« first day (3142 days earlier)      last day (2175 days later) » 

00:22
Can $\Bbb{Z}$ act nontrivially on any of its subgroups?
Obviously conjugation doesn't work, nor does left (or right) translation works, since translation might land outside of the subgroup....
00:34
@user193319 try translating by more than 1
I'm not sure I follow
01:13
Which statement is correct; "velocity is the derivative of position with respect to time" or "velocity is the derivative of distance with respect to time"?
how about neither, and velocity is the derivative of displacement with respect to time
I've seen all three of those definitions used ubiquitously.
definitely not distance, the units don't match up
@user10478 you'll see wrong things on the internet everywhere
Okie
But velocity does describe the way position changes over time, not the way displacement changes over time.
Displacement already has change built into it, right?
displacement is just position relative to a fixed origin
so ok the first one is correct
01:19
yo
limit question
If you move from point a to point b, and neither is the origin, then displacement is different than position, right?
$\delta x$
is displacement
it's the distance moved
@user10478 but the change in displacement is the same as change in position
position is a single point
01:22
(b-o)-(a-o) = b-a
I found a limit that wolfram alpha can't calculate
I was like hmm whats this limit
and wolfram alpha was like hmm idk
@Ultradark can you look at my stackoverflow question?
Okay, so to summarize, we can say "velocity is how position changes over time," but not "velocity is the derivative of position with respect to time"?
01:25
sure @Spencer1O1
@user10478 the latter is correct
@Ultradark Thanks!
I messed up at the beginning
Ah, okay, so for a state function, velocity is the derivative of position, distance, or displacement with respect to time are all correct?
In that case, what about your argument that the units for position don't match?
@Spencer1O1: Hint: itertools.combinations on a list with r=3...
01:28
@user10478 not distance
what is a state function?
it's a function of of state variables
used in physics
A function where the start and end points are all that needs to be considered, I think. For example, if you walk around town randomly, net distance is end point - start point. Total distance (the number on your pedometer) is not a state function.
like entropy is a state variable
path independence
internal energy is a state variable
I have a random question. How do you take the derivative of the prime counting functions
I would infer that for a state function, distance = displacement (at least up to sign).
I realize you can't because it's not continuous
but is there another way?
I guess I have to think about what is the rate of growth of $\pi(x)$
which I guess is asymptotically the rate of $x/\ln(x)$
so the derivative of $x/\ln(x)$ is $ \frac{\ln(x)-1}{\ln^2(x)} $
How do you deal with infinity over zero form in a limit
I can't apply lhopitals rule
01:45
@Dair so would I just do itertools.combinations("abcdefghi",3)?
@Dair it would just return abc,abd,abe,abf...
It just lists the combinations of letters...
Isn't infinity over zero just infinity trivially?
I think L'Hospital is only necessary because when both the numerator and denominator approach zero (or both infinity), it's not obvious which is stronger.
It's like a tug of war, but infinity over zero is tug of war with one team pulling and the other pushing.
02:16
Because as $x$ approaches infinity, $x/n$ also approaches infinity, but anything divided by zero is undefined.
But as $n$ approaches $0$, $x/n$ also approaches infinity :P
yes
It is in indeterminate form
the limit does not exist
I figured it out
03:20
Does anybody know of anything interesting going on at the interface between functional analysis and differential geometry? Seem to be some analogies between the two
 
2 hours later…
05:33
(need feedback on this meme)
 
3 hours later…
08:31
Hey guys!
This is pretty simple but any idea how I can show that the iterates of points with non-zero imaginary part map to infinity under $f(z)=2z^2-1$? I can do it using some complex analysis as was suggested in one textbook, but it seems like an overkill. I can't get a elementary estimate right now though.
09:01
I think I got it.
@JoeShmo have you tried looking into noncommutative geometry? I'm not sure how much diffgeo there is though
@AlessandroCodenotti My impression is that there is not that much diffgeo in that
hmm, though there might be some in the BV formalism (seeing as that is some mathematical physics, which tends to include some diffgeo), and that has been approached using noncommutative geometry (in the form of spectral triples)
 
