« first day (2823 days earlier)      last day (2184 days later) » 

12:24 AM
guys, is it true that since we have a uniform bound for $f_\delta$, we just have a constant function (the uniform bound), which is therefore integrable? because the bounded convergence theorem doesn’t even require a uniform bound right, it just wants $f_\delta(x,t)\leq g(x)$ for some integrable $g(x)$ (so just checking)
 
0
Q: Determine stability of the origin in a system with two zero eigenvalues

ALannisterI need to determine whether the fixed point $x^{*}=0$ of the system $$\dot{x} = 0 \\ \dot{y}=x $$ is attracting, Liapunov stable, asymptotically stable, or none of the above. For reference, a fixed point $x^{*}$ is said to be attracting if any trajectory that starts within a distance $\delta$ ...

 
actually, I'm pretty much convinced that's it. (fight me otherwise)
 
It's quiet in here ronight
tonight
@Semiclassical hey brah.
 
If I have a system with 2 zero eigenvalues, what can I say about the stability of the origin as a fixed point?
 
12:34 AM
What I notice in your example is that the trajectory is a vertical line
 
How would I know that? I'm still very confused about how to find trajecgtories
 
With the closest approach to the origin thus being at the time when y(t)=0
Your solution has x(t)=x0
That’s a vertical line.
 
But what about the y part?
 
That’s what makes it a line and not a point?
 
o..o
 
12:37 AM
Really? Because I thought x(t)=x0 was a vertical line independently of what y(t) is
 
Not if y(t)=y0 as well
Which only happens in your case if x0=0
 
it seems to be that if x0 is not zero then the trajectory is a line, and if x0 is 0 then it's a point
 
I really don't understand trajectories and my book is kind of useless too.
It's Strogatz
 
Hmm. Yeah. Which is kinda goofy
 
well, just find the solutions of the differential equation ?
x is clearly constant
so x(t) = x0
 
12:39 AM
@mercio I already did that
 
well then you draw them ?
 
It's x(t) = x0, y(t) = x0t+y0
 
And the y-velocity is proportional to the initial x-coordinate
 
I don't know how to draw those two
 
pick your favorite x0 and y0
like idk,
x0=41 and y0=3
 
12:40 AM
both equations together? separate?
 
well you usually draw them on a plane with x and y axis
and plot all the points x(t), y(t)
 
Like do I plot x(t) = 41 and y(t) = 41t + 3 on the same graph?
Or do I plot the points x(t), y(t) as ordered pairs for different t's?
 
It’s a parametric curve. So the latter
 
when t = 0 you have 41 and 3 so you put a dot on the point (41,3) of your refernce frame
 
OOOh ok
 
12:42 AM
(i'm not sure that's the correct word)
then you put another dot at the point corresponding to t=1
 
I’d also note that the equations amount to the x-coordinate and the y-velocity being equal and fixed
 
and you put a dot for all the points for each t
 
One guys said there's a line of fixed points at x=0, so it's neither. Somebody else said it's unstable
 
yep
they are all right
 
And someone else said it's an isolated fixed point
 
12:44 AM
I am less sure about that
 
Exactly.
 
“No it is, or no it isn’t??” “Yes.”
 
The isolated part really knocked me for a loop, because if there's a line of fixed points at x=0, there's no way it could be isolated
 
The statement about x=0 being a line of fixed points is definitely true tho
 
@Semiclassical Clue the Movie?
 
12:45 AM
Yuuup
 
Fine film :)
 
well the good thing about math is that if you actually do the work you get to discover when someone else is wrong
 
Yeah. Haven’t watched it lately though
 
@mercio until Semi explained to me it was a parametric graph, I had no idea how to do the work.
I head they were going to remake it. Big mistake.
 
I’m trying to think if there’s a physics system which would obviously work like this one
It’s not obvious there is, though
 
12:47 AM
physics systems usually talk about the acceleration of things
also I think I saw @Secret a few minutes ago
 
Well, you can write the y equation as y-acceleration = 0
 
i have to ask him if he got to look at the real points of those tangents of slope i to a real ellipse
 
So that much is consistent. But you also would need the x-position = y-velocity and I don’t think you’ll get that from forces
 
so it's a bunch of rocks freefalling in space with carefully designed speeds ?
 
