« first day (2879 days earlier)      last day (2134 days later) » 

12:01 AM
@Daminark our ANT2 course is 80% homological algebra and group/Galois cohomology so far
 
Nice
If we do group cohomology we're probably gonna be a bit less homological
Since it'd probably take a good chunk of time to actually learn the homological algebra
 
modular forms are really nice, but depending on what you do, there might be some trick/ugly calculations. There are of course calculations which are tricky, but not ugly. For example, to compute a certain Fourier series, you just have to say "Okay, let's just randomly take the infinite partial fraction decomposition of the cotangent and take the derivative of that $n$-times, that's probably useful for this series that seems to have nothing to do with the cotangent"
but they have all the magic of complex analysis amplified by NT magic, so pretty damn magical
@Daminark I don't think you can really make a wrong choice here, everything sounds great
 
This is true, yeah
Might also ask my Galois prof (the advisor to both of my mentors), if only so that I can just pick a topic and go for it
Also he's the one giving the talks on modular forms and seeing him talk analysis is so off-putting
 
@Daminark I think the way you're going to define group cohomology is to take the Eilenberg-Maclane space of the group, then take the infinite symmetric product and then homotopy groups of that and then voilà, that's group cohomology for you
really easy to compute, too
 
I know it's something people in NT interact with a good bit but going from hearing him talk about finite Galois groups and then boom harmonic functions was just like, whoa
Ah perfect!
Lol I've been reading a bit out of an appendix in Silverman and it just does H^0 and H^1
 
12:11 AM
yeah, Silverman does just what he needs
 
Also one of my mentors recommended this paper called "A Cohomological Viewpoint on Elementary School Arithmetic"
 
that's really fun
 
Which talks about a certain easy case of H^2
 
I can explain it a bit to you if you want, but not now, it's really late here
I wonder how those kids understand arithmetic without group cohomology
5
 
Yeah don't inconvenience yourself, I've got things to read. I did already read the paper I mentioned, it was surprisingly good for having seemed like a joke
Right? If you don't understand cocycles how can you understand such examples?
 
12:17 AM
anyway, if you do group cohomology or ANT, I can try to help you with stuff since I know a bit about that. (And I'd like to hear what kind of stuff you do in any case)
 
For sure, and thanks!
 
Hello @MatheinBoulomenos, don't recall seeing you before :)
 
@MatheinBoulomenos I'm an idiot. $N(z) = z\overline z = zz^p = z^{p+1}$, so $N:\Bbb F_{p^2}^\times \to \Bbb F_p^\times$ is just $\Bbb Z/(p^2-1)\Bbb Z \to \Bbb Z/(p-1)\Bbb Z ~:~ 1 \mapsto p+1$ which is coprime to $p-1$, so the image is full, so it sends $p^2-1$ elements to $p-1$ elements, so the answer is $p+1$
 
Hello @SimplyBeautifulArt
 
For 15 minutes I forgot that the non-trivial element is Frobenius
 
12:20 AM
@LeakyNun correct
 
52 mins ago, by Drew Brady
all you need to know is the Frobenius map lol
no, that isn't 15 minutes, that half an hour
I feel like a complete idiot now
@MatheinBoulomenos I was trying to solve $p^2-1 \mid n^2-1$
I even thought of primitive root of unity
of course $n$ has to be $p$, it's the only canonical choice
it's the only choice that makes it a homomorphism
nvm
@MatheinBoulomenos anyway, what's the next question?
 
