« first day (2516 days earlier)      last day (2505 days later) » 

1:00 PM
I just don't think this is serious, as in, an actual medical condition. I would think everyone feels this thing (especially us math-enthusiasts, it seems, from talking to people in this room. which is fair, mathematics is so isolated from the real world that that's entirely believable). Mine has gotten complicated and more than just "I am depressed, because hey cloudy sky" because of all the shit I read lol
@Just Nope.
 
perhaps you should try some form of meditation
for the sake of "peace of mind"
 
Meditation has the opposite effect on me
 
why do you think that is?
 
I can certainly imagine a few ways it can have the opposite effect.
 
it would depend on the type of meditation you choose
 
1:08 PM
Maybe I'm doing it wrong, but keeping myself busy prevents me from thinking too much about things that make me sad and puts me in a naive/ignorant state (which might be why I like maths so much, cause it requires so much thought process); whereas actually reflecting on who I am, what I'm doing etc makes me depressed about the absence of a meaning to life and thus to anything
 
That's more like what I thought, @Astyx. Unadulterated flows of thoughts can be potentially more harmful.
Well, harm is a strong word but at least less constructive.
 
It's some kind of morbid understanding of Sartre's "Man is condemned to be free."
 
Do you read Dostoyevsky?
I think I felt the existential thing a while ago but now my issue is something else.
 
I've only read "The Gambler"
 
That's a fantastic story, but maybe a little too dramatic.
 
1:15 PM
Agreed
 
I wouldn't suggest you the big tomes, but try Notes from The Underground.
 
What's it about ?
 
A man who went through paranoid existential crisis through 40 years and resurfaced from his own hell, with his convoluted and degenerate philosophy and spite, and a mass of memories.
It's broken up in two parts, the first his philosophical journal (14 smallish chapters) and the second an account of one of his memories.
 
Do you know about Sweig's "The royal game" ?
 
Oh yes, absolutely fantastic.
That's the only Zweig story I have read, though.
 
1:20 PM
Same goes for me
 
Hey
 
Hi
 
I played some drinking game last night
one player decides on a number between 1 and 50
then the other player(s) have 5 chances to guess the number
the player who knows the number, always responds by "higher" or "lower"
 
(@Astyx I think my existential perspective has not restricted to an inward view, though. I have been looking at various literature which deals with the exterior, and distortions of the classical view of reality)
 
What would be the best strategy?
 
1:25 PM
I think to start from the middle. 25 --> 12 or 13 --> etc. Keep dissecting.
 
dichotomia
 
I don't know much about game theory or so, I was just wondering^^
Yeah, I also found it the intuitively best way to just always guess the middle of the actual interval
say we have only 4 guesses
 
1) Start with $a=1, b=50$. 2) Play ${a+b\over 2}$. 3) If the result is higher, change $a$ to ${a+b\over2}$, other wise change $b$ to ${a+b\over 2}$ 4) repeat step 2
 
if you always choose the middle, you end up (if you are not "lucky") with some interval of about 6 numbers
maybe 5
dunno
and then in the last turn you have some quite small chance to win
wouldn't it be better to risk more at the earlier turns already?
 
Risk does not depend on where you choose it. You're lowering the probability by the same factor.
 
1:29 PM
It's a luck game anyway
 
hmm okay
 
The point is if you choose from 1, 2, 3, 4, 5 say, and you're supposed to guess in 2 steps, then choosing 3 divides it into 2 numbers in each set: {1, 2} or {4, 5}. So there's 1/2 probability for you to win after you know on which set it's in.
But if you choose 2, you will either have full probability (if the number is 1) or 1/3 probability (if it's 3, 4, 5).
Plus, the number lies on {3, 4, 5} with more probability than {1}
 
Did you know you can combine two losing games in a winning one ?
 
So it's a terrible strategy.
 
If I choose 3 and then in a possible second turn choose one of the two remaining numbers, my overall chance to win is 7/10, right?
or wait
6/10
3/5
I could also say I choose 3 and then either 1 or 4, so chances are 3/5
 
1:36 PM
Drinking is bad for your health anyway
 
If I choose 2 and then either guess out of 3,4,5 or win by naming 1, then chances are also 3/5
I could also say I choose 2 and then either 1 or 3
@Astyx I'm Aware of that^^
 
Guys, my book says that $t_P\circ \sigma_{\phi}$ not a linear transformation is of $\mathbb R^2$, because it is a reflection through a line that doesn’t go through the origin. I was wondering: why do we need $(0,0)$ to be our neutral element? Can’t we have a linear transformation of $\mathbb R^2$ where we pick a different neutral element, such that the congruence is still a linear transformation of $\mathbb R^2$?
 