Maybe from a fluid flow system
If you only worry about a finite range of x, I think you can understand it as...hmm
 
12:50 AM
And soon your irregularly scheduled ordinal collapsing function (which may be fairly lengthy, so brace yourself)...
 
It’s either Poisoiulle (sl) flow or Couette flow
 
Also how has everyone been?
 
nonono I want people to look at the tangents of slope i to real ellipses
 
Ah, Couette flow
With the additional wrinkle here being that both the top and bottom plates are in motion
 
Assume there exists a weakly compact cardinal $K$. Cardinals are treated as their initial ordinals and greek letters denote ordinals.
$${\rm cl}(A)=A\cup\{\sup(B) ~|~B\subset A\} \\B_0(\alpha,\beta)= C_0(\alpha,\beta)= \beta\cup\{0,1,K\} \\B_{n+1}(\alpha,\beta)= \{\gamma+\delta,\omega^\gamma, \Psi(\eta)~|~\gamma,\delta, \eta\in B_n(\alpha,\beta) \land\eta\in\alpha\}\\ C_{n+1}(\alpha,\beta)=\{\gamma+ \delta,\omega^\gamma,\Psi(\delta) ,\psi_\delta^\gamma(\eta)~|~ \gamma,\delta,\eta\in C_n(\alpha ,\beta)\land\eta\in\alpha\} \\B(\alpha,\beta)=\bigcup_{n\in\omega} B_n(\alpha,\beta)\\ C(\alpha,\b
 
12:55 AM
o..o'
the message got cut off so latex can't render
 
Well that's good for everyone who doesn't want to see it x'D
@Secret and no I do not plan to write some nice fundamental sequences for this mess of a monster
Also you can construct some nice extremely large numbers as follows: Let $\varepsilon_{K+1}=\sup(B(0,0))$. Define the following function:
$$H_\alpha(n)=\begin{cases} n,&\alpha=0\\ H_\beta(2^n),& \alpha\ne0,\beta=\max(C_n( \varepsilon_{K+1},0)\cap\alpha)\end{cases}$$
Then $H_\alpha$ is a hierarchy of $\mathbb N\mapsto\mathbb N$ functions that grow extremely fast.
 
You also need to pick different x0's and y0's, right? Otherwise, how would you get the origin at all?
OOh, each choice starts you out on a different line
then, you vary t to trace out the line
And then you have all these lines, and so you get a whole plane of fixed points
 
1:10 AM
Not quite. (1,0) isn’t a fixed point for instance
But every point on the line x=0 is a fixed point
(So a line of fixed points not a plane of them)
An imperfect analogy is to consider a slow-moving river with a strong wind blowing
The top of the water will be moving at a fairly constant speed due to the wind, while the bottom will more-or-less be at rest
 
ooh, is Semi getting theatric?
 
So if you were to cast a fishing line into the river, then the bait will move with the river, and the it’ll move faster the closer to the surface you are
 
my book said if you have two zero eigenvalues, you have a whole plane of fixed points
lies
 
...huh
Context?
You certainly have a (1D) subspace of fixed points
 
It says "on the other hand, if $\lambda = 0$, the whole plane is filled with fixed points"
 
1:20 AM
What example do the have?
On the face of it, that seems obviously bogus
 
I think it means in a general case, because they're all examples saying "sketch a typical phase portrait in the case that"
 
It's saying then though that the system is $\dot{x}=0$
So maybe I'm confused.
Whatever. I'm going to go think about this some more.
So, it's not a whole plane of fixed points, ,but it is a whole bunch of lines, right?
 
I mean, if you’ve got a system dot-v = Av with A=0 then yeah it’s a plane fixed points
But A=0 isn’t equivalent to having two zero eigenvalues
 
But if $A$ has o trace and 0 determinant but is not necessarily the zero matrix
right
I hate it when I get "user was removed" reputation changes
 
1:24 AM
It is if you restrict to symmetric A, but that’s not a typical assumption
 
And that's not the case in this particlar problem either
 
Right
 
Okay, so line of fixed points, but the other lines are not fixed points, they're just trajectories
 
Right.
 
Great. Think I got it now.
 