Here's the solution I expected: Let $G=\operatorname{Gal}(\Bbb{F}_{p^2}/\Bbb{F}_p)$ by construction, the cokernel of the norm is the 0-th Tate cohomology $\hat{H}^0(G,\Bbb{F}_{p^2})$, since $G$ is cyclic, periodicity of the Tate cohomology implies that $\hat{H}^0(G,\Bbb{F}_{p^2}) \cong \hat{H}^2(G,\Bbb{F}_{p^2})\cong H^2(G,\Bbb{F}_{p^2})$ Now the latter group is a relative Brauer group, so it classifies division algebras with center $\Bbb{F}_{p^2}$ and degree $2$.
But by Wedderburn's little theorem, all finite division rings are fields, so the group must be zero, thus the norm is surjective
@LeakyNun that was the hardest one
 
I'm laughing in front of my screen
where have I seen periodicity of Tate cohomology
 
I talked about it before
you can use it to prove that $\Bbb R, \Bbb C, \Bbb H$ are the only associative division algebras over $\Bbb R$
I mean, finite-dimensional
 
nvm, so what's the easier questions?
@MatheinBoulomenos I'll just memorize it and show off to my friends lmao
 
12:27 AM
let $\Bbb F_q$ be a finite field with $q$ elements, and $l$ be a divisor of $q$. How many elements does $\{x^l=x \mid x \in \Bbb{F}_q\}$ have?
 
$l$
 
you're kidding me
 
I'm not
 
except when $l=1$, in which case we all know how many there are
@MatheinBoulomenos still wrong?
 
12:29 AM
yes
 
yes as in
 
yes, still wrong
 
you must be kidding me
 
I'm not
 
$x^l - x$ is a non-zero polynomial that splits in $\Bbb F_q$
it has exactly $l$ roots
 
12:30 AM
are you sure about that?
 
0 is a root, and in the units group, $x \mapsto x^{l-1}$ is a map $\Bbb Z/(q-1)\Bbb Z \to \Bbb Z/(q-1)\Bbb Z$ sending $1$ to $l-1$, which has order $\frac{q-1}{l-1}$, so the kernel has size $l-1$, and including $0$ gives $l$ roots
you must be kidding me
oh
I'm kidding myself
that's embarrassing
 
Why is $\frac{q-1}{l-1}$ an integer?
 
yes, I realized
so the correct answer is $\gcd(q-1, l-1) + 1$
@MatheinBoulomenos am I right
 
I'd have to check if that's a valid way to write the solution
 
it isn't like there's a simpler way...
 
12:41 AM
yes, it seems right from a few test cases
 
what is your answer...
 
the way I wrote it was let $q=p^n$ and $l=p^k$, then the answer is $p^{\gcd(n,k)}$
 
oh
aha
 
There's a really nice proof, too
 
interesting
@MatheinBoulomenos wait, lemme think
ok nvm go ahead
I'm burning myself's time
and please post the remaining question
my own time. i can't english.
 
12:43 AM
@LeakyNun the first one was sorta split into two
 
go ahead
 
Hi can anyone verify my proof? I'd be so appreciative!
1
Q: Prove that a subset of $\mathbb{R}$ with positive outer measure contains almost a whole interval

Nicholas RobertsThis is exercise number 28 from chapter 1 of Stein & Shakarchi's Princeton Lectures in Analysis III: Real Analysis. Let $E$ be a subset of $\mathbb{R}$ with $m_*(E) > 0$, where $m_*(E)$ denotes the outer (or sometimes reffered to as the exterior) measure of $E$. Prove that for each $\alp...

 
Write $q=p^n$, $l=p^k$. Choose an algebraic closure of $\Bbb{F}_p$ and embed everything into that. Note that $\Bbb{F}_l$ is a splitting field of $x^l-x$ and consists of all element in $\overline{\Bbb{F}_p}$ such that $x^l=x$. This implies that the subset of elements $x$ in $\Bbb{F}_q$ that satisfy $x^l=x$ is actually $\Bbb{F}_l \cap \Bbb{F}_q= \Bbb{F}_{p^k} \cap \Bbb{F}_{p^n}$.
Now for every $m \in \Bbb N$, there's a unique finite field inside our chosen algebraic closure of degree $m$ over $\Bbb{F}_p$. Furthermore, we have $\Bbb{F}_{p^a} \subset \Bbb{F}_{p^b}$ iff $a \mid b$.
In other words, the poset of finite extesnsions of $\Bbb{F}_p$ under the partial order given by $K \leq L$ iff $K$ embeds into $L$ is isomorphic to $\Bbb N$ ordered by divisibility and we're talking the infimum here, so the gcd
@LeakyNun
 
