« first day (3864 days earlier)      last day (51 days later) » 
00:00 - 21:0021:00 - 00:00

12:00 AM
@CroCo The matrix representation is the same. The left side is different, but the right side is the same. They are either wasteful or its a misprint.
 
@user2103480 so messed up because this one lulls you into believing it could be something normal, but it has so much random crap tossed around...
 
wym this is completely cryptic to me
 
@user2103480 Is that Hairer
 
Yes, some 2016 paper on regularity structures
 
basically just that the characters at least look similar to usual characters/notation. $Ker$ is in there. There is an integral and a product in there. some disjoint unions, stuff like that. the other thing uses so many nonstandard characters it's just sad to look at
but yeah, I won't pretend like it is any more reasonable
 
12:09 AM
his notation really is the worst
 
@user2103480 it's theoretical CS from a Cousot & Cousot paper
 
thanks sounds way more reasonable
 
Filthy casuals posting memes without proper sources
It's on page 22 of the pdf
 
based Alessandro
 
after some digging, it seems in extrinsic rotations (i.e. eq 2.63), the rotation matrices multiplication should be done starting from the most-left but in intrinsic rotations (i.e. eq 2.70) is the vice versa.
https://en.wikipedia.org/wiki/Euler_angles#Definition_by_intrinsic_rotations
Some thing the author ignored to mention it.
 
12:24 AM
I'm not sure what an intrinsic or extrinsic rotation is supposed to mean
 
@AlessandroCodenotti this
physics undergrad I think
 
of course the authors are french
 
@MikeMiller my understanding is that in extrinsic rotations, the rotating frame rotates relative to a fixed frame. In intrinsic rotations, the rotating frame rotates relative to a moving frame.
 
@Astyx looks like a fine paper though
 
If they are profs at ENS and Polytechnique, it means they're good
 
12:28 AM
Sorry, I don't understand the language. It seems like you have reached an understanding now though.
 
@MikeMiller
See this
This is extrinsic rotation if I understood it correctly
Notice ZA,YA,XA is fixed
^ this is intrinsic rotation.
 
It's ok, you don't have to explain it to me. As far as I can tell there's no mathematical content to this distinction.
 
There is actually in the matrices multiplications order. That is why I've asked.
 
But that doesn't mean there's not content to it in computer graphics.
 
I think there are different but yield same final rotation matrix.
they* are different
 
1:06 AM
if you get your hands dirty with atan2, is it possible to still call yourself a mathematician
i submit, no
 
 
2 hours later…
3:21 AM
@leslietownes you're not a mathematician unless you're doing langlands
 
Have you met him?
i submit, yes
 
@RyanUnger did you see the Scholze paper that came out recently
 
4:24 AM
i'm not doing langlands :(
 
@BigSocks of course
 
never even met the dude
i did meet VI Arnold once
 
@RyanUnger yeah I don't really know any of that but it looks impressive huh
 
5:14 AM
finally I am able to compute the exact values of Bernoulli numbers in python
 
congrats
 
Now I will try to compute the exact value of zeta(s) at even and negative integers
making a computer algebra system is hard
I am still not able to define a "symbol" without using third party libraries
 
I have used Python to create a collatz numbers to 1X10^20 but its all been done before... I also write simple programs.
 
there's an enormous amount of collatz stuff on math.se lately. people seem to have been bitten by the bug
 
collatz bug
 
5:19 AM
I have worked on it for 6 years to be convinced its true, looking at stopping time, Mod 2 Mod 3 Mod 4, I created a new notation, and my professors say I'm crazy. but you know... silly math problems.
 
i don't judge, i put some crap about collatz in my phd thesis. i think i have been inoculated against further investigations
 
I even have a visual representation of it using silos and shoots, But I don't have the computer expertise to make it come to life in a 3d model.
 
graphviz is a very useful package, i would love to see more of it. i don't know that it goes into the third dimension.
 
The cool thing about chasing a unsolved problem is how much it forces you to learn, I never knew Python, anything about number theory, or dynamics, chaos. It spurred the investigation process.
I also looked at binary strings, and the start of my mod investigation, I have filled 50 - 50MB excel sheets with calculations, they take about 5 mins to load, I all this stuff because I was curious about a problem.
 