1 hour later…
10:29
What's the intuition behind regular rings? Let $A$ be local Noetherian, $\mathfrak m$ its maximal ideal and $k=A/\mathfrak m$, why should I expect $\dim A$ and $\dim_k\mathfrak m/\mathfrak m^2$ to agree in nice situations?
Also how do I think about regularity geometrically in terms of (co)tangent spaces/sheaves?
10:45
@AlessandroCodenotti m/m^2 is the cotangent space of Spec(A)
and dim(A) = dim(Spec(A))
one would want the dimension of the cotangent space to equal the (Krull) dimension of the space I guess
But we're looking at the dimension of $A_{\mathfrak p}$, not of $A$, since we want the stalks to be regular rings and this can be be different from the dimension of $A$, right?
the stalk is an “infinitesimal” neighbourhood of A... so my gut tells me that their dimensions should be the same?
Provided that p is a maximal ideal, I guess
Right, that works for $\mathfrak p$ maximal
But for $\mathfrak p$ minimal $\dim A_\mathfrak p =0$ and for general $\mathfrak p$ it can be nonzero and lower than $\dim A$
sure
but you asked about nice situations :p
Well I think that getting the intuition from the maximal ideals and generalizing is quite convincing
Here's something else I find confusing: Suppose I have $X$ a scheme over $Y$, I can construct the cotangent sheaf $\Omega_{X/Y}$ essentially by glueing together modules of Kähler differentials
But intuitively I'd like the cotangent space to actually be a scheme rather than a sheaf, can I just take the relative Spec of the cotangent sheaf? Is $\Omega_{X/Y}$ even an $\mathcal O_X$-algebra?
11:51
@JoeShmo This interface is essentially the fields of operator algebras and noncommutative geometry. See the works of Alain Connes.
12:55
How would I do the integral $\int \int _S x_i x_k ds$ where S is the surface of a sphere of radius 1 and x denotes the position vector
Oh nevermind I see it now
I’m actually stuck
13:37
What's $x_i$ and $x_k$
(Also you can do \iint for $\iint$)
 
1 hour later…
15:06
@AlessandroCodenotti The key is to take the symmetric algebra of your coherent sheaf -- as in the case when you're looking at the correspondence between locally free sheaves and vector bundles!
And take Spec of that?
That makes sense actually, as you said it's the same as the correspondence with vector bundles!
I forget if you want to take the dual (as in the vector bundle case) -- but probably doing some examples will make it clear
15:22
Even in the vector bundle case there is some debate on taking the dual or not or so I've heard
15:38
So for example if I look at $X=\mathrm{Spec}\Bbb C[x,y]/(x^2+y^2-1)$ (as a scheme over $k=\Bbb C$) I get that $\Omega_{X/k}$ is the module generated by $dx$ and $dy$ with relation $2xdx=-2ydy$
So if $y\neq 0$ we have that $\tilde{\Omega}_{X/k}$ is generated by $dx$ and similarly for $x\neq 0$ and $dy$, so $\tilde{\Omega}_{X/k}$ is an invertible sheaf (or a line bundle, which agrees with intuition so far)
So now I want to obtain a scheme back from $\tilde{\Omega}_{X/k}$ and that scheme should look like a cylinder
Which is where I start to get lost
 
2 hours later…
17:51
@JakeRose You can do it in spherical coordinates (reduce by symmetry to $i=k=3$ for easiest computation). You should see that by symmetry you get $0$ when $i\ne k$.
hi demonic @Alessandro
I think demonic is appropriate since I'm doing algebraic geometry, it is unholy and sinful indeed
@AlessandroCodenotti In this case you should expect to see that $\Omega_{X/k}$ to be trivial - since $X$ is isomorphic to $\mathbb{C}[u,v]/(uv-1) = \mathbb{C}[u,u^{-1}]$ which is a PID!
Now now now ...
Sloppy language, loch, saying a variety is isomorphic to a ring.
17:57
:p
@loch wait what
I'm not seeing the isomorphism
$\mathbb{C}[u,v] \rightarrow \mathbb{C}[x,y]/(x^2+y^2-1)$ sending $u\mapsto x+iy, v\mapsto x-iy$
Ah, this famous conic ...
Aha makes sense
I still don't see why $\Omega_{X/k}$ being trivial follows though
Can you give a global nowhere-zero section?
18:04
I guess I cannot :P
But looking at the explicit description of the module of Kähler differentials I don't see why it is trivial
howdy supanoid Eric
You can apply the same argument to $\mathbb{C}[u,v]/(uv-1)$, so the relation you get is $vdu+udv$, multiply this by $v$ gives you $dv = \ldots$, which kills $dv$