1:28 AM
And since x doesn’t change, you’re always the same horizontal distance from the y-axis
Ergo, a vertical line
 
And the origin is definitely not attracting, because none of the other trajectories are converging to it.
 
What I don't understand tho is why it's not Liapunov.
 
dunno the definition of that
Liapunov stuff is something that I don’t remember
 
It means that nearby trajectories remain close for all time
It makes me think of a Liopleurodon. Like from the old Charlie the Unicorn video.
 
1:31 AM
Ah. Well, if you start at (1,0), then one unit of Time later you’re at (1,1) etc
So you get farther and farther away
 
Ah, without any periodicity or cycling back
 
If you start st (0.1,1) it’ll happen ten times slower but you’ll eventually get away
 
I'm going to say something I always want to smack people for saying: "kewl beanz"
 
It also doesn’t help if you start at (1,-3). Your distance to the fixed point is initially decreasing, but ultimately you’ll pass it and start moving away
The other fixed points are a bit of an exception since you stay at the same point and thus the same distance for all times
But I’m guessing Liapunov as stated here is all or nothing
 
Well, it's only asking about the origin
 
1:38 AM
Well, I mean that if you restricted your choice of initial points to x0=0 then these (boring) trajectories would satisfy the Liapunov condition
But you’re not, so it’s a moot point
 
 
1 hour later…
2:53 AM
UGH PDES ARE STUPID!
 
@XanderHenderson context?
 
I am TAing an upper division class on PDE
it is titled "Introduction to Partial Differential Equations"
but I think it should more properly be called "Ad hoc: The Musical"
 
(obviously, I am not a PDE guy)
at 9 o'clock tomorrow morning, I am supposed to lecture on the wave equation
when did the guy who is actually teaching the course tell me this?
oh, about 30 minutes ago
it is currently 8 PM my time
and I want to go to sleep
(which I am going to do very soon now)
 
2:59 AM
and my poor students are going to get a sh*tty lecture in the morning
 
What all are you supposed to talk about?
 
(and that, honestly, is what is pissing me off; the university asked a post doc who doesn't give a sht to teach this class, and sht flows downhill to the graduate student TAs)
 
That, I buy
 
@Semiclassical I think I have it more or less figured out; he wants them to solve equations of the form $u_{xx} + a u_{tt} = 0$ with various initial conditions, using the method of separation of variables
 
Ah
So no Fourier transform stuff
 
3:01 AM
no, that was last quarter (I guess)
the problem is that I have never taken an undergraduate PDE class
 
are you following a textbook?
 
I took graduate PDE, so, like, I can prove existence and uniqueness of solutions in certain circumstances, but, like, who ever actually solves a PDE?
seriously...?
 
lol
Physicists
 
@s.patroller The book is by Pinchover and Rubenstein
it isn't great, but it isn't terrible, and there are a lot of examples
 
So basically I’d be a good person to give the lecture tomorroe
 
3:03 AM
anyone but me ;)
the other TA for the class has an 8 o'clock lecture, so I might drop in just to annoy him
he actually knows PDE
 
Main thing I remember is that in a lot of problems it comes down to writing the Fourier series for one of the boundaries and then using sep of variables to get the full solution by superposition
Which in turn is pretty close to how the 1D Schroedinger equation works, with the big difference being how the time derivatives enter in
 
@Semiclassical in this case, the primary instructor expects them to use separation of variables, which results in a (just barely) coupled system of ODEs of the form $f'' - \lambda f = 0$ and $g'' - \lambda g = 0$, subject to some initial conditions
 
Sure.
 
the ODEs are tractable, and the initial conditions give you everything else you need
 
Rectangular region
 
3:07 AM
yes, exactly
 
You can do it in other coordinate systems of course
And you get lots of ODEs in that way
 
I seem to recall that
 
One thing you may want to emphasize is the role of lambda
Eg you probably don’t need solutions with positive lambda, but you may well need one with zero lambda
 
how do you mean? Like, as far as I am concerned, it is just one more unknown that one ultimately has to figure out
oh!
yes!
indeed; this is why tomorrow's quiz is going to rock their socks
the only non-trivial solutions occur when $\lambda < 0$, which is a minor headache :\
I'm sort of tempted to say "Hey... uh... hint: $\lambda \ge 0$ leads to trivial solution... uh... ignore those cases."
 