@OneColdRuben That's great
I have never been sure why people find Collatz interesting
But if you are learning something from investigating / understanding things more, that's more than enough reason
 
5:41 AM
@MikeMiller I get why it’s curious at a glance: seemingly intractable despite being so easy to pose. I find it harder to understand why people -stay- obsessed with it
I mean, one must have a hobby. But it seems like a hobby with better prospects would he more appealing
 
Mathematics is puzzling
Mathematicians can proved super hard conjectures but the simple sounding collatz conjecture is still unproved
 
Yes, I am fond of assembling things, since then you can look afterwards
 
I like stuff I can compute
Makes it feel like progress :P
 
Still, all that matters is that your hobbies are fulfilling and enriching
And it sounds like that's the case here
So I can't give a hard time for that
 
Ya, fair
 
5:51 AM
is there a nice algorithm for algebraic simplification
 
that's pretty vague
 
I don't need an algorithm for very complicated expressions rn
 
 
4 hours later…
9:27 AM
Is My proof valid?: Let $R$ be PID and $(p),(q)\subset R$ be two distinct nonzero prime ideals. Then I want to show $(p)\not\subset (q)$ and $(q)\not\subset (p)$. But this follows from the fact that $(p)\subset(q)\iff q|p$ and R is UFD
I assume $R$ is not a field
 
you could be more explicit, but this works
 
Since $R$ is UFD and $p$ is prime, $p = uq$ where $u$ is unit. Hence $(p)=(q)$ which is a contradiction if we assume $(p)\subset (q)$.
 
you could be less explicit, but that works
 
9:55 AM
Any prime ideal in $R$ is principal, so by Krull's principal ideal theorem they all have height $\le 1$, so $\dim R = 1$ where $\dim$ is Krull dimension, so every non-zero prime ideal is maximal, so $(p)$ and $(q)$ are distinct maximal ideals, so they cannot contain each other.
probably don't need the step $\dim R = 1$
if $(p) < (q)$ then we can form a chain of prime ideals $(0) < (p) < (q)$, contradicting the fact that $(q)$ has height $\le 1$
 
Actually I wanted to show any PID which is not a field has Krull dimension 1
 
lol
accidentally proving your theorem
 
Any example of ring homomorphism from $\Bbb Q\to \Bbb Z$?
Want to see such homomorphism is injective
But it seems finding such map is not that easy
 
let p/q be in the kernel
 
I first wanted to see that map is injective but now I want to find a ring homomorphism from $Q\to Z$
 
10:07 AM
look at the image of 1/2
 
what do you mean
 
Assume you have such a homomorphism f. What is f(1/2) ?
or rather, what relation does f(1/2) satisfy?
 
f(1/2)= 1/f(2)
f(2) = 1 then
or -1..
 
I was thinking about 2f(1/2) = 1
 
Oh such ring homom does not exist
 
10:13 AM
right. stronger than that: there's not group hom
(other than the zero map ofc)
 
Then what would be an example of ring homom. such that $F\to R$
$R$ is commutative ring with unity
 
What's F ?
 
Field. sorry
 
That's too vague a question. You can always take F = R
 
I will let $R$ be a commutative ring with unity that is not a field
I want to see an example of $F$ and $R$ such that ring homom $F\to R$ exsits
 
10:20 AM
$F\to F[X]$
 
10:38 AM
any ring hom $f: \Bbb Q \to \Bbb Z$ must induce a group hom $\Bbb Q^\times = \Bbb Q \setminus \{0\} \to \Bbb Z^\times = \{-1,1\}$, which means $f(x) = \pm1$ for all $x \ne 0 \in \Bbb Q$, which means the image of $f$ is contained in $\{-1, 0, 1\} \subseteq \Bbb Z$, but no subset of that is a subring, contradiction.
 
11:35 AM
The moving sofa problem is the most important thing in mathematics
it can help you to shift your furniture
 
12:11 PM
@LeakyNun I was thinking of groups, and when I realized, I deleted.
 
12:27 PM
@robjohn probably a quick question, but is uniform convergence of an infinite product on K identical to saying the "partial products" are uniformly convergent on K?
I can't think up a better definition.
 
12:45 PM
Let $X,Y,Z$ be connected metric spaces and $f:X\to Y$ be an embedding and $g:Y\to Z$ be continuous such that the restriction map $g:\text{Im}(f)\to Z$ is a proper map. Is $g\circ f:X\to Z$ a proper map?
My guess is that $g\circ f$ is proper map.
 
@User sounds unlikely to me.
Can't you just do something like $B\hookrightarrow\mathbb R^n\to\mathbb R^n$ where $B$ is the open ball in $\mathbb R^n$, the first arrow the inclusion, and the second arrow the identity?
 
