« first day (1352 days earlier)      last day (3676 days later) » 

12:00 AM
k
 
@Chris'ssis can I ask for brighter color ;)
 
12:23 AM
@karl getting back into lee. the rank theorem is hideous
 
@Mike It is so god damn useful though
 
oh I believe that
it looks useful, and the global rank theorem he proves right afterwards shows its power
but like hell am I remembering the details of that proof
 
A spoiler implemented with toggles that doesn't toggle back to the original text: $\require{action}\toggle{\text{You will never see this again.}}{\toggle{1+1=2}{1+1=2}\endtoggle}\endtoggle$
 
lol invoking the Baire category theorem
no way this statement requires that much
 
12:43 AM
@mike: it's just a jazzy way of rigging the inverse function theirem ...
 
@karl is there a reason lee defines smooth emveddings as smooth immersions that are topological embeddings instead of diffeomorphisms onto their images?
nvm
ignore that
@Ted The idea is simple, the symbol-pushing is unpleasant
 
1:04 AM
Hi @Ted
 
Hi @user127001; @Mike: I don't consider that symbol-pushing at all.
@Mike: Super cool problem. Show that a smooth retract must be a submanifold, but a continuous one need not.
 
1:19 AM
@Ted Submanifold may have boundary?
 
No, no, no boundary
 
Oh, then the second part is snooze zone.
 
LOL, yes?
 
I kept thinking of weird retracts but could only get submanifolds... but sometimes they had boundary :P
Easiest: $f: \Bbb R \rightarrow [0, \infty)$, $x \mapsto |x|$
 
Huh?
 
1:22 AM
Unless you meant a deformation retract?
which I guess is still easy
 
your mapping is not what you're thinking
 
uh?
 
but, ok, in the smooth category if you start with a manifold, a retract must be a closed submanifold with no boundary
$x\to |x|$ is not defined on $\Bbb R$, and the image is $\{\pm 1\}$.
 
I have no idea what you mean.
 
you wrote $x\mapsto |x|$, did you not?
oh, what I wrote is nonsense, never mind.
 
1:25 AM
Yes. I did not write $x \mapsto x/|x|$, which is what I think you thought I meant
 
OK, you're right ... I had in mind a continuous retraction $\Bbb R^2$ to the axes. But the interesting thing is the smooth ($\ge C^1$) category.
 
Anyway I should get through this chapter first. @Ted Do I need any more tools for your problem than I have?
 
Nope.
 
Ok, I'll think about it when I finish what I gotta do.
 
Okey dokey.
 
2:00 AM
Does anyone know how to convert a integral into another?
 
@Victor You mean like $\int{af(x)}{\;dx}=a\int{f(x)}{\;dx}$ ?
 
@MickLH the relationship between $\int a(x)f(x)dx$ and $b(x)\int f(x)dx$
how to make them equal?
 
???
 
oh hold on
 
2:21 AM
@Victor sorry got a phone call, try researching "u substitution" however you'd like to
 
@MickLH right, good night
 
 
2 hours later…
4:39 AM
The "Puzzling" proposal in Area51 is quite close to being fully committed, and just needs an extra little kick: area51.stackexchange.com/proposals/45128/…
 
 
1 hour later…
5:43 AM
I created a community ad for the "Puzzling" proposal (since, as I mention in the comments, you already have the related and tags here.
It just needs two more votes to go public: meta.math.stackexchange.com/a/12775/55164
Note, that currently the link marks me as the "referral person"; if anyone has a problem with that, please, edit the link to remove me from it. I simply felt it would be a shame letting the rep go to waste if no one is listed as the referrer.
 
cap
6:06 AM
Hi, off topic: If $0\le x/y < u/v \le 1$ I do not see how it can ever be the case that $yv-ux=1$. For if we cross multiply the first inequality we obtain $yv-ux>1. Am I correct?
 
well, are $y$ and $v$ both positive?
or at least is $yv > 0$?
 
cap
Yes, all integers mentioned are positive.
 
im not sure how you're getting $yv-ux>1$ from that
multiply through by $yv$ and you get $0 \leq xv < uy \leq 1$
 
@Chris'ssis Have you seen this question? math.stackexchange.com/questions/470527/…
 
6:30 AM
Morning!
Can anybody answer this question please?
0
Q: Franck and Hertz graph function

Ionică BizăuI need a function whom graph is the following (the Franck and Hertz experiment graph): Currently I wrote the following function: function comptuteValue (value) { var period = 4.9 , low = 6 ; function addOrSubstract (x) { if (x < low) { if (x <= p...

