« first day (1966 days earlier)      last day (3350 days later) » 

17:00
does cos^4(x) - sin^4(x) = (cosx - sinx)(cosx + sinx)(cosx - sinx)(cosx + sinx) ?
@KinsonChan Expand in exp functions.
@DanielSank exp?
Use the fact that $\cos(x) = (1/2)(e^x + e^{-x})$.
By exp I meant $e^x$, i.e. the exponential function.
Have you learned that yet?
Huy
Huy
@DanielSank Is that a fact? I've never seen that equation before.
@Huy Of course! It's extremely important. It comes from $e^{ix} = \cos(x) + i \sin(x)$.
Read about Euler's identity
Huy
Huy
17:06
@DanielSank: Now I don't know if you're trolling or not.
Check your exponents. :P
Crap, I made a typo.
sounded like a piece of rush hour
Huy
Huy
sorry, I was just being pedantic before.
@KinsonChan I made a mistake.
It's $\cos(x) = (1/2)(e^{ix} + e^{-ix})$.
Huy
Huy
I think the cosh is defined that way though. :D
17:07
@Huy Yeah yeah.
I've been making a lot more typos than usual lately.
It's weird.
u guys are going WAY too far
im pretty sure u dont have to go that deep in order to find the answer for a school textbook
wat was the question @KinsonChan
@KinsonChan Perhaps but that identity makes questions like that one super easy.
does cos^4(x) - sin^4(x) = (cosx - sinx)(cosx + sinx)(cosx - sinx)(cosx + sinx) ?
You can always just fall back on iteritive uses of various trig identities, but if you use the exp functions you don't have to remember all those identities.
I'm a physicist, we survive by remembering a small number of basic things :P
17:10
@KinsonChan nope!
and im just studying double angle formula right now -.-
so what should it equal?
@KinsonChan Why don't you try to expand it out yourself?
i did
but im not sure if it is correct
cos^4(x) - sin^4(x) = (cosx - sinx)(cosx + sinx)(cosx^2 + sinx^2)
the second part is yet expandable
so you cant expand the second part?
17:12
@Agawa001 That can't be right.
The third factor is identically one.
ok this is the question: cos^4(x) - sin^4(x) = cos2x
^^^ original question
@KinsonChan That's not a question :D
you need to prove that the equation equals
@KinsonChan Ah.
using double angle formula
i was told that i should expand the left side
17:13
Ok, what's the double angle formula?
lol
its trigonometry
Huy
Huy
hey @BalarkaSen, you there? got a few easy questions for you
@KinsonChan I know that.
I'm here.
Write it down!
@KinsonChan You can do this yourself if you write down what you know.
Huy
Huy
17:14
@BalarkaSen: do you know how to prove that a hyperbolic metric with totally geodesic boundary on $\Sigma_{0,3}$ is uniquely determined by the boundary components?
Huy
Huy
ok
@DanielSank sin2a = 2sinacosa cos2a = cos^a - sin^a , 2cos^2a - 1, 1 - 2sin^2a
@KinsonChan What do the commas mean?
Huy
Huy
also, my prof defined the mapping class group as the quotient of orientation preserving homeos up to isotopy, whereas on wiki it's just homeos. that should make a difference, right? @BalarkaSen
17:15
means theres more than one
@DanielSank cos2a has 3 formulas
@DanielSank its right
@KinsonChan So they're equals signs :)
me too i got decieved firstly
u guys are confusing me
one of these mysterious triangular properties
17:17
so how should i solve it?
@Agawa001 Ok, I guess the cross terms all work out...
@DanielSank yes and sec pat is 1 indeed
@Huy Yes, it does make a difference. Your prof is working on the oriented category, while wiki is working with the unoriented category.
Huy
Huy
@BalarkaSen: can you elaborate on the difference it makes? if I always assume an oriented manifold, does it make a difference?
@KinsonChan ur convo with john sounded like an excerpt from rush hour im still laughing
17:20
@Agawa001 xD
@Huy Consider $S^1$. If you work with orientation preserving homeomorphisms, all of them are isotopic to the identity.
@KinsonChan You've got $\cos 2 a = \cos(x)^2 - \sin(x)^2$.
However, in the wiki definition, the antipodal map (an orientation reversing homeo) is not isotopic to identity.
Do you have chatjax turned on?
Huy
Huy
Hm.
17:21
@DanielSank what is chatjax? and no
Go here and drag the "start chatjax" link to the tab for this page.