Obviously not, take X = Z = R and Y = R^2, with f(x) = (x,0) and g(x,y) = x.
A month or two into talking about properness I would have hoped you'd have some intuition about what that means? At the very least you need to develop the skill to think of (counter)examples whenever you have a question. Playing with examples builds more intuition than pure formality.
 
Admittedly, it's kind of a hard definition to wrap your head around, sometimes.
 
For sure, but if it's a major topic you're working with you have to get there eventually!
(The intuition here was: proper means "sequences going to infinity get sent to sequences going to infinity." Since Im(f) doesn't know most sequences going to infinity, I had the map be well-behaved on Im(f) but crush everything else down to Im(f) to destroy properness.)
 
I am sitting here struggling with uniform convergence stuff, so "eventually" seems to sometimes be down the line a little (only maybe if you are as bad as I am, I guess :P).
 
12:59 PM
think of delta epsilon as fixed box
I don't remember if delta or epsilon or both should be fixed
I am half brain dead after car accident
 
Probably I am the only person who talks here proper related things. Thanks, @Mike miller for your suggestion. I should play with counter example while writing my master thesis.
 
Most people here who have studied manifolds in a lot of detail have seen properness, since it is the appropriate condition to generalize many standard results beyond the assumption of compact domain. (Isotopy extension, well-definedness of intersection numbers, etc.) That's not to say that it's instantly understandable but one ought to work with examples for the sake of understanding it
I just tend to think that if I can produce a counterexample within 10 seconds you can produce a counterexample in some number of minutes if you attempt to!
 
I don't get your counterexample. Isn't $g\circ f$ the identity? Is the identity not proper?
Or was he asking about $g$ before editing?
 
Lol I'm being a dick and didn't even read the question correctly
 
$g\circ f$ is proper because an embedding gives a diffeo on the image (so the restriction is proper) and the composition of two proper maps is proper
 
1:12 PM
But then the answer is yes by pushing definitions? If g|_{Im(f)} is proper, then for K subset Z compact, g^-1(K) cap Im(f) is compact. Hence f^-1(g^-1(K)) = f^-1(g^-1(K) cap Im(f)) is a preimage of a compact set, and f is proper, so this is again compact.
 
Right
 
OK, so examples/counterexamples don't help here and my advice on that doesn't help so much, but it's still a definition push
 
that is ok. I asked this type of question before, as I am getting contradictory statements before while using proposition 10.1.17 and do not know whether it is right or not. google.co.in/books/edition/Topological_Methods_in_Group_Theory/…
 
1:42 PM
what do you call a biholomorphism s.t. $f(z)=f^{-1}(z)$
is it a "holomorphic automorphism?"
 
Read the proof. If you can understand all of the details then it is right. If you don't understand a detail pursue a counterexample. If that pursuit doesn't work try proving the detail yourself.
 
@geocalc33 involution
 
okay so it's the same
ty
so $f(z)=1/z$ is an automorphism of the punctured complex plane $\Bbb C^*$
 
2:26 PM
@anakhro That's what it means. Same for infinite sums.
 
2:37 PM
@robjohn thanks. That's what I thought (that if it is that way for infinite sums, then it should be that way for products).
 
It's about the convergence of the partial sums/products.
 
It's a pretty impressive theorem no ?
 
Yeah definitely
 
Although outright horrible
 
2:45 PM
@EdwardEvans is something like this something you can improve on? Like it's deg 25 in 26 variables, so can you possibly make 25 or 26 smaller?
 
I think thereÄ's one on 20 variables
 
Does the degree go up as a result?
 
Yeah, apparently some dude showed it's possible to do it on 10 variables but the degree hits like 10^45
The record for the lowest degree of such a polynomial is 5 (with 42 variables [JSWW76]), and the record for fewest variables is 10 (with degree about 1.6 x 10^45 [Matijasevic77]).
 
Wild.
Are any of these minimal, or still possibly something smaller?
 
I dunno, someone just posted this theorem on facebook and I thought it was wild lmao
 
2:50 PM
nUmBeR tHeOrY
 
literally
 
Q: what does alternating caps like that mean in general?
 
@EdwardEvans ferocious even
 
when you HaVe LUncH
 
3:00 PM
> It is typically used online to convey a mocking tone.
 
