« first day (2378 days earlier)      last day (2637 days later) » 

1:00 PM
@Ramanujan Rewrite the system of equations as : $$\begin{cases}3a + 10b = -5c\\4a+6b = -2c\end{cases}$$
 
why not 9!!!!!!!!!!!!....!!! ?
how will you decide this is bigger or smaller than some other number
 
@s.harp if it gets hard, we'll ask M.SE
And that number is relatively small. Compared to most I've seen
 
Use Cramer's rule : $$a = {-5c\times 6 - (-2c)\times10\over 3\times 6-4\times 4}$$
and
$$b = {3 \times (-2c) - 4\times(-5c)\over 3\times 6-4\times 4}$$
Compute the quotient : $${a\over b} = {-30 + 20\over -6+20}$$
This rewrites as ${a\over 30-20} = {b\over 20-6}$
Do the same with $$\begin{cases}3a + 5c = -10b\\ 4a + 2c = -6b\end{cases}$$ to get the other equality
Typo two messages above : $b\over 6-20$
 
1:27 PM
To judge how large the numbers in the contest, they are mostly larger than Graham's number.
If anyone wants to look up how large that is
And that's the example code made in Ruby.
 
@Astyx thank you very much
And sorry for my behavior
 
@TedShifrin The set of singular 2x2 matrices fails to be a manifold at 0. What does the set look like near there?
It's the cone of something…
 
To be precise, real matrices or complex matrices? @AkivaWeinberger
 
1:45 PM
Real
 
Mmkay.
It seems like the best way to view it is that a 2-by-2 matrix is singular when the second row is some multiple of the first, in which case the problem is parametrized by the choice of first row and said multiplier.
 
Ooh, that should work
So it's the set of all $(a,b,t)$, with the points of the form $(0,0,t)$ quotiented together?
 
Well, i think you'd also have $(a,b,t)\sim $(ac,bc,t)$.
Since rescaling the first row shouldn't change whether the matrix is singular.
 
No, one corresponds to $\begin{bmatrix}a&b\\at&bt\end{bmatrix}$ and the other corresponds to $\begin{bmatrix}ac&bc\\act&bct\end{bmatrix}$.
 
If the first one is singular, so is the second (assuming $c\neq 0$.)
 
1:50 PM
and are distinct singular matrices
 
Hmm, fair.
 
Right, so they're both in the set of singular matrices. They're still different points of $\Bbb R^4$, though
 
Yeah.
Ultimately one is trying to describe $ad-bc=0$ in $\Bbb R^4$.
 
the quotient relation should be (a,b,0)~(c,d,0)
 
No, they have different first row @arctictern
 
1:52 PM
derp
 
It should be $(0,0,t_1)\sim(0,0,t_2)$
so 3D space with one axis quotiented to a point.
 
that
 
Hm. Does this mean that the set of nonzero singular matrices is homeomorphic to the interior of a solid torus?
Yeah. Cool.
The map to $S^1$ is given by $\begin{bmatrix}a&b\\c&d\end{bmatrix} \mapsto\frac{\langle a,b\rangle}{\|\langle a,b\rangle\|}$.
Wait, no
@arctictern @Semiclassical This is wrong
 
hello
 
The second row doesn't need to be a multiple of the first.
 
1:56 PM
o/
 
ah, right
 
We're forgetting about $\begin{bmatrix}0&0\\c&d\end{bmatrix}$.
 
The description that comes to mind for me is $(c_1,c_2,t)$, corresponding to a matrix $\begin{pmatrix} c_1 t & c_1 (1-t) \\ c_2 t & c_2(1-t)\end{pmatrix}$.
 
guys, any thoughts on my question
?
 
1:57 PM
That fixes it, yeah @Semiclassical
 
@Shobhit Write your transformation as matrix multiplication and compute the third power of the corresponding matrix.
To check that what I wrote gives a unique parametrization: If $(c_1 t,c_1(1-t))=(c_1' t',c_1'(1-t'))$ then $c_1' (1-t')=c_1(1-t)=c_1 - c_1 t = c_1- c_1' t'\implies c_1=c_1'\implies t=t'$.
 