 
7:10 AM
I made a root function that uses newtons method, on the wikipedia for newtons method is written: "the convergence is at least quadratic (see rate of convergence) in a neighbourhood of the zero, which intuitively means that the number of correct digits roughly at least doubles in every step.", however my algorithm dosn't have quadratic convergence, infact it takes it 125 steps to get to $\sqrt{10}$ with initial guess of 5
anyone knows why this is
 
I dunno what algorithm you're using
newton's method is getting me pretty damn fast convergence
correct to 50 digits after 7 steps
I don't think you're using the right algorithm, @Darksonn
For $\sqrt{10}$, you should be using $x_{n+1}=x_n-\frac{x_n^2-10}{2x_n}$
This doesn't fit your data points
 
7:28 AM
let me see, for a^(1/b) it should be
$x_{n+1} = x_n - \frac{x_n^b - a}{bx_n^{b-1}}$
if I'm not mistaken
 
@Mike Hey!
 
@Darksonn Yep - but that doesn't fit your data
 
Does conditional convergence mean anything significant? If you can just rearrange terms to get an arbitrary limit, it seems rather meaningless….
 
4.7 (using that) should output 3.413...
 
okay, I didn't derive that from my algorithm, I'll try and rewrite the step formula
 
7:29 AM
@Anthony then don't rearrange the terms!
 
@Mike But what does it mean :( It's so weird.
 
It means the sequence of the partial sums converges?
 
Just different partial sums…
It's so weird
 
What different partial sums?
I don't understand.
 
I just don't understand how rearranging terms can change what it converges to. That's so weird.
 
7:31 AM
Conditional convergence means that $\lim \sum_{k=1}^n a_n$ exists, but $\lim \sum_{k=1}^n |a_n|$ doesn't. That makes perfect sense.
Oh, well rearranging things.
That's a different story.
 
Why would rearranging things result in a different limit! :(
 
@Anthony Here's something to help you get a grasp on it. $\sum_{k=1}^\infty (-1)^{k+1} \frac 1k$ converges conditionally.
 
I know, I was reading wiki :P
I just get why if you're adding things up, the order should matter?!
 
But let's rearrange those, so that we do $1+1/3+1/5+1/7-1/2+1/9+1/11+1/13+1/15+1/17-1/4+...$
 
Agh!
 
7:34 AM
@Anthony If you rearrange finitely many terms, nothing changes.
It's when you start doing an infinite amount of shuffling that things get weird.
 
WHYYYYY
 
Dude calm down
 
:(
 
Look at the rearranged thing I gave you up there
 
ok, I really don't know what algorithm I was using, now it seems to converge in less then a second for $\sqrt{5000}$, instead of the 5 minutes it took before, thanks @Mike
 
7:35 AM
I know that it happens, I just can't fathom why it happens.
 
@Anthony Give me a second.
 
Aye aye captain.
 
By putting the negative terms sparingly - let's say putting the first one at the $10$th spot, the second at the $100$th, the third at the $1000$th...
We're essentially just adding up the harmonic series with some noise every now and then.
But it happens so rarely that it doesn't matter. You can prove that the series I'm describing diverges.
The thing is when you move around infinitely many things, you're totally changing what the partial sums $s_n$ look like.
 
@Mike what does "infinity" mean to you?
 
If you move around finitely many things, then the $s_n$ are eventually the same as they were before - so the series converges to the same value, like your intuition says it should.
 
7:38 AM
@Mike Is this really just a matter of not understanding infinity? Because god that's unbearable.
 
@Anthony No, it's a matter of our individual terms being, well, pretty big.
 
@Mike Whatchu mean?
 