NuMber TheorY
if I capitalize less letters that means I'm mocking less
 
fEwEr
 
dIfFerEnce gEomEtRy
 
Is there an easy way of seeing that $e^{z/n}\to 1$ uniformly on $|z|\leq r$?
Of course, since it is on a compact disk, it feels like it has to be uniform.
Something like "for sufficiently large $n$, $1 - e^{z/n} \geq 0$ and then use some branch of the logarithm?
 
it's bounded between $e^{\pm R/n}$, no?
 
3:11 PM
That would seem to make sense.
Oh yeah, of course.
 
plug in taylor?
 
That's just continuity of exp at 0 no?
 
The amount of question marks make me think these might not be as "obvious".
 
$\|1-e^{z/n}\| \leq |z|/n \sum_{k \geq 0} \frac{(|z/n|)^k}{(k+1)!} \leq |z|/n e^{|z|/n}$
 
what's wrong with what I said?
 
3:15 PM
Thorgott just gave a complete argument, I'm not sure why people are continuing to go
 
@Thorgott Nothing in particular, I am joking because with question marks, everything seems to be voiced in a doubtful tone.
 
oh, I see
 
Do you see, Thorgott, or are you a blind mathematician.
Joining the ranks of Giroux and Pontryagin.
 
and the late euler
 
It converges uniformly to 1 on the box $|\text{Re}(z)| \leq r, |\text{Im}(z)| \leq r$, since $$e^{z/n} = e^{\text{Re}(z)/n} (\cos(\text{Im}(z)/n) + i \sin(\text{Im}(z)/n);$$ the first term is a real number bounded between $e^{-r/n}$ and $e^{r/n}$ and the second term is a unit complex number with argument in $[-r/n, r/n]$.
These both give you concrete estimates on how far each term is from 1, multiplying them you again get a uniform estimate (converging to 0 as n goes to infinity) about how far these values from 1.
This is Thorgott's argument adjusted to correct for the imaginary terms he forgot about (his argument only bounds magnitude)
 
3:26 PM
So possible concrete estimates are respectively $\leq |e^{-r/n} - 1|$ and $\leq 2$, right?
And then of course, the product converges to zero, and depends only on n so it's a successful uniform estimate.
 
$|zw - 1| = |zw - z - w + 1 + [z-1] + [w-1]| \leq |z-1||w-1| + |z-1| + |w-1|$ which explains why you need to show both of these terms go to 1
Then $|e^{\text{Re}(z)/n} - 1| \leq \text{max}(1 - e^{-r/n}, e^{r/n} - 1)$, I dunno which is bigger
But they both go to zero
And $|e^{i \text{Im}(z)/n} - 1| \leq |1-\cos(\text{Im}(z)/n)| + |\sin(\text{Im}(z)/n)| \leq 2/n$ or something.
Make it 4/n if you're paranoid.
Your whole expression is bounded above by something that looks like C/n.
The punchline is that you need to get your hands dirty but none of the individual estimates are very hard. If this looks like your intro real analysis class that's because it basically is.
 
Thanks Mike. I will make sure to get my hands dirty.
 
That should say 2r/n since Im(z) can be as big as r.
And this is implicitly for large n.
The bounds I'm getting there can be argued from the Taylor expansion but you really only need a first derivative.
 
3:42 PM
In all my analysis courses, arguments about simple things like this always got swept under the rug as "it's easy to see".
And of course, I fell for the fool's view that "of course, if it is easy, I shouldn't need to bother to check myself".
<--- bad person
 
If f is continuous at 0, $x\mapsto f(x/n)$ converges uniformly to f(0) on any compact set
Of course you need to know that $\exp$ is continuous at 0, which Thorgott/Mike's method shows
 
@Astyx wouldn't showing that implication be just as much work as Thorgott/Mike's argument (modulo the continuous at zero part).
Or is there a really easy way of seeing that implication?
 
It's analytic at 0
 
3:59 PM
Ultimately you have to do something somewhere, I used a definition of exp which doesn't use power series
If you use the definition based on power series then you have to work with that instead
Either works
Astyx has the mmore sound phrasing
@anakhro I always say "CHECK: " when I say something like this. "Easy to see" is really an instruction.
 
Hope everyone's having a good day
polynomials are holomorphic mappings from C to (shining) C
 
@MikeMiller one really hopes in hindsight that it is easy to see that "easy to see" is really an instruction.
 