@Semiclassical and that will be equivalent to T^3?
 
If you can write $T(a,b,c)=Ax$ for $x=(a,b,c)^T$, then certainly $(T^3)(a,b,c)=A^3 x$.
 
@Semiclassical ok. on it!
@Semiclassical thank u :)
 
So the only way for two such matrices to have the same first row is to have the same $c_1,t$.
 
2:03 PM
@Semiclassical I'm not sure I follow, but I see that your parametrization is determined uniquely unless $c_1=c_2=0$.
 
Yeah, that's what I was aiming for.
And if $c_1=c_2=0$ then $t$ is free.
 
So we get 3D Euclidean space modulo the t-axis again, and nonzero singular matrices is 3D space minus the t-axis.
 
So all $(0,0,t)$ are equivalent.
Right.
 
Which is an open solid torus.
@Semiclassical No, wait, I'm confused again.
 
2:06 PM
@Shobhit FYI: In the above, I wrote $x$ as a column vector. You could have it be a row vector i.e. (a,b,c) without ^T, but in that case the multiplication will act on the right (so as to send row vectors to row vectors).
 
How do we write $\begin{bmatrix}1&-1\\1&-1\end{bmatrix}$ in your parametrizarion?
 
Should have $c_1=c_2$, so I'll just say $c$ for both.
So therefore we need $c t=1$ and $c(1-t)=-1\implies ct+c(1-t)=c=0$...bollocks.
 
Yeah….
 
What are we discussing about?
 
2:08 PM
Yeah, you can almost any line through the origin in terms of its intersection with $x+y=1$.
The exception, alas, being the line $x+y=0$.
(something something point at infinity)
 
@Velvet What does the set of singular matrices look like, as a subset of $\Bbb R^4$? It's a three-dimensional manifold everywhere except for the zero matrix.
 
@Semiclassical i have calculated A such that T(a,b,c)=Ax. now to calculate rank of T^3, i just need to find rank of A^3. is this correct?
 
Right.
 
ok
 
…Guys, which one of you is Kenny Lau
 
2:09 PM
Maybe: $\begin{pmatrix} c_1 \cos \theta & c_1 \sin \theta \\ c_2 \cos \theta & c_2 \sin \theta\end{pmatrix}$
 
Typo on top-right @Semi
 
And if you don't like parametrizing with $\theta$, you can reparametrize using $t=\tan\frac{\theta}{2}$
Fixed.
 
OK. We're sure this covers everything, right?
'Cause all you're saying is that the rows must have the same angle,
which they must, for it to be singular.
 
Right.
 
And then $c_1=c_2=0$ makes $\theta$ free.
 
2:12 PM
We still get freedom in $\theta$ when $c_1=c_2$.
Isn't $\theta$ also free if, say, $c_1=c_2=1$?
 
So it's $\Bbb R^2\times S^1$ (an open torus) with the center circle, $\{0\}\times\{0\}\times S^1$, quotiented to a point.
 
Oh, but then it's different matrices.
Sounds right, yeah.
 
Wait —
 
@Semiclassical thank you. i was able to correctly solve it. But A^3 had very large entries, finding rank was hard. one can only imagine how would it be at T^10. so my question is, is there some relationship betweeen null spaces, rank spaces of T and T^n
 
Polar representation isn't unique. $(r,\theta)=(-r,\theta+\pi)$.
 
2:15 PM
@Shobhit I should remember, but I don't.
 
@Semiclassical so there is one?
 
…Interior of cone of a Klein bottle?
 
shrug @Shobhit
Don't trust someone who can't remember!
 
ok
 
One thing you can notice, though.
If a vector is in the nullspace of $T^n$, then it's also in the nullspace of $T^m$ for $m>n$.
 