Then it's an affine transformation
That is, the composition (or sum) of translations and linear transformations
 
well it's still a congruence
just not a linear map on $\mathbb R^2$
I think I see
 
Just like de facto you require planes in $\Bbb R^3$ to go through the origin, the other "planes" actually being affine planes
And a linear function $\Bbb R\to \Bbb R$ is something that is $x\mapsto ax$ for some real $a$, while an affine one is any polynomial of degree $\le 1$
 
1:43 PM
because $(x,y)$ are the vectors of our vectorspace, we don't have a choice but to have $(0,0)$ as our neutral element. I dunno why I was confused, I think I had forgotten the definition of a linear map for a sec
 
Actually I think that the strategy does not matter as long as it "uses all possible intervals".
 
anyhow thanks I guess @Astyx sorry I was just needlessly confused
 
Glad to help
 
So always choosing the middle is an optimal strategy
among many other optimal strategies
(if the initial interval is big enough relative to the number of guesses)
if you have 1000 numbers and 3 guesses, guessing any numbers such that you always divide the actual interval into two nonempty intervals, will always give you a Chance of 3/1000 to win
@Balar
@Balarka Sen isn't this correct?
or actually this cannot make sense
then I could just as well choose 1,2,3
I think I'm still drunk
 
2:01 PM
Sorry, I was away.
It is true that the probability of one of those 3 guesses being the correct number is 3/1000. But I don't think this means it's the best strategy.
I think the point is using the choosing-the-middle strategy, you get the highest probability of winning at each of those 3 steps.
That should be the actual meaning of "best strategy".
 
And I mean, if you're drinking, the efficiency of the middle solution considering its simplicity might be enough, you most probably wouldn't think clearly enough to apply a better strategy if one even exists
This being said, I have to go, see ya later chat
 
lol
Bye.
 
haha^^
 
2:51 PM
for your drinking game its not only that you need to restrict the amount of numbers as much as possible to get to a solution faster, but you also need to prevent the information from your guesses from helping the others too much, who have a turn immediately after you
for example if you know that the number is between 1,2,3,4,5 and there are two other guys picking 3 either makes you win immediatly if its 3 or make it impossible for you to win
 
Hi folks
Anyone know a comprehensive textbook I could find to really learn about probability properly (from the fundamentals on). I'm finding myself asking "Why?" to some of the things the basic theory you get in school teach you.
 
@GhotiandChips Try checking the suggestions made in here, see if anything suits you: math.stackexchange.com/questions/31838/…
 
@John11 Cheers, @John11
 
@s.harp The way how I meant the game to be was basically a game of only 2 Players, the one who knows the number and the one guessing it
You just usually play it with more players, so one knows the number and the others guess as a team
But I also thought of some variations
You could say that the guessing players actually play against each other and only the one who guesses the correct number wins
or: The one guessing the correct number loses
 
 
1 hour later…
4:11 PM
[Chemistry] Ok the other program has finished the calculation. However it spits out another ungooglable error message. Plan to try out some TD-DFT stuff. If that still does not work I am going to contact the support centre
 
4:24 PM
For all the CompSci ppl please drop by

 theory salon

theoretical computer science. highlight reel vzn1.wordpress.co...
 
hi chat
 
lol @Semiclassical
 
Zee
My proof didn't work
Am sad... :(
 
Getting an error message is frustrating enough, but getting one which google have absolutely NO RECORD of is even more headscratching
 
@secret yeah, that's always "fun"
 
Zee
4:28 PM
A chance to help spread knowledge if you ask me
 
and for this latest one, I am no longer sure whether it is a software problem or a computational chemistry problem (because I have tried many flags I can find to localise the error, and it still kinda nebulous on how it popped up. It could be possible there is something really wrong with the wavefunction of this particular molecule). Hopefully will get that all sorted out soon
 
Zee
can someone clarify the answers to my question?
2
A: If $f \in L^+$ and $\int f<\infty$ then there exists a null set

user133281No, the proof doesn't work. If $f_n \to f$ and $\mu(f_n=\infty) = 0$ for all $n$, this does not imply that $\mu(f=\infty)=0$. As a counterexample, consider $f_n = n \mathbf{1}_{[0,1)}$ for all $n$. A correct argument could go something like this: suppose $A = \{f = \infty\}$ has positive measur...