Huy
Huy
@BalarkaSen: but they are both "common" definitions for the MCG then?
so not a typo or so
New topic: how does one treat $\delta ( f(x))$ where $f$ is a function in $N$ dimensions and $x$ represents an $N$ dimensional variable?
In general, the equation $f(x)=0$ defines an $N-1$ dimensional manifold.
@Huy I said something false above. Of course not all of them are isotopic to identity. But what I mean is that the wiki defintion gives you a bigger group, 'cause the antipodal map has a different isotopy class.
Therefore, $\delta(f(x))$ should be a function which is 1 on that manifold and zero elsewhere.
Huy
Huy
17:24
@DanielSank: isn't the delta function just generalized by the product measure or you can go the distribution way?
How then, would one integrate over, say, just one of the variables? Is there some way to factorize the expression so that variables can be integrated one at a time?
@Huy I don't know what you mean.
Meh, no, what I said was right. Every + orientation homeomorphism is isotopic to the identity. Descend down to MCG of [0, 1].
Apologies, working with too many things at once.
@KinsonChan did it work?
@Huy Yes, sure.
@DanielSank yea
Huy
Huy
17:26
@DanielSank: you are talking about that "function" that satisfies $$\int_{-\infty}^\infty f(x) \delta(x) \, dx = f(0)$$ or the physicists' definition $$\delta(x) = \begin{cases} \infty & x = 0\\ 0 & x \neq 0\end{cases},$$ right?
MCG doesn't just mean you have to consider homeomorphisms or anything like that. You can think about orientation preserving homeoms, diffeoms, homotopy equivalences, and whatever you want.
@DanielSank so howd u do it? :/
@Huy Yes of course.
@KinsonChan Yes.
It depends on the category you're working with.
@DanielSank so howd u do it? :/
17:29
Take the equation I wrote down and think about how to get more powers of $\cos$ and $\sin$...
I don't want to do your homework for you.
Suppose I had $a^2 - b^2$ and I want to get $a^4 - b^4$, what would you do?
Huy
Huy
@DanielSank I'd be careful when calling that thing "a function". I think you're right that this is the property the measure will have, but I'm no expert with measure theory.
@Huy Who said anything about measure theory?
Let's just treat the mathematical objects as things with properties.
Huy
Huy
well it is either a measure or a distribution?
@Huy Let's try a simpler case.
Go to one dimension. $\delta(f(x)) = \sum_r \delta(f'(r)(x-r))$ where $r$ are the roots of $f$.
This is obviously $\sum_r \delta(x-r)/|f'(r)|$ by thinking about how a delta function behaves in an integral and using a change of variables in the integral.
When we upgrade to $N$ dimensions, that sum over the roots must become some kind of integral over the $N-1$ manifold and the $f'$ factor is probably something related to the gradient of $f$.
Huy
Huy
yes.
17:35
Here's my attempt to work it out:
$\delta(f(x)) = \sum_{\text{"roots" }r} \delta( \sum_i (\partial_i f)(r)(x_i - r_i))$
@Huy with me so far?
Huy
Huy
somewhat, yes.
I'm just Taylor expanding $f$ about each root.
The $f(r)$ part (0th order in Taylor expansion) doesn't show up because it's zero by construction.
u guys are like doing math with english
Ok @Huy here's where I get mixed up. In the $\delta$ function we've got an expression which involves more than one variable.
I'm not entirely sure what to do with this.
Normally if I have $\delta ( \vec{x} - \vec{\mu})$ I would know to just write this as $\prod_i \delta(x_i - \mu_i)$.
Ahhhhhhhh
This is somewhat interesting.
$\delta(\vec{x} - \vec{\mu})$ has a vector as its argument.
That conversion to a product of deltas of scalars makes sense... but why?
I suppose it's obvious.
But we could do this:
Huy
Huy
I don't think you can multiply distributions, in general.
17:42
$\delta(\vec{x} - \vec{\mu}) = \int \frac{dk_1}{2\pi}\cdots \frac{dk_N}{2\pi} e^{i \vec{k} \cdot \vec{x}}$.
@Huy Sure we can.
anyone have a second?
Anyway, continuing my line of reasoning, that multiple integral factors to $\prod_i \int \frac{dk_i}{2\pi} e^{i k_i x_i}$.
(cos^2x - sin^2x)^2 = cos2x doesnt equal right?
Huy
Huy
Distributions may be multiplied by real numbers and added together, so they form a real vector space. Distributions may also be multiplied by infinitely differentiable functions, but it is not possible to define a product of general distributions that extends the usual pointwise product of functions and has the same algebraic properties.