Well, for positive lambda you get real exponentials as solutions
Which don’t oscillate and therefore won’t help much
 
3:12 AM
yup, and for negative $\lambda$ you get imaginary exponents which (though the magic of Euler) turn into sines and cosines
 
Right
The reason you might need lambda=0 is when the average value of the initial profile of the wave is nonzero
Since that’ll give a constant term in the Fourier expansion
 
The other part of this is how the boundary conditions on the endpoints determine what kind of Fourier series you want
Eg only cosine terms vs only sine terms
 
My biggest frustration with this class is not that I can't do the work, but that I can't do the work at with the level of quality that I expect of myself in the time that is given to me.
:'(
 
This honestly sounds like a class a Physics grad student would be in a better position to teach than a typical math grad student
 
3:17 AM
Heh. Probably.
 
solving pdes via separation is something that you have to git gud at
 
I am guessing that they gave me the class because I am one of only three or four grad students in the program at the moment who has passed the PDE qual
 
At least to get through E&M and quantum
Ah
 
but I have literally zero physics background
like, I took a semester of high school physics in 1997
 
crazy
 
3:18 AM
and a quarter of quantum mechanics last spring (like, unbounded operators on Hilbert spaces, amirite?!)
 
Trotter product formula, brah! Do you even commute?
 
@BalarkaSen Today's talk was about harmonic maps into CAT(1) spaces
@BalarkaSen You would have enjoyed this notion of a metric space-valued Sobolev function on a manifold
 
Okay, I need to feed the cat then go the f*ck to sleep. Thanks for letting me vent. ;)
 
3:22 AM
@XanderHenderson lol
You use that in QM occasionally
Night
 
@XanderHenderson I was born that year
@Semiclassical what do you make of this recent NK thing
 
vzn
@Semiclassical re QM did you see these? didnt realize its 5 separate experiments over both Science/ Nature. =O
 
 
3 hours later…
6:51 AM
Does theory of scalar fields and vector field hold for vector valued functions of one variable?
 
7:38 AM
$\lim_{n\to \infty} {\left(\dfrac{\sqrt 3}{2}\right)^n}= 0$
why?
 
7:51 AM
What's the difference between a category and a directed graph satisfying some conditions ?
 
8:07 AM
How do i find stats on the total percentage of internet traffic in the public domain attributed to a particular site like say i wanted to compare the global activity on twitter facebook, and reddit
 
8:21 AM
@Abcd Remember that continuous functions preserve the limit so if the limit exists you can do
$\lim_{n\to \infty} {\left(\dfrac{\sqrt 3}{2}\right)^n}= \log \left( \lim_{n\to \infty} {\left(\dfrac{\sqrt 3}{2}\right)^n} \right) = \lim_{n\to \infty} \log \left( {\left(\dfrac{\sqrt 3}{2}\right)^n} \right) = \lim_{n\to \infty} n \log \left( \left(\dfrac{\sqrt 3}{2}\right) \right) = 0$
since the log expression is some number with magnitude smaller one
you see that the limit exists, therefore the steps are valid
With problems like these you kind of assume already that the limit already exists even if you don't know, so that you can try to compute it.
Wait I forgot to take the $\exp$ afterwards
Correct would be
$\lim_{n\to \infty} {\left(\dfrac{\sqrt 3}{2}\right)^n}= \exp \left( \log \left( \lim_{n\to \infty} {\left(\dfrac{\sqrt 3}{2}\right)^n} \right) \right) = ... = \exp \left( \lim_{n\to \infty} n \log \left( \left(\dfrac{\sqrt 3}{2}\right) \right) \right) = \exp \left( -\infty \right) = 0$
so forget what I wrote above
only the last line is correct
 
8:53 AM
Hey math folks
I have a conundrum
What should I use ideally for the local trivialization of a bundle
$\psi : \pi^{-1}(p) \to M \times F$ or $\psi : M \times F \to \pi^{-1}(p)$
I've seen both used
and since it's a homeomorphism it doesn't matter that much in the grand scheme of things
But which one would you say is the most standard
Maybe I should call the first the local trivialization and the second the bundle coordinates?
 