Because $\sum |a_n|$ doesn't converge, the absolute values of the individual terms don't decrease "fast enough". That ultimately means that the "tail end" of the series - which is where all the action really happens! - doesn't just go to 0 as fast as we want. Fiddle with it too hard and it's not going to play nice back.
Have you seen the proof of Riemann rearrangement?
 
I was reading it/Also need to prove it for infinity.
 
"to prove it for infinity"?
 
7:43 AM
We don't talk about "infinity," instead we say "infinitely many." Am I right @Mike?
 
Show that I can rearrange something to diverge to infinity *
And do you not get why I'm so confused-ish? It's so weird that you can *reshape* something like that. I mean… it's just infinities?
 
Oh, no, I understand. But there's a difference between not understanding the rearrangement theorem which is weird as shit, and not understanding conditional convergence.
 
Nono, I'm sorry.
I understand.
It's just so weird.
 
Ah
 
Like, seemingly impossible.
 
7:48 AM
Well, that paper might give some insight
 
I don't get infinity, I guess?
:P
 
Meh, I think that's a copout answer. Do you understand why the harmonic series diverges?
If you can understand how to rearrange a conditionally convergent series so that it diverges, I think you'll feel less weird about it.
By "understand" I mean "see".
Do it yourself. Rearrange a series.
Show it who's boss.
 
Do you have me on ignore @Mike?
 
No, but I'm ignoring you.
 
7:52 AM
:D
 
thanks
 
@Mike I see how you can. Just. Crazy.
Because it's the same numbers. I guess it just isn't going anywhere anymore….
 
@Anthony Smoke less weed while doing math :P
 
lol
 
@Mike Well this is Berkeley.
 
7:55 AM
The good Bishop would be proud ;-)
 
after I rewrote this pow function, I can calculate roots insanely fast
 
:D
Thank Newton
 
sqrt(145064065156012560412310032121325462540823507238023541705234023805234108523‌​142052037182042305270725052870240205274027020123027315270270120725027052701270253‌​0527023702) is calculated to 117 digits after the comma in 4 seconds
@Mike do you know any good ways of calculating initial guesses for n^(1/x)
 
Nope!
 
:(
 
8:02 AM
$n$ is probably a good initial guess.
 
so sqrt(10), i think ill guess 10!
my current one is n*(1/x)
 
@Darksonn Well, 10 gets me to $\sqrt(10)$ to 200 digits precision in 10 calculations.
 
what if the initial guess is 5?
 
9
 
hmm, lets see how long it takes it to calculate sqrt(10) to 15000 digits
 
8:06 AM
Once you get down to the right range it's going to go the same rate
it'll take exactly one more step than 5 does
 
yea makes sense
okay 15000 digits does take some time
 
It'll be roughly halving the number until it gets close enough to the root
 
i know, I'll restart it with some logging
 
@mike Do you need differential equations so far when reading Lee?
 
@WillHunting I'm going too slow, I'm still early on.
 
8:10 AM
Ubuntu 14.04 will be out in a few hours
 
wow, the calculations begin taking long time when the fraction's so exact
maybe i shouldnt be using 15000 digits, each step increases dramatically in size
 
:P
 
Wow, I can vote to close questions now, I feel so powerful, lol
@JessicaK It seems you have not left this chat since yesterday lol
 
100 digits is more managable
350402861630453040153103^(1/4) in 11 seconds
 
@mike I am a bit sad that the Springer books sometimes are bound poorly, pages seem to start to fall off
 
8:13 AM
also, I got the log function working
 
..
 
Hey wait, if a function is differentiable, it's derivative has the intermediate value property… is the derivative continuous though?
Oh never mind.
 
8:30 AM
this dosent seem right 13210^(1/7) with newtons method... @Mike
1886.85714285714285714286
1886.71428571428571428571
1886.57142857142857142857
1886.42857142857142857143
@Mike my convergence slows down the higher roots i try to calculate
@Mike Halley's method apparently has cubic convergence, wouldn't using this one be even faster?
 
@Darksonn Yep, I believe that
@Darksonn Probably. I'm not an expert on algorithms, far from it.
 