Which is just $\prod_i \delta(x_i)$ (oops, I dropped the $\mu$'s).
@KinsonChan I tried to help you with this. Did you ignore my last hint?
17:44
i did what you asked
and im stuck at (cos^2x - sin^2x)^2 = cos2x
@KinsonChan Ok, so how do turn $a^2 - b^2$ to $a^4 - b^4$?
That's what I asked.
What do I multiply the first thing by to get the second?
by doing squaring
squaring
@KinsonChan Not quite.
Try it. Square it and see what you get.
i did
and i got a4 - b4
nvm i didnt
That's definitely not right.
17:46
@DanielFischer Hey!!! Let $A: \mathbb{R}^n \to \mathbb{R}^n \\ A(x_1, x_2, \dots)=(a_1 x_1, a_2 x_2, \dots)$. In order to show that A is a linear mapping is the following sufficient?
Let $x=(x_1, x_2, \dots), y=(y_1, y_2, \dots) \in \ell^2(\mathbb{N}), \lambda, \mu \in \mathbb{R}$. $A(\lambda x+ \mu y)=(a_1(\lambda x_1+ \mu y_1), a_2 (\lambda x_2+ \mu y_2), \dots)=(\lambda a_1 x_1+ \mu a_1 y_1, \lambda a_2 x_2+ \mu a_2 x_2, \dots)=\lambda (a_1 x_1, a_2 x_2, \dots)+ \mu (a_1 y_1, a_2 y_2, \dots)= \lambda Ax+ \lambda Ay$
@KinsonChan sometimes u can see people doing maths with elvish. its called topology
@KinsonChan How would I go from $a - b$ to $a^2 - b^2$?
@Agawa001 its ok, i got used to that everyone here are einsteins
Can you think of a way?
erm
i give up
17:47
@Huy So as you see, $\delta(\vec{x} - \vec{\mu})$ turns into $\prod_i \delta(x_i - \mu_i)$ rather naturally.
@KinsonChan Try $(a-b)(a+b)$ and see what you get.
i get a^2 - b^2
but that doesnt apply to ur application
Now how would you go from $a^2 - b^2$ to $a^4 - b^4$?
@KinsonChan I don't know what "ur application" refers to.
^^^
that
(a^2 - b^2)(a^2 +b^2)
17:50
Go on...
= a^4 - b^4
?
but im not done yet
17:52
No, but you're really close.
im at (cos^2x - sin^2x)^2 = cos2x
which doesnt make sense because they dont equal
@KinsonChan Where did that come from?
Also, would you mind please using mathjax so your math is easier to read?
from my original equation
i dont know how to use mathjax :/
like i dont know how to type it
this is the original equation: cos^4(x) - sin^4(x) = cos2x
i need to prove they equal
Huy
Huy
I think this is a proof for what you're trying to show/see @DanielSank
@KinsonChan Right, I just showed you how to link the difference of two squares to the difference of two fourth powers. Use that.
17:56
@DanielSank which got me to (cos^2x - sin^2x)^2 = cos2x
@Huy What does that star mean?
where im stuck
@KinsonChan No, that's wrong.
Go back to the $a$'s and $b$'s.
17:56
oh i see
thanks
You're welcome.
See, you can figure it out yourself with a little bit of fiddling around.
Huy
Huy
@DanielSank: It's a pullback on distributions, but when you apply it to a continuous function, it is just composition.
@DanielSank YAY!
@DanielSank thank you
@KinsonChan You're welcome. Next time when people give you a hint, please assume they are trying to help you.
@DanielSank will do!
17:59
@KinsonChan Good. Hints are meant to help you learn ;D
@Huy So what does $\rho^* \delta_0$ mean?
@Huy Yeah but what's $\delta_0$?
Is that a delta function at the point 0?
Huy
Huy
@DanielSank: the Dirac measure, yes.
Huy
Huy
(as a product of measures)
sorry, I'm not enough of an expert in this area to explain it in a different way from the one I know
if you want to read about it yourself, see Theorem 6.1.5 on page 136 in Hörmander's Analysis of Linear Partial Differential Operators I.
18:07
Ok this is quite helpful.
Huy
Huy
it is exactly what you're trying to show, just in a more mathematical notation, I guess.
@DanielSank you still there?
@KinsonChan Yo.
sin2x + sinx = 0
i need to find the exact value of x
0 < x < 2pi
@KinsonChan What have you tried?