 
@BAYMAX Heya. Were you able to make sense of the rest of that problem?
 
Hi@Semiclassic - & - all
yes
I did it :)
leaky helped me
actually I was jammed
hey
I actually posted a question in main but I could not appreciate the answer any help!
in the first answer
why it exists for finite positive time?
it = solution :)
 
Not sure what you're asking. Do you mean, why do solutions for $x_0 \in(0,1)$ never leave that interval?
 
4:41 PM
I heard that solution for this ODE exists for finite backward times ?
 
Sure.
If you run it backwards, the flow directions reverse.
 
like finite time
why not all times
 
So unstable fixed points become stable and vice versa.
 
hi @TedShifrin
 
Well, at infinite time you can reasonably expect that some solutions will diverge to infinity.
 
4:43 PM
after a long time
yes@Semiclassical
 
Hi @BAYMAX, Semiclassic. I'm still traveling in Europe.
 
hi @ted
 
nice
I mean @Semiclassical there is something intersting in this problem!
 
@GhotiandChips: Good for you for educating folks about GB Shaw :)
 
Zee
@TedShifrin hello :D
 
4:45 PM
Hello
 
Zee
How's Europe?
 
Blazing overheated.
 
like
 
Mar 2 '15 at 17:51, by David Wheeler
In oversimplified terms: if we have $\alpha = \{x \in \Bbb Q: x < \alpha\}$ (ingnore the circularity for now), and similarly for $\beta$, we define $\alpha + \beta = \sup(\{z \in \Bbb Q: x \in \alpha, y \in \beta\})$
If there's one thing that ordinals taught me, is that supremums are weird
 
Zee
Do the people there have different mentality? For better or worse
 
4:46 PM
I think it helps to draw out families of solution curves.
 
I haven't read GB Shaw but I should
 
@Semiclassical for $x_{0} < -1$ when is the solution valid
 
like computationally i see that solutions diverge when $x > 1$
and $x < -1$
 
My guess is that it's valid for all finite times. (As $t\to-\infty$ it goes to $-1$ and goes to $-\infty$ as $t\to\infty$).
 
4:48 PM
@Balarka: he's the one who used the word GHOTI to prove how idiotic English is.
 
@ted tbf, is there any language which isn't idiotic?
 
Mar 2 '15 at 18:04, by David Wheeler
Also: "clock numbers" (finite fields of prime order) also obey the same scope of arithmetic rules as rationals do, with much different results
O, these things have a name instead of $\Bbb{Z}_p$...?
 
for ex, in Croatian there are no silent letters or diphthongs — letters are always pronounced the same.
 
@TedShifrin I just googled, haha.
 
4:49 PM
This is fantastic
 
$x_{0} > 1$ the soln blows up in finite positive time?
 
I don't think so.
 
I've told people about that for decades, Balarka, including here, I Thought.
 
I could be wrong, though.
 
"In Finnegans Wake, James Joyce alludes to ghoti: "Gee each owe tea eye smells fish.""
HAHAH
 
4:50 PM
@TedShifrin Who?
 
Easy to check, though. If $x(t)$ is sufficiently large, then $\dot{x}=-x+x^3\approx x^3$.
 
Oh, Balarka, I had heard it credited to Shaw.
 
@TedShifrin Yeah, it seems it's originally by Shaw. Wikipedia references James Joyce in notable uses.
 
Hi @TedShifrin!
 
Somehow, I knew Joyce could not miss this pun.
I was almost looking for him in there.
 
4:52 PM
Is it common in mathematics to lose the distinction between holonomy groups and their restrictions to null-homotopic loops?
 
't was a Joyce reference when I created it ^^
 
Ah, Hi @Danu
 
If I assume a solution $x\sim at^p$ then that'd be $\dot{x}=pat^{p-1}=x^3 = a^3 t^{3p}\implies p-1=3p\implies p=-1/2$
 
Usually a different term. Geometry cares about holonomy around all loops.
 
So $x\sim 1/\sqrt{t}$...huh.
Methinks I did something wrong.
 