4:19 PM
can someone help me set up an integration problem correctly
 
Depends on the integration problem. :P
 
actually I think I can set it up correctly just had to think for a sec
I need to evaluate it via contour integration!
this is going to be the difficult part
 
Hi, does anyone who know how to prove an algorithm called IDA*(iterative deepening A*), which is a graph searching algorithm in computer science. I really like to know a formal proof of it. If anyone could share it to me, I'm willing to formulate this into a question and provide bounty. Thanks!
 
4:36 PM
No bounty required.
 
5:25 PM
I have a function $F(z_1,z_2) := z_1^2\cos^2(z_2)+\frac{3}{4}z_2^2$.

I need to show that $\forall \alpha > 0\, \exists \gamma > 0\, \forall z\, \|z\| \geq \alpha \implies F(z_1,z_2) \geq \gamma $.

Aside from using Lagrange multipliers for this function $F$ with restriction $z_1^2 + z_2^2 \geq \alpha^2$ that give a basically unsolvable (?) equation over $z_2$:
$-(\alpha^2 - z_2^2)\sin(2z_2) + \frac{3}{2}z_2 -z_2\cos^2(z_2) = 0$ do you guys have any other ideas I could use here?

I'm asking this because this is an example taken from a book so I feel like it should not be that hard, maybe so
 
In the universal coefficient theorem for Homology, to show that $H_n(C;G) \cong H_n(C) \otimes G \oplus \operatorname{Tor}(H_{n-1}(C) , G)$ this splitting is not natural, I was asked to examine the map given by identifying the boundary of the disc in, $H_2(\Bbb{R}P^2, \Bbb{Z}_2) \to H_2(S^2, \Bbb{Z}_2)$. I have reduced it to showing that $H_2(\Bbb{R}P^2, \Bbb{Z}_2) \to H_2(S^2, \Bbb{Z}_2)$ is not the trivial map. How does one do that?
 
Compute kernel :)
The map is $\Bbb{RP}^2 \to \Bbb{RP}^2/\Bbb{RP}^1$
 
Greetings, a @Balarka, @Sayan, demonic A.
 
Does someone know quick intuition for Cayleys formula for number of spanning trees
In mathematics, Cayley's formula is a result in graph theory named after Arthur Cayley. It states that for every positive integer n {\displaystyle n} , the number of trees on n {\displaystyle n} labeled vertices is n n − 2 {\displaystyle n^{n-2}} . The formula equivalently counts the number of spanning trees of a complete graph with labeled vertices (sequence A000272 in the OEIS). == Proof... ==
 
proof should give intuition
 
5:38 PM
A typical way to product a spanning tree in $K_n$ would be to pick a vertex, flail your arm in $n$ directions, then stand at each of the other vertices, flail your arm in $n$ directions, so on. At some point you need to stop else you're going to create cycles, which spanning trees do not have. You have to stop at the $(n-2)$-th stage exactly, whence $n^{n-2}$
Roughly, this is the intution
 
famesyasd, are you able to show that $(0,0)$ is the only root of $F$? (is that true?) if so it would seem by continuity of $F$ enough to establish that $\inf F$ is bounded away from zero on any set of the form $\{z: \|z\| \geq \alpha\}$
 
@famesyasd Just to be sure. Your $z_1,z_2$ are real, right?
 
Isn't there a neat double counting proof we talked about once? @Balarka
 
@famesyasd if $\cos(z_2)\geq 3/4$, then $F$ is bounded by $\alpha$. If not, you have lower bounds on $z_2^2$
 
The usual proof is some coding thing
I never remember, but I am also not trying hard
 
5:42 PM
@Jakobian You mean $|\cos z_2|\ge\sqrt3/2$?
 
sorry, $\cos(z_2)^2 \geq 3/4$, not that it matters much
 
Well, it matters to someone who isn't knowing how to do it.
To be clear, I'm clarifying for famesyasd, not for you.
 
when i see "yasd" all i can think of is yet another sudden death. a phenomenon common to some of the computer games i like to play
 
Good thing I lead a sheltered life.
 
the games are uncompromising. like math, sort of. i do think that identifying the set of roots of $F$ is close to solving the problem. i would not want to explicitly construct the best $\gamma$ or anything like that. but i could be missing something
 
5:48 PM
It's equivalent just to bound below when $z_2\approx 0$ and then have an obvious lower bound when $|z_2|$ is bounded below. Again, I'm assuming all real variables.
@leslie Can you make heads or tails of this? ... I'm not going to tell the OP, but I'm guessing what someone told him/her to do is to look at the ODE that $\int_{-\infty}^\infty e^{-\pi x^2}e^{2\pi ix\xi}\,dx$ satisfies.
puts Balarka, Sayan, and demonic Alessandro on the naughty list
 
i'm confused by the problem statement, as you are. i wonder if they are hiding the ball a bit here. "someone told me to do X" would be much more preferable to "here's my hat, and now a rabbit is hopping out of it."
 