9:16 AM
@philmcole I didn't understand why you took logarithm there.
$0 \times \infty \ne 0$ :/ ?
@LeakyNun Please give it a look.
I thought its 0.
and I think $0\times \infty$ is not equal to $0$ — Frank Moses 8 mins ago
 
@Abcd Look only in the last line. I took $\exp(\log(..))$ which are inverse functions and therefore not changing the limit. But it helps to compute it, because of the log property that $\log(a^p)=p \log(a)$
 
@philmcole okay thanks
@philmcole how did you bring lim outside log btw?
 
that's what I meant with $\log$ and $\exp$ being continuous functions. This is a characteristic of continuous functions.
 
@philmcole Please brief me about that characteristic. I am unaware of it.
 
9:32 AM
In mathematics, a continuous function is a function for which sufficiently small changes in the input result in arbitrarily small changes in the output. Otherwise, a function is said to be a discontinuous function. A continuous function with a continuous inverse function is called a homeomorphism. Continuity of functions is one of the core concepts of topology, which is treated in full generality below. The introductory portion of this article focuses on the special case where the inputs and outputs of functions are real numbers. A stronger form of continuity is uniform continuity. In addition...
A function $f:X \to Y$ is continuous at $x_0 \in X$ $\iff$ for every convergent sequence $(x_n)_n$ in $X$ with $\lim_{n \to \infty} x_n=x_0$ also $\lim_{n \to \infty}f(x_n) = f(\lim_{n \to \infty} x_n) = f(x_0)$
The property we use above is the $\implies$ direction.
 
10:18 AM
@philmcole that's the right answer, but the wrong answer for this situation
 
why?
 
everyone is unique, you can't use the same answer that works for A for an answer that works for B
this answer may work for other people but not him
 
okay...
 
:P
different people like different answers
 
I'll leave it to you then :P
 
10:20 AM
@Abcd for $(\sqrt3/2)^n < \varepsilon$, one only needs $n > \log_{\sqrt3/2}\varepsilon$
this works because $\sqrt3/2 < 1$
so $(\sqrt3/2)^n$ is decreasing
I suddenly thought of another answer:
since $(\sqrt3/2)^n$ is decreasing and it is bounded below by zero, it has a limit (by monotone convergence theorem), so let that limit be $L$
then, $(\sqrt3/2)L = (\sqrt3/2) \displaystyle \lim_{n\to\infty} (\sqrt3/2)^n = \lim_{n\to\infty} (\sqrt3/2)^{n+1} = L$
so $L = 0$
 
11:00 AM
@LeakyNun Why is 0* infinity not defined?
 
11:30 AM
Could you please help me check about whether the note I'm reading is wrong? The comments below let me wondering there is something wrong?
0
Q: About the dimension of a Jacobian matrix in the note of CS231n

NiingI'm reading a class note of CS231n, and I'm wondering the dimension of the Jacobian matrices $\Large\frac{\partial Y}{\partial X}$. In the given example he provided a formula $$\frac{\partial L}{\partial X}=\frac{\partial L}{\partial Y}\frac{\partial Y}{\partial X},$$ use the formula my calcula...

 
12:15 PM
@Abcd Consider: $\lim_{x\to0}x\cdot\frac1x$, versus $\lim_{x\to0}x^2\cdot\frac1x$, versus $\lim_{x\to0}x\cdot\frac1{x^2}$
Each one is of the form $\lim_{x\to0}f(x)g(x)$, with $\lim_{x\to0}f(x)=0$ and $\lim_{x\to0}g(x)=\infty$, but the answers are different.
Consider also $\lim_{x\to0}x\cdot\frac2x$
 
What is the DTFT of x[n]*(cos(2*pi*v0*n))^2
oops, meant: x[n]*(cos(2*pi*v0*n))^-2
 
 
2 hours later…
2:19 PM
How to show that the $\Bbb Q$-vector-space
$\{f : \Bbb R → \Bbb R | f \text{ is continuous and Image}(f) ⊆ \Bbb Q\}$ has dimension $1$?
 