But probably better is to realize that $\mathbb{C}[u,u^{-1}]$ is what you get when you invert $u$ in $\mathbb{C}[u]$, and use something about Kahler differentials and base change (which in this case is just a long way of saying you have a trivial bundle over $\mathbb{A}^1$ and you're taking an open set)
18:09
what's the difference between an associate and an assistant prof
associate is higher ... almost always after tenure.
@loch Nice, I like the open set picture, thanks
I'm confused now though, how should I interpret $\Omega_{X/k}$ being trivial, geometrically? This doesn't agree with my intuition for what the tangent bundle should look like
Well, cotangent bundle, actually :P
The curve is isomorphic to the affine line, which has trivial (co)tangent bundle.
Oh, of course
I guess I should think about $\Bbb P^1$ for a nontrivial example then
Yuppers.
And you should understand how the overlap map $z=1/w$, $dz = -1/w^2\,dw$, tells you the whole story.
18:18
I did this once already actually, the cotangent sheaf of $\Bbb P^1_{\Bbb C}$ is one of Serre's twisting sheaves, namely $\mathcal O(-2)$
Yup, and what I typed shows that.
You get a pole of order $2$ for the obvious differential ($1$-form).
every line bundle on $\mathbb{P}^n$ is of the form $\mathcal{O}(n)$ :p
Yeah I've been told $\mathrm{Pic}(\Bbb P^n)=\Bbb Z$ but it wasn't proved
How do I find the order of the poles of $\frac{z^2}{(1+z^2)^2}$?
Factor the denominator, @Jake.
(I assume you meant $^2$.)
18:22
@TedShifrin I’ve done that
oooh yes thanks for pointing that put
So let's say $\Bbb P^1$ has coordinates $x_0$ and $x_1$, I look at the two usual affine patches $x_0\neq 0$ with coordinate $z=x_1/x_0$ and the other patch defined similarly with coordinate $w$. I look at $dz$ on the first patch, this is the same as $d(1/w)$ on the overlap and what you wrote follows @Ted
You got my answer to the surface integral question earlier?
Yup, @Alessandro.
yss I did thank you very much for that
Okey dokey.
with this, when you factor it, I don’t understand how that tells me the order of the poles?
18:23
What order pole does $1/z^2$ have at $0$?
So, what order pole does $1/(z-i)^2(z+i)^2$ have at $z=i$ or at $z=-i$?
So we're done.
How do you show that that is equivalent to looking at its Laurent expansion?
oh actually I see it now
thank you ted a salwsgs
18:25
Let $u=z-i$ and write $z^2/(z+i)^2$ in terms of $u$.
as always*
These are good computations to know how to do in your sleep.
@TedShifrin do assistants take students
Risky if they're pre-tenure, @Eric. But possible ...
Generally, it's preferable to have an adviser who's had a few students and knows how to do it ... :)
And whose name carries weight later on ...
18:28
@TedShifrin is the best way of getting the residue simply by using the formula with the derivatives?
an assistant prof at ucb whose work i thought was cool emailed me out of the blue like whaddup
I never do that, @Jake, nor do I teach it.
so i was confused cuz i guess i thought they didnt take students
If he's a star who's expected to be promoted and stay, not so much a risk, @Eric.
Oh, they certainly can. You might ask how long he expects to be at UCB :)
18:29
@Jake: I prefer to have people see it from the Laurent expansion. Of course, when there's a simple pole, it's clear that the residue of $f(z)/(z-a)$ at $a$ is $f(a)$ [assume $f(a)\ne 0$ and $f$ holomorphic, of course]
@TedShifrin i looked it up and i c now he has multiple students lol
I had a feeling you were gonna say that
Aha ... Still a good question whether he knows he's leaving in a few years or expects to be promoted. One of course never is sure.
seemed like quite a mindless formula to use
I don't like mindless formulae, Jake. :P
18:31
if it’s not simple how would you tackle it?
@TedShifrin we have something in common
@TedShifrin for sure
Laurent expansion, of course, like the one we were just talking about. Put in $u$ and compute for a minute.
@Eric: Even though the PhD program at UCB is smaller than when I was there (approx 400), it is still easy to disappear there. I doubt you'd have a problem, but be aware.
they were kind of off my radar but then they offered me a big money no teaching thing that forced it back into perspective
Oh, that's unusual.
Of course, I don't know how things have changed there over 40 years ... :)
i think it was "ur good and brown u make us look good" situation
18:34
You want some teaching, of course, but not excessive amounts.
I know people routinely making ritual sacrifices in the woods to receive such an offer
yeah of course
Mhmm ted when I do it I get 3 but it’s i/4?
wtf am i doing
You're doing the residue at $z=i$?
Yes
i binomiallynexpanded thenbottom fraction, multiplied out the numerator and then found the x^-3 terms
18:43
Huh?
I mean x terms not -3
OK.
$x=z+i$?
I use just the geometric series every time.
So you have $-1/4$ times $(-1+2ix+x^2)(1+ix + \dots)$, so the $x$ coefficient is ....
Specifically that’s what I did
oh I didn’t square it
I see now thanks ted
18:45
Yup, I was about to point that out.
 