Hmmmmm, I'm going to try and implement halleys method
 
8:46 AM
Hello everyone.
 
hi
 
@Mike this halleys method seem to be as fast at cube roots as newtons is at square roots
 
try at fourth roots
 
Ooooh, there is a generalization of newtons and halleys method, which gives the fastest for the root x
ok testing fourth root, newton vs halley
halleys method: 1.4 seconds
newtons method: 8.7 seconds
@Mike
what does this mean: $(1/f)^{(d)}(x_n)$
specifically the $(1/f)$ part, $f$ is a function
 
dunno
I gotta go to bed
Night man
 
9:00 AM
okay, thanks for the help
 
9:29 AM
0
Q: n'th and (n-1)'th derivative of $x^n - a$

DarksonnI have a function of the form $f(x) = \frac{1}{x^n - a}$, and I need to programmatically find the n'th and (n-1)'th derivative of the function. Since the function has this specific form and that the power is tied to which derivative it is, I was wondering if there's an easier method than simply t...

 
 
1 hour later…
10:49 AM
Is there a proof of $\pi_n(x) \sim \frac{x}{\log(x)}\frac{(\log\log(x))^{n-1}}{(n-1)!}$ of approximately the same complexity as the prime number theorem, where $\pi_n(x)$ is the number of positive integers $\le x$ who have $n$ prime factors?
 
11:48 AM
Without context @N3buchadnezzar there can be no size.
 
Even though then size does not exists, it still matters
 
Only in theory, my friend :-)
 
yo
 
wazzup?
 
da ceilin
 
11:58 AM
da roof, da roof, da roof is on fire!!!
 
@Alyosha how old are you?
 
lol!
A fun topic of debate for the day: Should something be considered mistreatment if the being in question has never known any different? Alternatively, how do mistreated beings react to proper treatment after having known nothing else?

Related topics: human rights, animal rights, plant rights (is that a thing?), Aristotle, Plato's "Allegory of the Cave"
also dystopia, social change
 
@meer2kat specifically, you can't miss freedom or well-being until you experience it
 
@GabrielR. exactly my question. hence the allegory of the cave reference
 

 The Symposium

A Party Space for Philosophy.SE! Both philosophy and mundane c...
 
12:10 PM
place to debate?
 
yep
 
yay
 
@meer2kat When given the chance to avoid mistreatment, would a mistreated sentient being take it?
 
@KarlKronenfeld I moved the debate to the link that @skullpatrol posted :)
could you move there as well?

 The Symposium

A Party Space for Philosophy.SE! Both philosophy and mundane c...
 
12:25 PM
hi @sawarnik and @khallilbenyattou
 
@Karl is getting serius
 
@PedroTamaroff hop on over and join the debate :)
 
@meer2kat Good afternoon. I presume it's the morning where you are?
 
@KhallilBenyattou 8:30 am :)
@KhallilBenyattou It's what...1:30 PM for you?
 
gooooood morning everyone :)
 
12:32 PM
Cannot now. Tennis
 
@MickLH hi!
 
hey :)
hows things?
 
It's 9:30 am here,.btw.
 
5:30am here
 
@MickLH well. ypi?
*you?
 
12:34 PM
excited to put another day in on my arbitrary precision rational solver
 
Any way to get the roots to
$$
(z^2+1)^n \cdot i^n + (z^2-1)^n = 0
$$
where $n \in \mathbb{N}$?
In particular I am looking for the roots that lie within the unit circle..
 
@meer2kat It is indeed!
Is there any way to evaluate the following integral via complex partial fractions? $$ \int \dfrac{1}{x^2 + 1} \text{ d}x $$
 
..
 
...
 
....
:15007582 What are you doing so early?
 
12:43 PM
So far I have: $$ \begin{aligned} \int \dfrac{1}{x^2 + 1} \text{ d}x & = \dfrac{1}{2i} \int \dfrac{1}{x-i} \text{ d}x - \dfrac{1}{2i} \int \dfrac{1}{x+i} \text{ d}x \\ & = \dfrac{1}{2i} \log \left| \dfrac{x-i}{x+i} \right| + \mathcal{C} \end{aligned} $$ I know that the integral of a real valued function is a real valued function so how do I take the real part of this final result?
 
