formulations, one is ideal-theoretic and the other idele-theoretic
19:37
It seems none is interested in the question Mike asked (good, field's clear for me), so I'll write my epic solution here.
Suppose $f : \Bbb C \to \Bbb C$ is an entire function with simple zeroes $\lambda_i \neq 0$ and order of vanishing $n$ at $z = 0$. Take some other entire function $g$ with precisely those simple zeroes and order of vanishing at $z = 0$. $f/g$ is a nonzero entire function, so $f/g = e^h$ for some entire function $h$ by holomorphic lifting along the covering map $\exp : \Bbb C \to \Bbb C \setminus \{0\}$.
If $f$ has simple zeroes at $\lambda_i \neq 0$, then one usually constructs, as in the proof of Weierstrass factorization theorem, $$g(z) = z^n \prod_i E_{a_i}(z/\lambda_i)\;\text{where}\; E_{a_i}(z):=(1 - z) \exp (z + z^2/2 + \cdots + z^{a_i}/a_i)$$
We need to pick $\{a_i\}$ appropriately so that the product converges to an entire function. Notice that $E_{a_i}(z) := \exp(\log(1 - z) + z + z^2 +\cdots + z^{a_i}/a_i) = \exp(-\sum_{n > a_i} z^n/n)$. If $|z|< 1/2$, then the term inside the exponential is bounded by $2|z|^{a_i+1}$. Therefore $|1 - E_{a_i}(z)| = |1 - \exp(-\sum_{n > a_i} z^n/n)| \leq c|\sum_{n > a_i} z^n/n|<C|z|^{a_i + 1}$.
In particular, $|1 - E(z/\lambda_i)| \leq C |z|^{a_i+1}/|\lambda_i|^{a_i+1} = C |z/\lambda_i|^{a_i+1}$. If we can choose $a_i$ so that $\sum |\lambda_i|^{-1-a_i}$ converges we're done because by the simple lemma that $\prod (1 + a_n)$ converges whenever $\sum |a_n|$ converges.
Hadamard tightened what $g$ can be: Let's call the order of growth of $f$ to be $\rho = \inf \{r : |f(z)| \leq A\exp(B|z|^r) \; \forall z\in \Bbb C\}$. Then one can pick $h$ to be a polynomial of degree $\leq \lfloor \rho \rfloor$, according to the Hadamard factorization theorem.
@MikeMiller In our case, let's say $f(z) = \sin(z) - z\cos(z)$. This has order of growth $\rho = 1$. Notice that since $f$ is odd, the real zeroes of $f$ are $0, \pm \lambda_1, \pm \lambda_2, \cdots$.
I claim there are no non-real complex zeroes of $f$; this is just Rouche; consider the square $K = [-n\pi, n\pi]^2 \subset \Bbb C^2$. $|-z\cos(z)| < |\sin(z)|$ on $\partial K$, so $f$ has as many zeroes as $\sin$ in $\text{int} K$, which are exactly the integer multiples of $\pi$, and since there's a solution of $f(x) = 0$ between any two solutions of $\sin(x) = 0$ (except at $x = 0$, where those sets coincide), we're done by simply counting.
Since it's already given to me that $\sum \lambda_i^{-2}$ is finite (:p also easy to just prove by hand), I can take the constant sequence $a_i = 1$. The refined Weiestrass product theorem gives $$\sin(z) - z\cos(z) = e^{az+b} z^3 \prod_{k = 1}^\infty (1 + z/\lambda_k) e^{-z/\lambda_k}\prod_{k = 1}^\infty (1 - z/\lambda_k) e^{z/\lambda_k} = e^{az+b} z^3 \prod_{k = 1}^\infty (1 - z^2/\lambda_k^2)$$
The coefficient of $z^3$ in the Taylor expansion of $\sin(z) - z\cos(z)$ is $1/3$, and the coefficient of $z^4$ is $0$. Since the constant coefficient of the product term, when expanded out, is $1$, this forces $b = -\log(3)$ and $a = 0$.
Therefore $\displaystyle \sin(z) - z\cos(z) = z^3/3 \prod_{k = 1}^\infty (1 - z^2/\lambda_k^2)$
Compare coefficient of $z^5$ on both hands of this equality. One is, by writing down the Taylor expansion, $1/30$. The other is, by expanding out the product, $1/3 \sum_{k = 1}^\infty 1/\lambda_k^2$
$$\sum_{k = 1}^\infty \frac1{\lambda_k^2} = \frac3{30} = \frac1{10}$$
mic drop
(This is of course an emulation of Euler's proof of the Basel problem)
19:55
I suspect Mike has a more succint and clever proof in mind but I am admittedly not clever :)
I liked that I had to freshen up my memories with concrete complex analysis to do this
20:17
from modular forms, all my complex analysis knowledge has been overwritten by computations with Fourier series of modular forms and computing horrendous integrals wrt the hyperbolic measure on the upper half plane
we spent over two lectures basically computing integrals for one big proof
God that sounds painful
Hmm, so your class is doing modular forms, is it looking at it as an application of complex analysis (essentially, is it a class which at the core aims toward the analytic and is just including this as a fun topic?)
20:33
the methods are analytic, but we mention some connections with Riemann surfaces and the results are often number-theoretic
20:44
Ah
it's not as bad as it sounds, it's actually pretty cool just that one proof was really ugly
and it's not just including modular forms as a topic, it's a whole course on them
I think the prof studies modular forms just for their own sake
I see
@Daminark If you're doing group cohomology, you can discuss this stuff here, I like that stuff and I'm sure others here are interested as well
21:11
Yeah once I start getting to the cooler bit I'll definitely post about it some
21:48
@Balarka: And your reading list is missing my papers I assigned to you to read, too. :D
Hi @Ted
My plan for doing alg geo over the summer is this:
- start with Miranda's book on Riemann surfaces and Algebraic Curves, which starts in the analytic category and then gradually moves into the algebraic category
- proceed with Milne's notes on alg geo (http://www.jmilne.org/math/CourseNotes/ag.html) I like that he works with algebraic varieties which he defines as locally ringed spaces, so he's basically doing the modern formalism, but just over an algebraically closed field and only looking at closed points
does this sound good?
I want to prepare for the alg geo class which starts in the winter which has a reputation for being hard with lots of people dropping
@BalarkaSen This was EXACTLY my approach
Just from the starting point that someone once knew how too sum 1/n^2
Which is the zwroes of sin(pi x)
22:04
Hi @Adeek
Hi @MikeMiller
@MikeMiller oh
22:30
@BalarkaSen I guess nobody else has the IQ to compete with Chad proofs of this problem like ours, so they didn't comment
Not all of us have PhDs in animated adult television
LMAO
@TedShifrin Your paper on the Whitney umbrella gave me a very crucial insight on the Nash blowup of stratified spaces along their frontiers
More evidence that I should read your papers more!
"every time someone mentions Nash blowup my mind wanders to an extremely terrible joke about how Kaczynski blowup is totally different" - Fargle
2
oh no now it's public record
22:46
Unablowup
23:04
@BalarkaSen Holy shit lmao
I'll be here all week
23:23
Geocalc is just legit trolling right?
idk
i feel like he is a troll
The fact that he had no reaction to my counter trolling
23:40
@BalarkaSen The good place to start for sheaf theoretic inteesection homology is the book by Kirwan

« first day (27 days earlier)      last day (1556 days later) »