18:12
2sinx = -tanx
stuck there
not sure if thats the correct path tho
i expanded the sin2x to 2sinxcosx
Perhaps apply $\arcsin$? :P
havent learnt that yet @Danu so it could be completed without it
@KinsonChan For how many values of $x$ do you know an exact expression for $\sin(x)$?
I'm guess it's essentially four.
@DanielSank howd i apply mathjax again?
howd u make it stay when i close the page
You could just write $0=\sin (some stuff...)$... (effectively the same as applying $\arcsin$)
Just click the thing again, it'll load it again.
You have to reload every time.
18:15
what thing?
ChatJax
i dont have the link
It's on the top of the starred board
on the right
the problem is i dont know what sinx =
it only gave me the range
range of x
The goal is to find $x$ such that $\sin(x) + \sin(2 x) = 0$, right?
18:17
yes
x with a range of 0 <_ x <_ 2pi
How many values of $x$ do you know such that you are able to even write the value of $\sin(x)$?
For $0 < x < 2\pi$ I only know 12 possible values.
i checked the answers and there is 5 possible answers for x
That's not even very many to check.
@DanielSank I think that's pedagogically really bad, dude :P
I need to know 5/7 = 42 ??
18:19
@Danu Why?
@lopata its 21
ok thanks
I can't think of another way to do the problem, and if you draw a circle and think about the expression geometrically the solutions become obvious and clear.
How is that pedagogically bad?
@lopata i rait ure question a perfuct 5/7
Huy
Huy
reddit is leaking
18:21
@DanielSank we are taught to do it with the cricle
5/7 thanks m8
but i dont know how to apply it here
5/7=7/7=21
Huy
Huy
what about with rice?
u forgot 21/7 = 3
3 = 3 side triangle
triangle = illuminati
illuminati confirmeeddd
18:22
:o
MATH BITCH
@KinsonChan Draw a circle and draw the points at angles $\pi/3, \pi/4, 2\pi/3, \ldots$.
my life just changed
@DanielSank howd u come up with those radians?
@KinsonChan They're the only angles for which I know the value of $\sin(x)$.
18:23
@DanielSank tell me how u found the values of sinx
@DanielSank You don't know $\pi/6$?!
Come on, then you're missing out on the nicest heuristic ever.
Huy
Huy
^
$\sin 0 = 1/2 \cdot \sqrt 0$, $\sin \pi/6 = 1/2 \cdot \sqrt 1$, $\sin \pi/4 = 1/2 \cdot \sqrt 2$, $\sin \pi/3 = 1/2 \cdot\sqrt 3$, $\sin \pi/2=1/2 \cdot \sqrt 4$
and the reverse for $\cos$
@Danu Oops, forgot that.
@Danu Cute.
@DanielSank This was amazeballs when I found out at age 15 :)
18:26
@danu what are those numbers again?
something to do with perfect triangles?
@Danu My personal favorite is how you can use matrix multiplication of two rotation matrices to recover the angle addition formulae.
@DanielSank Neat, too.
@evinda Yes, it is. I assume your first line is a typo and should read $A colon \ell^2(\mathbb{N}) \to \ell^2(\mathbb{N})$.
a colonel?
@Danu You put the square root on the wrong numbers, $1/2\cdot\sqrt{0},\, 1/2\cdot\sqrt{1},\, 1/2\cdot\sqrt{2}$ etc.
18:34
Yes, I am sorry... I don't know what I wrote wrong.. Thank you!!! @DanielFischer
@DanielFischer I heard you know everything. I am trying to understand why $\delta(f(x)) = \int_{p|f(p)=0} \left( \prod_i \delta(x_i - p_i) \right) / |(\nabla f)(p)|$.
@DanielFischer Hahaha!! Great.
I can prove $\delta(\vec{x} - \vec{\mu}) = \prod_i \delta(x_i - \mu_i)$ (using Fourier transform), but I'm stuck in the case with a function inside the delta function. See above transcript if you're interested.
@DanielSank I know that I know nothing. Or so. I suppose $\delta$ should be the Dirac distribution? What assumptions do you have about $f$?
@DanielFischer Let's live in Happy-Land where $f$ is as nice as we want.
The $\delta$ means Dirac delta function.
or distribution... whichever you like.
18:38
^say distribution when not* around physicists
I am a physicist.
in a math chat room :P
When in Rome...
Your first comment lacked the not*.
I know...