Feeling nice today, understood the rigorous definition of limits.
 
@Sawarnik @meer2kat @skullpatrol @MickLH
 
@KhallilBenyattou Hi :)
 
yo @KhallilBenyattou
 
@Sawarnik @skullpatrol Yo! Have you guys got any idea of how to proceed with my above integral?
Also, which text are you reading @Sawarnik that helped you grasp the rigorous definitions of limits? :)
 
12:47 PM
@Sawarnik I like to get up early, like I said I'm excited to work on my project
 
@KhallilBenyattou My book is the worst for understanding the defintion of limit! I mostly the idea learned it from Khan Academy, then by thinking upon it. My book just states the defintion and gives one proof using it :(
@KhallilBenyattou I haven't done any calculus with complex numbers! :/
 
Greetings
 
Greetings: "Super sis"
 
Good morning
 
Greetings, greatest one!
 
12:55 PM
Hi all :-)
 
@Chris'ssis Are you from US?
 
@Chris'ssis Hey :)
 
@Sawarnik No
 
@Chris'ssis Are you from the US?
 
@meer2kat Yes, it is mistreatment because the act is inherently wrong as it has the potential to cause suffering.
 
1:00 PM
@Sawarnik I've never liked Khan Academy.
 
@KhallilBenyattou @Sawarnik I'm not from US, but from a country in Europe.
 
@ParthKohli I know, i don't like it much as well. But its a good resort for just gaining the basics and study properly.
 
@N3buchadnezzar $$\left(\frac{z^2-1}{z^2+1}\right)^n = -i^n \iff \frac{z^2-1}{z^2+1} = \exp\left(\frac{(2k+1)\pi i}{n}\right)\cdot i =: \zeta_k$$ Then you get $$z^2 = \frac{1+\zeta_k}{1-\zeta_k}, \; 0 \leqslant k < n.$$
 
@ParthKohli there isn't much better on Youtube
 
@Chris'ssis Is the country in the United Kingdom?
 
1:01 PM
@KhallilBenyattou No, she told.
 
@685-252 MIT?!
 
@Chris'ssis That's pretty cool. I know a teacher who went to teach over there for a month very recently. He said it was nice over there :)
 
@KhallilBenyattou :-)
 
@ParthKohli he is a graduate from there
 
@685-252 No, I meant the MIT courseware on YouTube.
 
1:03 PM
@ParthKohli It's a bit boring, you know.
 
I particularly feel just like Vipul does here.
 
@ParthKohli I know, but he puts a lot of work into it, imho
 
I'm going to propose a mind-blowing question, it will seem impossible at first sight.
Compute elementarily, without pen and paper $$\lim_{n\to\infty} \frac{1}{n}\sum_{i=1}^{n} \sum_{j=1}^{n} \frac{1}{i+j}$$ It's a magnificent question!
5
 
@Chris'ssis I will use chalk and blackboard, lol
 
how about a pen, but i write on my hand, lol
 
1:09 PM
@WillHunting :-)))
 
Or a computer, maybe.
 
There is a very very easy way to compute it.
 
@Chris'ssis Yes, I'm aware that Wolfram is really easy. :)
 
$0$ ?
 
@GabrielR. It's not $0$.
 
1:16 PM
@Chris'ssis damn! I thought the double sum would be $2H_{2n}$
 
@DanielFischer Thanks!
 
@GabrielR. OK :-)
 
@GabrielR. Same as you, hence I particularly admire your knowledge in analysis.
 
@Chris'ssis I can't crack it. I tried a matrix point of view, but even summing diagonals fail.
@Alyosha are you in college ?
 
@GabrielR. No, I will be next year. You?
 
1:32 PM
@Alyosha The French education system being a bit different, I would say I'm a freshman. Do you do linear algebra as well ?
 
@GabrielR. Not officially in school, the British A-Level is literally just calculus, vectors and complex numbers (and a bit of differential equations).
I've done a bit outside school, though.
 
@Alyosha I'm from the UK too! high five Which university are you planning on going to?
 