Yes, I tried to make my accusation less accusatory :P
 
i also personally cringe on 'thanks in advance.' i'm working with my therapist on that, too.
 
It's the people who want us to do their homeworks/exams who tend to say that ... or very polite people.
I truly was confused by the $\xi\in\Bbb C$ at the beginning and the $\xi\in\Bbb R$ at the end.
 
also the superfluous inclusion of a theorem number. it's fishy. i'm not saying it stinks, but it's fishy
 
5:56 PM
Well, they are citing Stein & Stakarchi.
 
didn't know 'thanks in advance' was so controversial
 
To me, it feels presumptuous, @Jakobian.
Especially given the current "do my exam question for me, and hurry" all over MSE.
 
for some people, 'thanks in advance' sounds more or less like an advance guarantee that no acknowledgment of anything will be coming later.
which yeah, 'do my homework,' i don't know. i do see it in my professional life where it is less cringeworthy.
 
Anyhow, do you agree that they're probably angling for (or being told to do so) the ODE approach?
I think Stein/Stakarchi do prove differentiation under the integral sign for integrals $\int_a^b$.
I no longer have that book.
 
yes that does appear to be the setup. it's so weird when people link solutions to the problems they're asking, and just say 'instead of'
the USPS lost about half of my math books during a cross-country move. i have approximately 50% of my previous mathematical knowledge. the curse was i managed to get most of the books i didn't even want anymore.
 
6:00 PM
They actually mistyped the theorem. They have $[0,1]$ and then $[a,b]$.
I got rid of literally thousands of my beloved books when I retired. Sigh.
 
there wasn't time. the answer is needed NOW. a,b,0,1, just answer.
i did find one of my lost books in a bookstore. that was a trip. i wanted to tell the guy at the counter that i was simply repossessing stolen property.
 
LOL ... so you packed all the good books in the boxes they lost, huh?
I nostalgically remember many hours browsing at Cody Books on Telegraph. Back when we had books and bookstores.
 
cody's was good. and moe's. and shakespeare & co (not great for math, but great)
i saw kurt vonnegut in moe's once. he was browsing history books
 
Moe's must be after my time.
 
same block as cody's i think. and now all of it is gone
i literally lost all of the expensive books i bought and retained books that i had been given for free
i have something henry helson gave me on dirichlet series. it looks like it's a book. i don't know if it was really published. i lost all of the $100 tomes they made me study in graduate school.
 
6:09 PM
Oh, Moe's webpage says they're still there. The webpage says they've been there since 1959. I honestly do not remember it at all.
 
they did not have a mathematics section. which cody's definitely did (halfway in the back on the first floor)
 
I was a French lit major in college, so I know how to read a non-math book :)
 
@TedShifrin That is one book I don't have. That was after I was there.
 
I taught graduate complex out of it one time, @robjohn. Halfway through the course, I regretted my choice pretty seriously.
 
I may get it at some point because it does present a lot of the things he taught in class that are not in his previous books.
 
6:11 PM
I knew that the geometry side would be weak, but I didn't realize how much they underplay the multivaluedness of log, which is something I want to play up from week 2.
There were a few serious mathematical errors, but I guess the strength (which didn't matter to me) is the inclusion of the number theory material.
 
when i was in graduate school i embarked upon a fairly aggressive campaign of copyright infringement, in partial revenge for losing my books. they are on a hard drive under my desk now. my favorite complex analysis book is conway.
 
I never liked Conway, either.
But since he's a functional analyst, I imagine he'd appeal to you :)
 
bak and newman is a really good undergraduate treatment.
and yes, that's how he hooked me
 
@TedShifrin I attended one of his seminars (about FRACTRAN) when he visited Princeton, and I wrote a Life implementation, but other than that, I don't know much of what he did.
 
Actually, Lang's book is surprisingly good. I was going to teach out of Gamelin my last time teaching the graduate course, but then I got cancer and almost didn't teach at all that semester, so I gave the course to someone else.
Different Conway, @robjohn.
 