2:17 PM
yes, good point
 
So it's $\Bbb R^2\times S^1$, quotiented by $(0,0,t)$, and quotiented by $(c_1,c_2,\theta)=(-c_1,-c_2,\theta+\pi)$.
Ignoring the first one:
Oh, I see. It's not a Klein bottle, it's still a Torus
(good thing it's an even number of dimensions)
Which takes us back to the open cone of a torus.
(Note that we can't simply require $\theta$ to be in $(-\pi,\pi]$.
 
If I do the Weierstrass sub I said earlier, then $(\cos \theta,\sin \theta)=\left(\frac{1-t^2}{1+t^2},\frac{2t}{1+t^2}\right)$.
 
If we do that, we'd forget about matrices where the top and bottom rows are parallel but opposite.)
@Semiclassical Don't you also need $t=\infty$ for that to work?
 
Yeah.
 
To get $(-1,0)$.
 
2:21 PM
And you still get hit with what you said, just with $(-c_1,-c_2,-t)\sim (c_1,c_2,t)$ instead.
 
one of these days, i'll talk with u guys on your level of mathematics (i hope)
 
But, yeah, final answer: open cone of a torus. The nonzero singular matrices then form a neighborhood of a torus.
@Shobhit Our level of mathematics, apparently, is "confidently stating the answer, before realizing that the answer is wrong, and then doing the same thing again multiple times"
4
But this time I'm sure we got it right.
 
@Shobhit What level are you on now?
 
It wouldn't be called research if we knew what we were doing!
4
 
haha, nice one
@Velvet my level is below u guys
i know it was not the best way to convey what i wanted to say, but i think u get what i wish to say
 
2:24 PM
@Shobhit You mean those guys, my level is below theirs as well.
 
i want to learn study discover more, more, more mathematics
 
Probably not a great sign for the rest of the day when it's 8:30am and I already feel like taking a nap.
(still need to finish grading this quiz uggghhh)
 
2:41 PM
What kinds of "chores" does one have in a mathematics department?
 
@SimplyBeautifulArt I mean when working at a university?
 
Lol, I'm still in high school, sorry.
 
2:54 PM
How can we prove (or disprove) that $|\kappa| \ge |\Bbb N| \implies |\kappa \times \kappa| = |\kappa|$?
 
That requires AC
 
and then?
 
I think it's actually equivalent to AC over ZF but I don't remember the details
 
would this be easier to prove? $|\kappa \times \kappa| = |\kappa| \implies |2^\kappa \times 2^\kappa| = |2^\kappa|$
 
27
A: About a paper of Zermelo

Asaf KaragilaThe following is a theorem of $ZF$:  The axiom of choice holds if and only if for every infinite set $A$, there exists a bijection of $A$ with $A\times A$. (i.e. $|A|=|A|^2$)  Let us overview the theorem of Zermelo, namely if the axiom of choice holds then $\kappa=\kappa^2$ for every infinite...

You prove that it holds for well orderable sets and then use AC by saying that every set is well orderable
 
3:04 PM
hmm... I only speak English
 
3:15 PM
Hi there.
Is $\displaystyle \int_2^\infty \frac{x+y}{(y)(y^2-1)(\ln(x+y))} dy$ uniformly convergent?
I proved that is convergent, and I bounded it with $\dfrac{1}{2}$
 
@Akiva Your starred message is so accurate
@Ramanujan Your behavior is fine
@Topologicalife What do you mean by an integral being uniformly convergent ?
 