1:24 AM
@MatheinBoulomenos I went to practise my presentation
and you know what, I spoke to the microphone for 36 minutes
took way more time than I have
now I only have 5.5 hours of sleep
 
@DrewBrady you still there :D can you continue helping me? , I can see that c^T x=0 iff c=0, but why do we have to suppose that?
If someone could help here please would be very appreciated math.stackexchange.com/questions/2827744/… ? It's about cone of tangents and a set of gradients multiplied by directions
 
Hello, everyone. I'm stuck on my PhD research but I feel like, if I share the problem, there's a chance I'll get scooped :/
 
scooped?
 
It's when someone published a result you would have done if you were quick enough.
*publishes
 
ah
then don't do it
If I were you I would not :P
 
1:33 AM
Yeah, I'm not going to. It sure is frustrating though.
It's about the product over the nth roots t of unity of f(t) for some polynomial f, where n and f are determined by a group. The product gives the order of the abelianisations of certain groups.
 
sounds like a combination of complex analysis and abstract algebra
haha
 
 
2 hours later…
3:16 AM
Good morning all!
 
3:30 AM
Good night ;p
 
@mercio: I have thought a bit about these "E" irreps. So over {\Bbb C} they are either 2-dimensional real or pairs of 1-dimensional complex (you agree)? For "us" they are just summarized because we actually always investigate hermitian (selfadjoined, there is as I know again some issue it this is the same, but lets say in our cases always) operators, that is with pure real eigenvalues.
So we do not notice the difference between the "two types" of E. What is essential is that the operators always commute with the all group elements. So for me all E can branch from one "degenerate state" (that is a "subspace" of the eigenvectorraum of the hermitian matrix) with a 2-dimensional wave function basis but one and the same eigenvalue into states with two (in the generic case) different eigenvalues. This is why my intuition comes closer to classifying groups over $\Bbb R$.
 
@SimplyBeautifulArt what's up?
 
3:47 AM
@mercio: So maybe you are right after all, that is not only best but also inevitable to mention that one works over $\Bbb R$.
 
@TheGreatDuck made fast growing functions with currying
 
ah cool
@SimplyBeautifulArt I broke the chain rule.
 
probably not
 
elaboration
I was examining operators similar to differentiation and couldn't find any useful ones that obeyed the chain rule. Frustration ensued.
 
3:50 AM
I'm not surprised
one cannot construct facsimiles of mathematical objects without breaking a few 'axioms'.
non-Euclidean geometry taught us that much
 
Well, if you're interested I'll show you a sneak at what I've recently done
 
@SimplyBeautifulArt "fast growing functions" scare the heck out of me ...
 
@Rudi_Birnbaum he has a contest where people attempt to make the functions with the fastest growth they can
he's like an expert at googology.
 
Jon
In Linear algebra what does $M \circ C $ mean if M and C are linear operators?
 
@Jon in general
\circ means composition
 
Jon
3:58 AM
Sure
 
that wasnt a question
i just split my post in half by accident
 
Jon
Thanks
 
@Jon however
i think circ also relates to something weird with tensor operations
you doing anything with cube shaped matrices called tensors?
 
@TheGreatDuck: I once had also a go in the buisness (for myself). Conway is only the starting point. Then you start a series of hillarious exeggerations to finally notice that all that ever comes out is about zero compared to infinity. Then you think you don't want to live forever ...
 
@Rudi_Birnbaum i prefer engaging in efforts to find useful diff. eq. methods
or rather
integration methods that may double as diff. eq. methods
 
Jon
4:01 AM
No
 
@TheGreatDuck: simple example?
 
@Jon then it's likely just composition
function composition
@Rudi_Birnbaum Suppose F is a continuous function such that there exists a function g(y,z) and a step function C(x) such that $F(x) = g(x,C(x))$ and $f(x) = \partial_y g(x,C(x))$ for some f. Then f integrates to F.
or rather
F is an integral of f
 
@TheGreatDuck: So whats the "doubling" part.
 