6:13 PM
Oh
 
there are too many conways
i like lang's algebra a lot. he was very prickly in person, i don't know that i would recommend him as a person. but the algebra book is good
 
The one you're thinking of was the topologist/algebraist, who just died. The one leslie's talking about is (was?) a functional analyst.
I think the Algebra book is only good if you already know the material.
 
this is probably true
 
Yes, he was prickly. I sorta liked him, though. Although the seminar talk he gave at Berkeley where he ("teasingly") put down Chern was a bit shocking to me. Chern just smiled in his usual way.
 
he was kind of creepy around certain people. i'll leave it there. he also, i think, prioritized saying unpopular or controversial things over the kinds of compromises that most of us make when communicating with an audience. he found some joy in that
he was very funny
 
6:17 PM
I should point out to @robjohn that it's worse than he realizes. Both of these are John Conways. The algebra/topology/Life one is middle initial H, and the analyst is middle initial B.
 
Bak and Newman contains a really cool proof (by Newman) of the fact that the Fourier transform is injective using complex analysis in the last chapter
 
Creepy? Just because he was closetedly gay?
 
bak and newman is the coolest book ever. i'm surprised more people aren't onto it.
 
I actually never even saw that book, although I've heard of it, of course. I think it appeared after I taught undergraduate complex many, many years ago.
 
@Thorgott Could you sketch it?
 
6:18 PM
Salut @Astyx. Guten Abend @Thor.
 
Salut Ted
 
absolutely not. nothing creepy about being closeted, especially for a person of his vintage. he was sometimes creepy on top of that.
but not too creepy. i'm not canceling him
serge lang struck me as the kind of person who never met a bull without wanting to wave red banners in front of it
 
@TedShifrin yes, $z_1,z_2$ are real
@jakobian didn't quite get your message
 
is yasd a reference to another sudden death?
my therapist and i can process this together if the answer is no. i'm ready for either answer
 
damn these messages
 
6:26 PM
@famesyasd Ted Shifrin elaborated on what I was saying, maybe that'll be more understandable for you
what kind of atlas do we give to a coordinate slice? The one map from non-constant coordinates of coordinate system?
the book doesn't really explain that
 
If we know that ${\displaystyle P(s)=\sum _{n>0}\mu (n){\frac {\log \zeta (ns)}{n}}},$ then what is $P(sz)?$
 
So $f\colon\mathbb{R}\rightarrow\mathbb{C}$ is in $L^1$ and we assume $\hat{f}(x)=\int_{-\infty}^{\infty}f(t)e^{2\pi ixt}dt=0$ for all $x$. Fix some auxiliary real $a$ and now define a function of a complex argument $z$ by $\int_{-\infty}^af(t)e^{2\pi iz(t-a)}dt$ for $\operatorname{Im}(z)\le0$ and $\int_a^{\infty}f(t)e^{2\pi iz(t-a)}dt$ for $\operatorname{Im}(z)\ge0$. For real $z$, these agree by the hypothesis.
Dominated convergence implies continuity of this function on $\mathbb{C}$ and we see it is bounded by $\int_{-\infty}^{\infty}|f(t)|dt$. In fact, it is holomorphic on the upper and
 
The emoji with the hearts in the eyes
 
@Jakobian Yes. If you're looking at the $(n-k)$-dimensional slice given by $x^{k+1}=\text{const}$, ..., $x^n = \text{const}$, then of course $(x^1,\dots,x^k)$ give coordinates on the slice.
 
I think the idea behind this proof is motivated by Paley-Wiener theory, but I never learned that proper
it's a very neat trick in any case
 
6:33 PM
ok, thanks
 
"For real $z$ these agree by the hypothesis" you mean they're opposite of each other right?
 
oh, I forgot a minus sign on the second integral
 
My advisor once showed me a book from Wiener that he had had signed by Wiener. I do not have many mathematical heroes but it did feel cool to touch the page.
 
@leslietownes just to be clear we pick a sequence $z_n$ by contradiction that tends to zero and then apply continuity of $F$?
 
Very neat trick indeed!
 
6:39 PM
If I ask question can 10 people agree to upvote it instantaneously?
 
no
 
@famesyasd Too hard. Just do a lower bound when $z_2\approx 0$ (so Jakobian's suggestion is $|\cos z_2|\ge \sqrt3/2$) and then give a lower bound for the rest of $|z_2|$.
 