4:54 PM
Mar 3 '15 at 12:43, by Balarka Sen
Define $C_n(X) \stackrel{\partial}{\longrightarrow} C_{n-1}(X)$ by mapping $M \to X$ to $\partial M \to X$. From this, we have the chain complex of abelian monoids : $$\cdots \to C_{n+1}(X) \stackrel{\partial}{\to} C_n(X) \stackrel{\partial}{\to} C_{n-1}(X) \to \cdots$$
I am curious what this will look like in pictures... would I see a space filled with loops...?
 
Huh. Looking at Mathematica to save time, it appears the solutions should be of the form $x(t) =\pm\frac 12 (c-t)^{-1/2}$
 
@TedShifrin But all classification results are for simply connected (complete) manifolds, so we only have classifications of the restricted groups (?)
 
@BAYMAX Which would indeed blow up in finite time.
I'm perplexed by that, I'll admit, but I probably shouldn't be.
 
no time factor in sloution
 
typo.
 
4:56 PM
They see the geometry rather than the topology, Danu. I guess fundamental group can create a group extension or something. Ask Mike.
 
ok
what $x_{0}$ you took@Semiclassical
?
 
And checking Mathematica's full solution of the original ODE seems to bear out that solutions escape in finite time.
I didn't take a specific one. Note that $x(0)=\pm \frac 12 c^{-1/2}$, so one can parametrize $x_0$ by $c$.
 
ok
 
@TedShifrin The restricted holonomy group is the connected component of the identity, and coincides with the holonomy of the universal covering space
 
But if I write it in terms of $x_0$, I get $x(t)=\dfrac{x_0}{\sqrt{1-2t x_0^2}}$
So it blows up at a time $t=\frac12 x_0^{-2}$
That's for the x^3 dominant solution, though.
So I only expect it to be a relevant approximation when $|x_0|>1$
 
5:00 PM
Black dots are zeta zeros
Looks like it could be a vector field.
 
nice @Semiclassical
 
Right @Danu.
 
@BAYMAX If I plug the exact ODE into Mathematica and solve, I find that solution with $|x_0|>1$ diverge in a time $t=\frac12 \log\left(\frac{x_0^2}{x_0^2-1}\right)$
 
@Secret do you want an integral challenge?
 
Which is positive and finite so long as $|x_0|>1$, as it should.
 
5:03 PM
oh,so what conclusion we get!
 
As for how one could predict that the solutions would escape in finite time...tbh, I don't know.
 
@LeakyNun Depends on what, cause you know I am not that good at integrals, but shoot
 
I guess the fact that $\dot{x}\sim x^3$ for large $x$ is the point.
 
@Secret $$\int \sqrt{\tan x} \ \mathrm dx$$
 
That integral is a problem in Spivak.
 
5:04 PM
so for $0<x_{0}<1$
we have steady state solution
but for $x_{0} > 1$
 
@TedShifrin interesting
 
No. You're asymptotic to the steady state solution.
 
yes
sorry
 
But the solution never actually reaches there because $\dot{x}$ gets smaller and smaller the closer you approach.
 
@TedShifrin Did you see my plot?
 
5:05 PM
More generally, I think that $\dot{x}\sim x^p$ for large $x$ leads to finite-time escape for any $p>1$.
 
for $x_{0} >1$,the solutions blow up in finite time as per yours $t = \frac{1}{2}\log\frac{{x_{0}^2}}{{x_{0}^2 - 1}}$
and for $x_{0} < -1$
oh nice
the solution blows up
 
Right.
 
but not at finite time
 
for $x_{0} < -1$
 
5:08 PM
Should still blow up at finite time.
 
like denominator $t \rightarrow \infty$
 
For $x<-1$, $\dot{x}<0$.
 
@TedShifrin I'm quite confused why, givne that the classification of Berger concerns Hol^0, people instead study manifolds with Hol contained in his groups.
 
More generally, note that your ODE $\dot{x}=-x+x^3$ is invariant under the reflection $x\mapsto -x$.
 
yes
gotcha
 
5:10 PM
So if you've got solutions with $x_0>1$ which blow up in finite time, then there's a reflected solution with $x_0<-1$ which also blows up in finite time.
 
solid!
 
