« first day (2855 days earlier)      last day (2463 days later) » 

00:00
any open set is a countable union of open intervals
I don't know, ask Balarka
I want a handle-able basis, not a list of all opens.
Don't involve me in this
Nor me.
@LeakyNun I think that a topology should be specificed by an algebra over the ultrafilter monad on the double category Rel of sets, functions and relations
I withdraw my comments. goes back to being invisible
00:01
Leaky always thinks about messy stuff and traps me into saying something wrong when I try to interact
I ain't playing with you anymore
Does that include me too?
@LeakyNun change my mind about specifying topologies as algebras over the ultrafilter monad on the double category of sets, functions and relations
00:02
@MatheinBoulomenos I actually want understand what you mean
I won't change your mind
I like algebra and category
I have no idea
Pollution is time-irreversible
but it's a remark
I can't change polluted minds
00:04
who is Rel?
Hi again yall
Any link from mathein is worth looking at :D
mathein I made a folder after you =p
@KasmirKhaan trust me, you won't want to go there
many things u said , some I understood and some I hope I understand later ><
@LeakyNun haha why ?
00:06
@KasmirKhaan I'm mostly trolling right now
@KasmirKhaan it's a defecation-hole
Haha ><
ncatlab is an elaborate joke
Okay I wont go there
0
Q: If a = b, does cos(a) = cos(b) too?

Mohammad Areeb SiddiquiI know that if $a = b$ then $ln(a) = ln(b)$. Is this true for any function? Or atleast the trignometric? If yes, then how?

00:08
@MohammadAreebSiddiqui it is true for any function
that's part of the definition of a function
@MohammadAreebSiddiqui also defintion of the "="sign
i just realized
How about the mlab tho?
how stupid
that question was
saying a=b means they are the same, there is condition equality such as
2x=3
that is true only for x=3/2
00:09
@Daminark that's the higher analog of a joke, it's an $(\infty,1)$-joke
@Mathei I don't know, but it wouldn't surprise me. The proof isn't difficult (you use the resolution given by a CW decomposition of the sphere, and patch it together infinitely many times to make a free resolution; the same thing happens in the negative direction. The periodicity is clear.)
and there is equal sign, that is true for all =p
@KasmirKhaan yeap XDDD idk why i did that
apologies
I asked / still ask worst questions :D so no need to be sorry :D @MohammadAreebSiddiqui
00:12
@KasmirKhaan is it?
Can someone remind me what the formal duality of homotopy fiber and homotopy cofiber is? If $f : X \to Y$, let the homotopy fiber be $F_f$ and the homotopy cofiber be $C_f$. Is the homotopy fiber of $Y \hookrightarrow C_f$ homotopy equivalent to $F_f$ (and vice versa)? I think so.
On that note, the homotopy cofiber of the covering map $S^n \to S^n/G$ is the $(n+1)$-skeleton of (a CW model of) $BG$, so there is definitely something along the lines of what I suggested going on there.
I guess my hiding is working pretty well.
@MikeMiller while we're at it, let's prove Wedderburn's little theorem. If $\Bbb F_{q^n}/\Bbb F_q$ is an extension of finite fields, then the norm map is given by $x \mapsto x^{1+q+\dots+q^{n-1}}=x^{\frac{q^n-1}{q-1}}$. So the kernel of the norm on multiplicative groups is the subset of elements which satisfy $x^{\frac{q^n-1}{q-1}}=1$, of which there can be at most ${\frac{q^n-1}{q-1}}$.
But $\Bbb F_{q^n}^\times$ has $q^n-1$ elements and $\Bbb F_q^\times$ has $q-1$ elements, so the norm $\Bbb F_{q^n}^\times \to \Bbb F_q$ is surjective.
@TedShifrin Ted :D
Maybe not so well.