@Silent find its basis and show that the basis only consists of one element :P
hint: Q is totally disconnected
 
@Silent write down some examples of functions in this space
 
ok
 
@GFauxPas hi
 
Morning
 
2:22 PM
5
Q: Arc contribution in $\int_{-\infty}^\infty \mathrm{d}z \frac{e^{-z^2}}{z-1}$

Just AskConsider an improper integral with a pole on the integration contour at say $z=1$, $$ \tag{1} I = \int_{-\infty}^\infty \mathrm{d}z\ \frac{e^{-z^2}}{z-1+i\epsilon},~~~~~\epsilon>0. $$ Let $$f(z) = \frac{e^{-z^2}}{z-1+i\epsilon}$$ then $$ \sum_{residues~inside~\Gamma} = 0 = \oint_\Gamma f(z) ...

regarding Waiting's challenge, I actually have truoble visualising what the contours:
$\Gamma_{\epsilon}$ and $\Gamma_{\infty}$ look like
I mean, the white line here is I, but where are the other two contours are?
 
@AlessandroCodenotti Only examples come to mind are like $f(x)=\frac pq$ for all $x\in \Bbb R$
 
@Silent Those are the only examples that come to my mind as well and there's a good reason for that ;)
 
Hey dumb question, Bezout's theorem doesn't work in the classical form over $\Bbb Z[x]$, right?
 
Hi @Balarka and @Dami
the whole meme crew is here
@BalarkaSen Which Bezout's theorem?
 
@AlessandroCodenotti no other examples exist! Thank u
 
2:29 PM
@Silent You should prove it though
 
Bezout's lemma I meant. I guess the point is $\Bbb Z[x]$ is not a PID
So eg $x$ and $2$ are counterexamples
 
Yeah, the keyword to google is Bezout domain
 
(They have gcd 1 but you can't produce polynomial $f(x)$ and $g(x)$ such that $xf(x) + 2g(x) = 1$, for obvious reasons).
is googling
 
@BalarkaSen yo dawg I heard you like indices i.gyazo.com/8c6511c23634bdb7376bb08b21ce882b.png
 
Are there like fourteen implied sums inside the integral?
 
2:32 PM
@0celo7 This is sickening
 
is A a tensoid thingy?
 
@AlessandroCodenotti implied sum over $i,j,\mu,\nu$, explicit sum over $\alpha$
 
god, so many things need to catch up. I really have to stp doing politics starting from 30 April\
 
@MatheinBoulomenos do you have a good reference with a complete proof of the Minkowski bound? We're only doing the quadratic number fields case in class
 
@BalarkaSen proof of the Hodge theorem via indices
 
2:40 PM
does anyone have some reference to filters in topology?
@BalarkaSen are you familiar with limits done via filters?
 
@Alessandro Introductory Algebraic Number Theory by Alaca and Williams has a chapter on it
 
@LeakyNun I have one but I need my computer to find it again and I don't remember whether it's in English or Italian
@ÍgjøgnumMeg thanks, I'll check it!
 
@AlessandroCodenotti penso che posso leggere italiano..
 
Hi why does the Banach fixed point theorem imply that when you throw a map on the ground some point on the map lands exactly above the one it’s representing?
What is the function in this case?
 
@philmcole earth to map
 
2:45 PM
I'll see if I can find it later, I have to catch a train now @Leaky
 
ok
 
Leaky can you expand a bit? :P
 
@philmcole sending each point on earth to the point representing it on the map
 
And why is it true that one point lands exactly above itself (except that the theorem guarantees it)? I mean I can’t imagine it being true yet
Also in the theorem the map goes from the metric space $X$ to $X$. So we regard the points on the map as miniaturised earth?
And that’s why the map is a contraction since on the map the distance between points is smaller?
 
@Alessandro yo
 
2:54 PM
I think I'm being silly, but if $2u \notin \mathfrak{p}$ (where $\mathfrak{p}$ is a prime ideal of some ring) then this doesn't guarantee that $4u \notin \mathfrak{p}$, right?
 
@Daminark are you familiar with filters?
@ÍgjøgnumMeg if 4u in p, then 2 in p or 2u in p.
if 2 in p, then 2u in p.
if 2u in p, then 2u in p.
so, 4u in p implies 2u in p
 
@LeakyNun yeah I just got this after typing it
hahaha
thanks
 
lol
 
Not really
 
ok
 

« first day (2823 days earlier)      last day (2184 days later) »