@famesyasd see Ted's comment. it is probably the simplest solution.
A key in that problem seems to be that $F$ is a continuous sum of nonnegative things. This does put a lot of pressure on when those things are both going to be zero.
 
I can't believe I got 10 upvotes on a question on the Mathoverflow
I must have asked an easy to state but hard to answer question
it's definitely not research level lol
Do you think if someone solved like the Novikov conjecture or the Hodge conjecture in a math S.E. exchange post it would garner many upvotes?
I think it would get like no upvotes cause nobody would understand lol
 
7:04 PM
you may be right about that. if i look at my upvotes it is mostly for the dumb stuff that i say and not the smart stuff that i say. which isn't to say that i say smart stuff. people do upvote dumb.
and if it seems technical and confusing, forget it
 
@leslietownes okay I think I get, thanks guys. btw, yes, "yasd" means "yet another stupid death", an acronym coming from roguelike games, I used to play them a lot when I have created this nickname. My favourite game is Rogue, I personally think it's the best game of all time. I have added yasd to my nickname when the system considered "Fames" being too short or taken. I don't really think it is a great addition but whatever.
BTW I was really edgy when I was creating my nickname so fames means "hunger" not fame and it's spelled diffrently not how you think
 
if the solution was posted it on Mathoverflow it would probably be recognized as legit and people would upvote it a lot
 
i love that yasd meant what i thought it meant. i am a nethack fan.
mathoverflow is a strange place. i forgot the password to my account there
 
geocalc33 is derived from geo-metry calc-ulus and the number 33
 
I don't like nethack don't beat me :)
 
7:09 PM
because calculus (with analytic geometry) was my favorite subject in college
 
Leslie Townes is the first and middle name of the actor known as "Bob Hope"
 
thought that name sounded familiar
 
i don't like the new nethack. 3.4.3 was OK
 
7:37 PM
@BalarkaSen Okay I think I get it now. Tell me if this is correct. When I consider $\Bbb{R}P^2/\Bbb{R}P^1$, the 1-cell is being collapsed, so the CW structure is just $e^0,e^2$. Since the quotient map $q$ is taking the original 2-cell to the 2-cell in $S^2$, I am guessing it should induce an isomorphism between $H_2( e^1 \cup e^2 / e^1)$ (for $\Bbb{R}P^2$) and $H_2(e^2/e^0)$ (for $S^2$). Now when I take $\Bbb{Z}_2$ coefficients the degree maps are all 0 for $\Bbb{R}P^2$ and same for $S^2$
Which gives me that both $H_2(\Bbb{R}P^2,\Bbb{Z}_2)$ and $H_2(S^2,\Bbb{Z}_2)$ are isomorphic as there are no higher cells in the CW complex.
 
Yes
If $A \subset X$ is a subcomplex then $C_*^{CW}(X/A) = C_*^{CW}(X)/C_*^{CW}(A)$
All chains reduced
So you're ust doing an algebra computation
(The natural map $C_*^{CW}(X) \to C_*^{CW}(X/A)$ is identified with the canonical quotient map)
 
8:07 PM
I've got 12 upvotes for a smart thing once
but I've put a lot of bounties on that question because no one seems to be able to answer it
only easy questions with hard answers attract audience
and it makes sense
not really a mystery that the further you go, the less and less people know what you're talking about
 
I had a series of questions about modular forms that only one guy was answering
 
it's official: I passed my physics exam in the absence of understanding
 
Glückwunsch
 
is anyone solid at evaluating contour integrals
I guess I have to draw a contour and sum the residues
 
8:34 PM
@Thorgott what physics?
 
newtonian mechanics
 
langrangian wish I learned
instead of newtonian
1
Q: On evaluating $\rho(s)=\frac{1}{2\pi i}\int_{c-i\infty}^{c+i\infty}\Gamma(z)P(sz)\,dz.$

geocalc33How do you evaluate the following integral? $$\rho(s)=\frac{1}{2\pi i}\int_{c-i\infty}^{c+i\infty}\Gamma(z)P(sz)\,dz.$$ $P(\cdot)$ is the prime zeta function. I got the integral from taking $\rho(s)=\sum_{\rho\in\Bbb P}\exp\big(-\rho^s\big)$ and using the Mellin transform to extend the domain o...

okay I just asked my contour integration exercise
I think I almost got it but I think I messed up somewhere
 
8:58 PM
@Thorgott I take this statement with several grains of salt.
 
00:00 - 21:0021:00 - 00:00

« first day (3864 days earlier)      last day (51 days later) »