00:18
lol
@TedShifrin I idont know how you did it but i really did not notice you are here ><
If my statement about the duality is true, the homotopy fiber of $S^n/G \hookrightarrow BG^{(n+1)} \hookrightarrow BG$ would be homotopy equivalent to $S^n$, as I claimed.
Kasmir: Your days of yearning for my knowledge are done. You've moved on ...
@Balarka I dunno. I would write EG as S^{infty n}.
If and when you get back to geometry, Balarka, we can talk again :P
00:19
@TedShifrin haha thansk but no ! :D
Then EG has a sort of built-in periodicity map.
I am too tired/unclever right now to know if your idea works
@BalarkaSen so if you want to be very vague, you could say that the reason why all finite division rings are commutative is that $\Bbb Z/n\Bbb Z$ acts freely on $S^1$
@MikeMiller No worries. I'll ask you to explain your idea in more detail tomorrow.
I have to sleep too
@BalarkaSen do you?
Haha
I like that irritating vaguery
00:22
@TedShifrin I hope that I shall be back at it, and therefore that we could have some nontrivial mathematical conversations. I wrote something as a self-reminder here that I actually hadn't thought of before, on a geometric note.
If you want to check it out
@MatheinBoulomenos Algebra still sucks my dude
@BalarkaSen right now, algebra and topology seem to be kind of symbiotic
I do want to sit down and understand your argument though
It seems really interesting
not sure where you're coming from
But I don't know how one patches up cohomology and homology (in +ve, -ve degree) togather to get the Tate cohomology. I suspect it's not hard, just that I have never worked it out before
With those on my to-do list, I shall depart to sleep
Hi. Is the test used here alternating series test? If they were using the alternating test, they needed to show $A_n$ is decreasing, but there's no mention about that. Also at the end it says consequently $\lim_{n\rightarrow \infty} a_n=0$ which itself does not imply that $\sum a_n$ is convergent.
00:25
@Balarka, I don't recall that argument.
At least algebraically, the fact that your group is finite is crucial for patching together cohomology and homology, so how is this used in the topological picture? @MikeMiller
@LeylaAlkan better work in a non-archimdean setting, then $\lim_{n\rightarrow \infty} a_n=0$ does imply that $\sum a_n$ converges
@LeylaAlkan you can ignore the guy above me ^^
@LeylaAlkan you can show that A(n+1)/A(n) < 1
to show $A_n$ is decreasing?
yes
you just need to multiply both sides by A(n) to get the desired conclusion
But I think they are using another method here, aren't they?
00:35
@Leyla: I don't think we're talking about the series $\sum a_n$. Aren't they doing ratio test here?
This is the question, we were considering the case when $x=-3/2$
And it's about the convergence of the series
Oh, then it appears there is something garbage-y.
OK, so the ratio test on this series fails because the limit ratio is $1$.
@LeylaAlkan it's the alternating series test
@Leaky is right.
But did they check the sequence is decreasing?
I just told her how to prove it
00:43
No. but it's indeed.
Then Leaky wins. I don't like that text much.
How can we guarantee that there exists $c$ such that $\int_a^b[\alpha(x)-c]dx=0$
@ManeeshNarayanan by letting $c = \frac1{b-a} \int_a^b \alpha(x) \ \mathrm dx$
They defined it for you in the first line.
$c = \dfrac1{b-a}\displaystyle\int_a^b \alpha(x)\,dx$.
Never mind .:)
Leaky beat me because I had to type displaystyle.
00:50
whatever
what was the name of that? fundmental lemma of calculus of variations? I remember we proved some version for locally integrable functions as an analysis exercise
Thank you @TedShifrin and @LeakyNun
Yeah, @Mathein, basically it's nothing but the fact that orthogonality to all vectors makes you $0$.
@TedShifrin the proof wasn't as easy you make it sound
00:53
@MatheinBoulomenos it is, the proof is "including yourself"
I was talking about the version we proved
It basically is, @Mathein. In the land of continuous functions, you need to know that $\int_a^b f^2(x)\,dx = 0$ means $f=0$.