Zee
If increasing sequence of simple functions converge to f point wise , and they are infinite only on a null set, does that imply f infinite only on null set as well?
 
a cake for you :) from me
@Semiclassical
 
sorry but its imaginary
 
5:11 PM
@Zee what is a null set?
 
For $t$ equal to imaginary part of a Riemann zeta zero:
$$t = 2 \pi e \exp \left(W\left(\frac{\frac{\arg \left(\frac{\zeta (c)}{\zeta \left(c+i t+\frac{1}{2}-1\right)}\right)}{\pi }-k+n-\frac{\vartheta (t)}{\pi }+\frac{t \log \left(\frac{t}{2 \pi e}\right)}{2 \pi }-1}{e}\right)\right)$$
c->1
 
Zee
@LeakyNun set of measure 0
 
what does it mean for f to be infinite?
 
Zee
@LeakyNun intuitively a set with 0 area
 
@BAYMAX To quote a tired old meme: the cake is a lie.
 
Zee
5:13 PM
F = infinite
 
oh
 
and I'm asking you what that means
 
Zee
F(x)= infinity
 
so i offer you Mpie
Mathematicalpie
$M\pi$
 
loool
 
5:14 PM
hey thatts a palindrome
 
so it is.
 
k=1/2
 
@Zee what is F a function on?
 
Zee
@LeakyNun a measure space, but I suppose for simplicity you can say R
 
and the codomain?
 
Zee
5:17 PM
The positive real line
Including infinity
And zero
 
alright. I have no idea then.
 
Zee
@LeakyNun I appreciate the effort!
 
Morning @MikeMiller
 
Hi @MikeMiller, Ted just deferred me to you, when I asked him some stuff about holonomy.
I have a naive question: Given that the classification results of Cartan & Berger are for simply connected manifolds (or equivalently for restricted holonomy groups), why are the classes of spaces that people study those with global holonomy a given group (one from Berger's list), rather than the restricted holonomy group?
 
5:39 PM
@Danu Is that true? I'm guessing it probably depends on person.
In any case, holonomy corresponds to geometric properties, while restricted holonomy does not.
(Is the Enriques surface hyperkahler? Does it have a nonvanishing holomorphic 2-form? I dunno.)
no, it doesn't
 
One incarnation of my question is, for instance:
Why say a manifold has $G_2$ holonomy if its global holonomy is contained in $G_2$, rather than just the restricted holonomy?
Idem for, for instance, quaternionic Kähler manifolds and $Sp(n)\cdot Sp(1)$
@MikeMiller I'm not sure what that means...
 
[Some unconventional integral experiment]
To be figured out why it does not work
$$\int \sqrt{\tan x} dx$$
\begin{align}
&\int \sqrt{\tan x} dx\\
& = \int \sqrt{\frac{\sin x}{\cos x}}dx\\
& = \int \sqrt{\frac{(\cos x)'}{-\cos x}}dx\\
& = \int \sqrt{-\frac{f'}{f}}dx\\
& = \int \sqrt{-(\ln(f))'}dx\\
& = \int \sqrt{(\ln(\frac{1}{f}))'}dx\\
\text{Let }
u^2 & =(\ln(\frac{1}{f}))'\\
2udu & =(\ln(\frac{1}{f}))''dx=2udx\implies du=dx\\
& \int \sqrt{(\ln(\frac{1}{f}))'}dx\\
& = \int \sqrt{u^2} du=\frac{u^2}{2}+C=\frac{\tan^2x}{2}+C \neq ANS\\
\end{align}
 
christ chat exploded
 
@BalarkaSen ?
 
was referring to Secret's message
 
5:52 PM
ANS ?
 
means "answer". It is a scary thing that consists of hyperbolic sinh cosh, while for another version a lot of arctan and tan
Interestingly, I am actually using the correct substitution $v^2=\tan x$, however, somehow it screwed up in the middle and give the wrong ans for my case
 
But that doesn't work does it ?
Well $\sqrt{u^2} \ne u$ in general
 
no, the problem isn't there
the problem is, of course, $\mathrm du = \mathrm dx$
the $=2u \ \mathrm dx$ simply does not follow.
your attempt basically reduces to $u^2 = \tan x$, and then how the rest fails is obvious.
$2u \ \mathrm du = \sec^2 x \ \mathrm dx$
$\sec^2 x \ \mathrm dx = 2u \ \mathrm dx$?
 

« first day (2516 days earlier)      last day (2505 days later) »