I don't want to get into function vs. distribution right now. As far as I care, $\delta(x)$ means $\int (dk/2\pi) e^{ikx}$, and don't do the integral over $k$ until the end of your calculation :D
18:40
can someone tell me why sin(30) = 1/2 ?
And if you want, add an imaginary part to $k$ to make things nicer.
when i put sin(30) in a calculator it doesnt equal to 1/2
@KinsonChan That's in degrees. Mathematicians/physicists usually work in radians
Anyway, none of this matters for the problem at hand.
@Danu You should slow down, bro :)
Seriously
typos all around
Don't usually make that many.
Huy
Huy
18:42
well it is Christmas time
Anway, we can Taylor expand $f$ near its zeros to get:
$\delta(f(x)) = \int_{p|f(p)=0} \delta(\sum_i (\partial_i f)(p) (x_i - p_i))$.
Using the Fourier representation of the $\delta$ we can replace the $\delta$ with:
$\int (dk/2\pi) \exp \left[ i k \sum_i (\partial_i f)(p)(x_i - p_i)\right]$
Compare this to $\delta(x - p)$ whose Fourier transform is
@DanielSank Okay. The integral still doesn't make sense since I'm not a physicist ;) The point is that you extend composition with $f$ by analogy - for nice enough $f$. If you have a distribution that is given by a suitably nice function $g$, you can compose that function with $f$ and get a distribution on the domain of $f$: $$\int_U g(f(x))\cdot \varphi(x)\,dx.$$ Now if $f$ is a diffeomorphism $f \colon U \to V$, you can make a variable transformation $y = f(x)$, and the integral can be written
$$\int_V g(y)\cdot \varphi(f^{-1}(y))\cdot \lvert \det D(f^{-1})(y)\rvert\,dy.$$ Then you generalise that to arbitrary distributions. If $T$ is a distribution on $V$ and $\varphi$ a test function on $U$, you define $$(T\circ f) \colon \varphi \mapsto T\bigl[ (\varphi\circ f^{-1}) \cdot \lvert \det D(f^{-1})\rvert\bigr].$$
Set $T = \delta$.
Question for you all
So I have shown that any square matrix $A$ can be written in the form $A = S + K$, $S$ being symmetric, $K$ being skew-symmetric
Any hints of how I can show uniqueness of $S$ and $K$ for a fixed $A$? There's a hint to use the fact that the zero matrix is the only matrix that is both symmetric and skew-symmetric
$\int (dk_1/2\pi)\cdots(dk_N/2\pi) \exp \left[ i k \cdot (x - p)\right]$
(here $x$, $p$, and $k$ all live in $R^N$)
@Clarinetist You know that sums and differences of symmetric resp. skew-symmetric matrices are again symmetric resp. skew-symmetric I suppose. Assume two representations, and fiddle a bit with the equation until you get something where the one side is symmetric and the other skew-symmetric.
Huy
Huy
18:50
@Clarinetist: $0 = (S - S') + (K - K')$
@Huy I've tried playing around with that. So because $0$ is symmetric, we have $$0^{T} = S^{T} - (S^{\prime})^{T} + K^{T} - (K^{\prime})^{T} = S - S^{\prime} - K + K^{\prime}$$
@DanielFischer I don't understand the part where you "define" the meaning of the third equation.
Huy
Huy
@Clarinetist: Just take one of the brackets to the other side and then apply your hint.
@Huy OH, I see now. Thank you @DanielFischer and @Huy!
What an interesting proof technique.
@DanielSank If $T_g$ is the distribution defined by $g$, that is, $T_g(\psi) = \int_V g(x)\psi(x)\,dx$, we define $T_g \circ f = T_{g\circ f}$. (If $f$ is nice enough.) We can do that because we know how composition of functions works. Then we look at the formula we get for $T_{g\circ f}$ by the change of variables, and use that formula to define $T\circ f$ for an arbitrary distribution $T$.
18:55
@DanielFischer I understand the meaning of what you've written. I guess I would avoid "defining" anything by regarding the $\delta$ as a limit of something already defined, but the idea is the same.
@DanielFischer Yeah, I see what you mean.
Again, in my little physics brain $\delta(x)$ is short hand for a limit (which doesn't need special defining).
Huy
Huy
hey @Mike
@DanielSank Yes, you can also take a sequence/net/filter of nice functions converging to $\delta$ and say $\delta\circ f$ is the limit of $g_\alpha\circ f$. Of course one needs to check that that works out (is well-defined).

« first day (1966 days earlier)      last day (3350 days later) »