Maybe there's a boundary condition here or there. No big deal.
okay, of course that's trivial
Well, I've certainly taught this in the context of calculus of variations ... with boundary conditions and adjoint boundary conditions.
hmm, the proof of dominated convergence is interesting
pointwise convergence surely does not imply uniform convergence, but it can be approximated to be uniform in some sense
00:56
Lusin's theorem?
hmm
in the sense that for any $\varepsilon$, if $A_n$ is the set of points whose image is trapped between $\varepsilon$ of the limit beyond $n$ iterations, then the measure of $A_n$ goes to the whole domain
I'm still trying to visualize what this means
Of course analysis is easy with strong regularity assumptions, the version we proved was this: let $f \in L^1_{loc}(\Bbb R^n)$ such that $\displaystyle \int_{\Bbb R^n}f \varphi d \lambda^n = 0$ for all $\varphi \in C^\infty_c(\Bbb R^n)$, then $f(x)=0$ almost everywhere
this wasn't quite as easy
Yeah, for anything I've ever taught, I didn't need that generality.
But if $f\ne 0$ on a set of positive measure, you can find a simple function that's nonzero near it on that interval and then choose $\phi$ appropriately.
you have to worry about the sign of $f$, too
no continuity assumed
But you can use all the density theorems ... continuous function near $f$ ... or Lusin's theorem or whatever.
01:01
it's cutting away a cofinite set again
I see this technique a lot
I'm not going to get into a technical real analysis contest with you. My playing with that stuff is 40+ years old.
@TedShifrin I believe you that you could solve it, of course
Silly you. :P
but choosing $\phi$ appropriatly meant a few pages of calculations with convolutions and dominated convergence
That seems a bit too much.
And most calculus of variations is in some smoothness/Sobolev setting, so this is a bit overly general, most likely.
01:03
I tend to write large and spaciously
Uh huh.
I was already asked if I hate trees
you hate a lot of things.
trees are the best
red black trees <3
I like trees where the root is on top and the leaves are at the bottom
01:04
Basically, if $f_n \to f$ pointwise, then for any given epsilon, the set of points whose image is further than epsilon away from the function has arbitrarily small measure as $n$ gets large
it's hard to wrap my head around this
because the convergence surely is not uniform
so we can cut away a set with unbounded image but arbitrarily small measure, and then we are left with a set with bounded image
this is very interesting
This reminds me of the Baire category stuff about the pointwise limit of a sequence of continuous functions.
so guys im getting a degree in computer science and mathematics, what job should I am for?
proof assistants
We're not the ones to advise you on that, @Mozzie. Many of my advisees who did that went into consulting.
i think I'm overqualified to be the president so lets shoot a little lower
01:07
LOL
I'm about to cry over that one.
everyone is qualified to be the president these days
if Trump can do it, everyone can do it
you guys ruined it by actually answering
damn it
I honestly have no backup plan if I can't get something in academia
@MatheinBoulomenos me neither
@Mozzie can't help, we type too fast
@Mathein: I don't think we should chase everyone out of academic pursuits. But you definitely need talent and determination. I think you have both.
01:08
I often type way faster than my brain thinks
what ever was still funny haha
which usually leads to tragedy
no comment @Leaky
no comment is a comment
@Ted thanks
01:09
@MatheinBoulomenos you would also need determinant
(can't stop myself misreading that)
@LeakyNun yes, that's why you need twisted tensor products
linear algebra is fake math
Um, not.