2 hours later…
20:45
♫ all by myself ♫
hello folks
i want to "truncate" a taylor series using the MVTI on the double integral :
f(x+h) = f(x) + f'(x)\cdot h + \int^{x+h}_x \left(\int_x^{s} f''(p) dp\right) ds
oups
you need dollar signs around your equation
$f(x+h) = f(x) + f'(x)\cdot h + \int^{x+h}_x \left(\int_x^{s} f''(p) dp\right) ds$
can't you use Taylor's theorem or something
no i dont want a residue, i want to end up with the exact value with MVTI
can the normal MVTI be applied to the double integral ?
20:55
Here's what I think is an easy bounty for whoever knows the concept I describe . . .
1
Q: A positive integer "modulo a sequence".

ShaunMotivation: The (principal) value of $$m\pmod{n}$$ for some positive integers $m> n$, might well be viewed as the value $$m-\sum_{i=1}^{M_{m,n}}n,\tag{$\Sigma$}$$ for some $M_{m,n}\in \Bbb N$ with $M_{m,n}n\le m$ but $(M_{m,n}+1)n> m$; indeed, we have just subtracted a suitable number $M_{m,...

@MehdiSlimani what is the normal MVTI?
$\int^a_b f(x)dx = f(c)\cdot (b-a) \qquad c\in ]a, b[$
mean value theorem for integrals, sorry. didnt realize i was the only one using that abbreviation
looks good to me
interesting
oups my bad one sec
$f(x+h) = f(x) + f'(x)\cdot h + f''(\alpha)\cdot\frac{h^2}{2} \qquad \alpha\in ]x, x+h[$
*i know you can end up with this but i dont know how
isn't that just Taylor theorem or something
21:15
taylor theorem talks about the residue of a truncated taylor sum
here, i start derivating the taylor sum, but at the second integral instead of continuing i want to say something along the lines of "there is an average value of the integrand"
i think i found what i was looking for. formulating my problem helped me a bit, thanks !
isn't that the Taylor theorem with the Lagrange form of the remainder?
yes ! i didnt know about this, thank you
21:52
$\lim_{x\to \infty} \pi(x)^{\frac{1}{\ln(x)}}=e$ Wow!
that means as $x$ tends to infinity $\pi(x)$ raised to the rate of the logarithmic integral is equal to a constant
22:05
pretty useless result but cool nonetheless
 
1 hour later…
23:21
I think the point is that you want the dual there if you want to recover your locally free sheaf from your bundle $V$ as sections of the map $V\rightarrow X$.

Taking duals is bad if you want to apply the same construction to arbitrary coherent sheaves though, since taking dual tends to kill information.
As an example if you take the skyscraper sheaf at the origin for $\mathbb{A}^1$, if you do the "dual" construction you end up with nothing because the dual of the skyscraper sheaf is 0 (on the level of $k[t]-$modules, there are no maps $k\rightarrow k[t]$)

Whereas if you do the non-dual construction, you should get a space that looks like $k[t,x]/(tx)$, so you genuinely see a space that looks like what you'd expect from the skyscraper sheaf

(and I think you can recover the skyscraper sheaf by taking "cosections")
Fix a commutative ring with unity $R$ with module $M$ and consider the category $\Lambda_M$ whose objects are $(m, \alpha)$ with $\alpha:R^m \to M$ and morphisms are commuting triangles
Is this category filtered? @loch
brb looking up defn of filtered category
the corresponding directed graph is directed (trivial) and every pair of parallel morphism coequalizes somewhere above (I can't prove this)
i.e. I have $f, g : R^m \to R^n$ with $\alpha:R^m \to M$ and $\beta: R^n \to M$ such that $\beta g = \beta f = \alpha$
I want to find $k$ with $h : R^n \to R^k$ and $\gamma : R^k \to M$ such that $hf=hg$
hmm this requires too much thinking
is it obvious if you take nice R / M ?
not to me...
@loch the problem is, the text I'm reading uses $\varinjlim$ for both filtered direct limit and colimit
and it doesn't specify in this case so I'm left assuming that it's a filtered direct limit but my argument can't go through
which leads me to suspect that this isn't filtered at all
23:40
lol what text are you reading

« first day (3142 days earlier)      last day (2175 days later) »