@Rudi_Birnbaum same thing but with a differential equation
or a solution anyways
it gets messy
 
4:06 AM
@TheGreatDuck sounds interesting
 
it is
one can even look at that operation as a sort of differentiation that has a set of derivative functions rather than a unique derivative funvtion
and that's nice and all
until one realizes the chain rule doesnt apply to it
which just results in complete infuriation and nothing makes sense anymore
probably just a matter of finding something similar to "step function" with the right closure requirements
or it's a pipe dream and the only sets to have such closures is the set of all functions and the set of constant functions
in which case i have to bite the bullet and eventually attempt to prove that method works without the chain rule
and for some reason i think that might be necessary.
ugh
<<<<<< hence my current avatar
 
hehe
When those functions do intergration and differentiation how can the chain-rule not be fulfilled?
-- I should leave now, am on some kind of road trip ... cu later
 
@Rudi_Birnbaum i mean the weird partial differentiation operation
cause it's not actually differentiation
it's a weird facsimile
it's more closely comparable to "anti-integration"
but this is assuming im using step function properly
it's the one thorn i cant seem to fix
 
@Rudi_Birnbaum >.>
@Rudi_Birnbaum that's why we stick strictly to the finite... or to the infinite. $\ddot\smile$
 
@SimplyBeautifulArt i will be posting a question in a minute or two
perhaps you'll have insights with your analysis knowledge
 
4:19 AM
👀
Eh
It's past 12
 
it will only be a few minutes
and it's not
 
But past 12
 
it's 11:20
just wait 5 minutes
 
It's past 12 here
 
poleeease
 
4:20 AM
Also No Game No Life takes higher priority
🤷‍♂️
 
@SimplyBeautifulArt meh
life is overrated
supposedly
i havent heard any actual conclusive stuff on that yet
:-P
and that's just a joke on your statement
not a suicide joke
 
:p
@TheGreatDuck you ever did function currying?
Also good night!
 
@SimplyBeautifulArt not in math. Just in Haskell and only briefly
like
maybe a couple days on it
@SimplyBeautifulArt think you can answer this? math.stackexchange.com/questions/2828059/…
 
 
4 hours later…
8:11 AM
Does anyone know any interesting facts about the kernel of the comultiplication in a {Hopf algebra, coalgebra}?
 
 
1 hour later…
9:26 AM
What are polynomial one forms?
 
context ?
 
 
1 hour later…
Zee
10:40 AM
@Daminark your sleep schedule is quite impressive
 
Good morning chat
 
yoo
can i ask your opinion about something
 
Mine? Sure
 
i got accepted to continue to grad school for theoretical math to my undergrad university
i also applied for another one in Athens
in athens it will be harder and with 3 more courses
but in my uni i know the proffesors etc
its a safe space
i know ill do well
and that i can get a good grade
in Athens i got nothing in my side f. Ill start from scratch but the lvl will be better.
 
10:56 AM
That can be a hard decision to make. On one hand your success might be more of a sure thing if you stay, but on the other, you lose an opportunity to network with a different body of people.
 
where would you go?
yes exactly
 
Personally, I would probably stay. But that's because I am personally more predisposed to familiarity than I am to change. I think the better decision in the long run might be to go to the Athens one.
More networking, possibly higher level--I'll let you speak for that, I don't know--and you still do get to maintain contact with old professors. Speaking from experience they tend to like former students communicating with them.
 
ive got friends that are grad students already in my uni and would be fun to have the same courses without pressure
 
That's true too.
I mean, either way, you're getting a graduate level mathematics education.
 
haha
thats true
 
11:00 AM
@Zee lmao yeah it's a total mess. Also hey everyone!
 
Heya @daminark
 
what do you say @Daminark
 
I'm still reading the last couple messages since I just barely got on, but say, my school tends to strongly recommend that students go elsewhere
 
I've noticed it's almost policy, at least among many universities here in the States, not to accept grad students from their own undergrad.
 