Who is this charlatan?
t r i g g e r e d
like what the hell is an eigenvalue
01:10
linear algebra is module theory over boring rings
agger
what IS an eigenvalue
IS
I can see that @Mozzie is going to be unemployed.
when we destroy isis, it will be known as waswas
lolz
its a number duhhh
01:11
6 mins ago, by Leaky Nun
Basically, if $f_n \to f$ pointwise, then for any given epsilon, the set of points whose image is further than epsilon away from the function has arbitrarily small measure as $n$ gets large
if we assume that the map is non-singular, then an eigenvalue is a fixed point of the projectivization
can anyone convince me that this is true
(I can prove that this is true, but I'm not convinced)
$f_n : [a,b] \to \Bbb R$
@MatheinBoulomenos that's undecidable
what is a computer scientist doing with maths
the reals do not have decidable equality
the two subjects inherently contradict each other
ummm
no
haha
01:13
lol
why do you need $\Bbb R$ to do math?
linear algebra then
theoretical computer science is literally all discrete math and combinatorics...
Let $k$ be a field and $V$ be a vector space over $k$
computer graphics is all linear algebra...
01:14
it's all a contradiction
besides all you maths people needs to get away from matlab and start using C :)
hi @Secret
not so with your presence
I never used matlab
I use magma and maybe GAP
java is my first language
I learnt it since like when I was 10 (maybe below)
01:15
what if you want your formulas to come to life
your going to program a real time system
i don't think so
@MatheinBoulomenos can you convince me
@Mozzie C is completely useless for the stuff that I need programming for
like?
I like Python
python is good
01:17
but C is closer to the ground so it is faster
i like it for prototyping
python is very slow
Computing Galois groups, class groups, (normal, characteristic etc.) subgroups, computing Fourier coefficients of modular forms, characteristic polynomials of Hecke operators etc,
at first you will not need C to do that kind of stuff
but if its gets implemented into some sort of real time application
your not going to use mathematica
this stuff doesn't get implemented in a real time application
this chat is a defecation-hole of productivity
01:19
no kind of embedded system needs to compute Galois groups ...
I haven't used mathematica, either
there's no programming language like fasm
the man who developed cryptography, a mathematician, said his work was useless...
Doing arithmetic geometry in C is like writing a modern 3d video game in assembly
3
Where C is the programming language, not the field
you wouldn't start in C
@MatheinBoulomenos lol
01:23
btw.... all modern 3d video games break down to assembly
:D
everything breaks down to machine language
but that's not the point
whatever, i enjoy math so it doesn't bother me
01:49
I didn't understand how this conclusion(higlighted one) made. Does anybody know this? Btw, T stands for the gamma function
@LeakyNun do you have any idea?
and another similar solution
 
2 hours later…
03:27
Is a 4 connected planar graph easy to find the existence of a hamiltonian path
04:25
@Daminark sup
How's it going?
pretty good
my talk is in a few hours
So I'm going over the material again
And for you?
Reviewing for my combo final, then my life is basically gonna be pure number theory/Galois for a little while
sounds like heaven
Unless you have to do some ridiculous stuff like that exercise ...
"So here's a polynomial where the Galois group of the splitting field is the monster group. Compute all subgroups, corresponding fixed fields and for each intermediate field, compute a primitive element and its minimal polynomial"
(There are some papers which actually construct such a polynomial over $\Bbb Q$, it's still unknown if it exists for any finite group)
05:21
@Mathein hey sorry I was out, and lol hopefully that won't happen
I've got analytic number theory to learn for complex analysis, which is kinda cool but the actual practice of it is one of those things which seems more pleasant to see than to do
Then there's Galois theory which is hopefully gonna be more fun, and then my algebra prof has been teaching some number theory. He said to learn p-adics because he wants to do Kronecker-Weber next, seems like the path he's taking is to do it locally?