@Alyosha How far have you gone in calculus ? Do A-Level classes (is it still considered part of high school btw ?) approach calculus with rigor (I mean epsilon-deltas, proofs...) ?
 
@GabrielR. No, ocr.org.uk/Images/67746-specification.pdf this is the specification. It's very routine. It's still considered high school, in America the equivalent of 12th grade.
 
@GabrielR. None, whatsoever. The most rigorous thing in the A-Level is a very brief look at the derivation of the derivative via first principles.
 
1:37 PM
@Sawarnik the problem you told me yesterday, does not need Holder Inequality, it can be handled with Chebyshev Inequality too
 
@KhallilBenyattou Cambridge.
 
@Alyosha @KhallilBenyattou ah, it's just like in French high schools then... boring
 
@Sawarnik Weighted AM-HM and Chebyshev Inequality.
 
Yes, I'd hate to be pre-university before the internet.
 
@Alyosha That's pretty cool. I myself am aiming for Warwick but they require STEP and I haven't prepared for it at all :P
 
1:38 PM
@Hawk Forgive me! I don't know that stuff. But did you give the answer?
 
@KhallilBenyattou It's not that bad when you get into the swing of it.
 
@Sawarnik 'Forgive'? Either you are very poor in english or being extremely humble, beyond my tolerance level...
@Sawarnik Yes, I will answer it...
 
@Alyosha I reckon just over 2 months should be enough to crack it!
I've done a few questions here and there, but they were pretty nice integrals and differential equations.
 
@KhallilBenyattou I've not started serious revision for it yet, although I've done enough to find the questions becoming a little repetitive.
 
@Hawk Nice :) You may get a bounty.
 
1:40 PM
@Alyosha are A-level high schools that selective ? Is it only for gifted students? Is it possible to enter OxBridge without them ?
 
@GabrielR. Pretty much everyone does them. There used to be S-Levels (of which the STEP mathematics paper is a remnant of), which were for the best people, but they were stopped a few decades ago.
 
@Sawarnik Can you please post me the link to the problem again?
 
@Hawk Check r9m's most recent question.
 
So when You say 'pretty much a freshman', you mean you take these high-school selective courses?
 
@Sawarnik Okay...thanks!
 
1:43 PM
@Alyosha Thanks for the info. Yeah I took a special math course for advanced students when I was in high school two years ago. Now I'm doing this en.wikipedia.org/wiki/…
 
@GabrielR. Looks excellent. When are you applying to les grandes ecoles?
(Apologies, I stopped French two years ago)
 
@Alyosha I already am. Exams are next week actually :(
 
Ah, we have two months.
 
@GabrielR. Bonne chance!
 
@KhallilBenyattou thanks
 
1:47 PM
@Alyosha I find your questions and answers very nice :) You ll definitely pass.
 
r9m
@Chris'ssis is it 2 ?
 
@Sawarnik Thank you, I'll try to ask more.
 
@r9m $$\lim_{n\to\infty} \frac{1}{n}\sum_{i=1}^{n} \sum_{j=1}^{n} \frac{1}{i+j}=2\log(2)$$
@r9m by the way, you need to know nothing about integrals to finish it.
 
Probably easy, but $$\lim_{n \to \infty}\frac{1}{n}\sum_{k=1}^n\left \{ \frac{n}{k} \right \}^2.$$
 
@Alyosha Yeah, it's elementary.
$$\lim_{n \to \infty}\frac{1}{n}\sum_{k=1}^n\left \{ \frac{n}{k} \right \}^2=\log(2 \pi)-1-\gamma$$
 
1:52 PM
@Alyosha What do the 'curly' parentheses mean?
 
I'll post it later on.
 
Fractional part
 
brb
 
@Chris'ssis did you use the polygamma function to solve the other problem ?
 
@GabrielR. Definitely no. High school knowledge only.
 
1:56 PM
Indeed, I didn't see at first how easily it turns into a Riemann sum.
 
brb
 
$$\int_0^1 \left \{ \frac{1}{x}\right \}^2 dx$$
 

« first day (1352 days earlier)      last day (3676 days later) »