11:02 AM
Familiarity is definitely comforting which I can understand, but I guess moving is sorta seen as better by virtue of exposing you to new types of folk
For example, in my school, we don't really have anybody doing stuff like operator/C* algebras, so that's a subject whose existence you're liable to never hear about if you just stay. And so on
 
I think another part of it is wanting to prevent conflicts of interest with acceptance to grad programs, and additionally the colleges in question wanting more people to have been in contact with them in total (if all your grad students were undergrads, you don't really have any new students through grad programs, but if all your grad students come from elsewhere, then every grad student represents a new person added to the university community).
 
For what it's worth, I don't think either choice will necessarily hurt you.
 
Also, in America at least, grades in grad school feel like less of a thing
You need to pass everything and then your focus should be on research
So to that end I've been told that the most important choice in that regard is to make sure that you'll have a really good advisor
 
11:10 AM
ohh
 
11:21 AM
pff really tough one
2 proffs want me to stay..
they are really good to me here
 
If you think it'll be good for you, that's ultimately your judgment call to make.
 
0
Q: Show $\lim \frac{ n^{n-2} + n^{n-1} + n^n}{(n-2)^n + (n-1)^n + n^n} = \frac{e^2}{1 + e + e^2}$

mickShow that $$\lim_{n \to \infty} \frac{ n^{n-2} + n^{n-1} + n^n}{(n-2)^n + (n-1)^n + n^n} = \frac{e^2}{1 + e + e^2}$$ Does l’hopital work here ? Taking log ?

 
Yeah keep in mind that most of what I said above was hearsay, like I've still got a year of undergrad left. If you really have good reason to believe that in your particular case, one decision is better than the other, then go for it, even if a few general principles here and there suggest otherwise
 
Any ideas ? Looks simple ?!
 
thanks people!!
dami you are an undergrad?
 
11:37 AM
End of page four
They defined a polynomial mapping between banach spaces in page 3. However, I don't quite understand the definition, since $D^kp(y)$ lives in the banach space U and the tensor power lives in the tensor product of the banach space U. How is the multiplication between them defined?
 
what multiplication ?
 
between $D^kp (y)$ and the quotient in the definition of $p(x)$
polynomial function
in the beginning of page 3
 
maybe $D^kp(y)$ lives in $V \otimes U^{*k}$ ?
 
And still the multiplication
Is it understood as tensor product?
 
I guess so ?
 
11:46 AM
the quotient lives obviously in $V$ tensored k
 
$p$ is from $V$ to $U$
oh i got it backwards lol
 
what branch would you classify the above paper -Geometry-analysis?
 
It is rough path theory
To solve ODE
 
I have never heard of a rough path theory before
 
driven by paths with p-finite variation
It is quite new. It formulates a generalization of Riemann-Stieltjes integral to p-finite variation functions.
instead of bounded variation.
So one tries to find class of integrands, which can compensate the roughness in the integrator. This leads to the notion of smooth 1-forms.
and such 1-forms can be approximate by polynomial forms. Thats the context.
 
11:57 AM
guys
'For example, a linear regression model fits the problem when you’re trying to understand how two points are related.'
what.the.fuck.
is this satire?
 
12:10 PM
@ManolisLyviakis yup
Hey @mercio!
 
oh i definitely know how to 'build my data science team' if i know how to rub two points together
 
And @quallenjäger
 
Hi @Daminark
 
@PeterSheldrick how many points do you want?
:P
 
i think even with some low number > 2 a combinatorial approach is better
 
12:15 PM
not statistical?
 
well combinatorial is still statistics but linear regession is kinda convex optimization imo
 
hmm
thanks for the link, by the way
:-)
hi @TheGreatDuck
 
hey @AkivaWeinberger
 
1:14 PM
Hey
 
1:57 PM
@PeterSheldrick well...for dummies, I guess? :/
 
mfw you learn a trick online and wish your professor taught it to you
finding the first few terms of the power series of 1/polynomial
 
What trick do you have in mind?
 
or 1/powerseries for that matter
long division
 

« first day (2879 days earlier)      last day (2134 days later) »