I never saw a purely global proof of Kronecker-Weber
Wait no, I think Hilbert's original argument was global
fun fact: the result is named after Kronecker and Weber, but the first complete proof was by Hilbert
But learning p-adics is a good idea, anyway
they're simpler both algebraically and analytically
This is true, yeah
He said for now he'd rather approach it more from the naive analytic viewpoint as expansions to the left instead of worrying about inverse limits of Z/p^nZ
Since it's a bit less rigid and you can get the idea
analysis over $\Bbb Q_p$ is too good to be true
$\sum_{n=0}^\infty a_n$ converges iff $a_n \to 0$ and you can rearrange as you wish
@MatheinBoulomenos, Is it true: A series $\sum_{n=0}^\infty a_n$ converges if $|a_{n+1}-a_n|<\alpha^n$ where $\alpha$ is a fixed real number in $(0,1)$.
@Silent no idea
05:31
ok
@Silent no, if you just fix $a_n$ to be a constant non-zero sequence then this breaks
If you also stipulate $a_n\to 0$, which I assume you meant... then hmm
Nope
I think $a_n = \frac{1}{n}$ breaks it
Right?
Or no maybe not actually hmm
@Daminark Wow! This was amazing.
$\frac{1}{n} - \frac{1}{n+1} = \frac{1}{n^2 + n}$
@Daminark now some weird geometry: every triangle in $\Bbb Q_p^n$ is isosceles. Every point in an an open ball is a midpoint. Two open balls of same radius are either disjoint or equal to each other. Every open ball is closed and vice versa
Good lord
05:37
balls around the origin are actually closed under addition and they are actually fractional ideals
What is $\Bbb Q_p$?
in particular the set of elements with $ \|x \| \leq 1$ is a clopen subring and it's called the set of p-adic integers. It's a local ring with maximal ideal consisting of elements with $\|x \| < 1$
@Silent it's the field of p-adic numbers
ok
@Daminark So, it seems like $a_n=\frac1n$ does not work as counterexample
@MatheinBoulomenos Where can i learn this stuff? Is this covered under abstract algebra?
It sounds amazing
@Silent it's more of a number theory thing
05:43
There's a really elementary treatment by Gouvea
I think it's called "p-adic numbers, an introduction" or something like that
It won't hurt to know what a metric space and a ring is
but it's really accessible
Thanks for that recommendation
they are absolutely essential in modern number theory
@Daminark do you want some super weird defintion of p-adic numbers?
more of a joke, tbh
Let's hear it
Okay, let's take the subgroup $\Bbb Z(p^\infty)$ (don't worry about the notation) of complex numbers $\Bbb C^\times$ where the order is finite and some power of $p$ (so actually a subgroup of $S^1$) Explicitly $\Bbb Z(p^\infty)= \{ e^{\frac{2\pi i k}{p^n}} \mid k,n \in \Bbb N\}$
that's an abelian group
if we have any abelian group $A$, then the set of group homomorphisms $\operatorname{Hom}_{\Bbb Z}(A,A)=\operatorname{End}_{\Bbb Z}(A)$ is a ring, where addition is pointwise addition and multiplication is composition
now $\operatorname{End}_{\Bbb Z}(\Bbb Z(p^\infty))$ turns out to be an integral domain (nothing about this is obvious, not even that it's commutative), so if we take the quotient field, we get $\Bbb Q_p$
Really intuitive, right?
I think that's the other approach my prof told me about? It seems to me like $\mathbb{Z}(p^{\infty})$ is gonna also end up being the (inverse limit? limit? colimit? something like that) of $\mathbb{Z}/p^n\mathbb{Z}$?
well, it is the colimit of $\Bbb Z/p^n\Bbb Z$, but it's actually impossible to turn it into a ring itself (you just have an abelian group)
$\Bbb Z_p$ is the limit of $\Bbb Z/p^n\Bbb Z$ as rings
that's quite different actually
for $\Bbb Z_p$, you have the projection maps $\dots \to \Bbb Z/p^{n+1} \Bbb Z \to \Bbb Z/p^n\Bbb Z \to \dots \to \Bbb Z/p^2 \Bbb Z \to \Bbb Z/p\Bbb Z \to 0$
these are actually ring homomorphisms

« first day (2855 days earlier)      last day (2463 days later) »