What set is $x$ in ?
 
in $\mathbb{N}$
 
3:37 PM
I don't think it's uniformly convergent
Because if you fix the lower bound, and make x go to infinity, the integrand goes to infinity
 
4:00 PM
@Akiva: DogAteMy — with a little linear algebra you can prove that the set of singular matrices $(a,b,c,d)$ on the unit sphere in $\Bbb R^4$ is diffeomorphic to a torus, indeed (you can write down the map to $S^1\times S^1$). So, yes, it's a literal cone over this torus.
 
hullo @Ted
 
Hi @Ted, @Alessandro
 
Hi @Balarka
 
Hi to you too
 
4:06 PM
Hi @Balarka
 
Hi @people saying hi
 
4:23 PM
Hi
 
Can anyone help me out with a chain rule question? math.stackexchange.com/questions/2132003/…
 
4 hours ago, by Simply Beautiful Art
If anyone is interested, I'm hosting a big number contest. Code your number in any language (no experience needed), maximum of 256 characters, not including spaces, and try to reach the largest number you can. First submissions due Saturday. :-) http://chat.stackexchange.com/rooms/51337/this-is-the-realm-of-simply-beautiful-‌​art
 
@SimplyBeautifulArt how are you gonna compare them?
 
@DHMO they will either be obvious or I'll ask MSE
@DHMO (join the contest...)
 
Hanks got his ascending chain of anes
At the end there is a pro-p ane
 
4:42 PM
short for profinte p-adic ane
 
4:56 PM
@s.harp I don't think it is p-adic but rather p-ane and suffering
 
5:25 PM
Let $g:\mathbb R^3\to \mathbb R$ be smooth and $T$ be a linear isomorphism of $\mathbb R^3$. I'm trying to determine conditions on $g$ such that the fiber $(g\circ T)^{-1}(0)$ is a graph of a smooth function about a point $p$.

The obvious way to do this is to differentiate $g\circ T$ w.r.t $z$ using the chain rule, which gives one answer.

I think there's another possibility, of looking at the right entry of $Dg\cdot A$, where $A$ is the matrix representing $T$ in the standard basis. Thing is, this method gives me a different answer. Can anyone help me out?
Is the differential of a linear map maybe not equal to the matrix representing it w.r.t standard bases?
 
@Calc Differential of a linear map is equal to the matrix representing it w.r.t standard basis. In either case you should get that $Dg$ is nonsingular as a sufficient condition.
The point is to invoke the implicit function theorem.
 
5:44 PM
@BalarkaSen not the transpose? Could you take a look at my question? I explained there how I get different results
0
Q: Confused with chain rule while using implicit function theorem

CalcLet $g:\mathbb R^3\to \mathbb R$, $g=g(u,v,w)$ be a function and $(x_0,y_0,z_0)$ a point satisfying $$g(x-y,y-z,z-x)=0.$$ I need to give conditions on $g$ to enable expressing $z$ as a function of $x,y$. Using different methods I get different results. Differentiate the above equation w.r.t $z...

If I replace the matrix with its transpose then both methods give the same result
 
@Balarka: You're misusing the word nonsingular.
 
@TedShifrin ? I just mean $Dg$ is not zero.
Full rank in the image
@Calc You shouldn't get the transpose; derivative of a linear map is exactly itself. I am not going to read the question, but probably someone else (@Ted?) would help
 
@BalarkaSen thank you. @TedShifrin any chance you can help me out? The question is very short
 
Hey, does anyone know a book/website that shows how to graph the range of a parametric/vector function ?
I have the steward book on multi variable calculus but It doesnt explain much
 
@Ted I guess nonsingularity isn't the right word for non-square matrices, so I get your objection.
Anyway, replace that with "nonzero".
 
5:54 PM
@BalarkaSen could you help me if I write down the details here?
 
I really should be working on other things, so I can't help you right now; sorry about that
 
no problem
 
@Calc However, haven't you written down the wrong matrix in there for $T$? That eats (x, y, z) and spits (x - z, ...)
Whereas you should be spitting (x - y, ...)
 
Then I think I have a linear algebra problem.. I just wrote the matrix whose columns are the images of the standard basis vectors along $T$
i.e $x\mapsto x-y$ so the first column is $1,-1,0$
Is that incorrect?
 
That is not correct. What is [1, 0, -1; -1, 1, 0; 0, -1, 1]*[x, y, z]?
 
6:01 PM
It's $[x-z;-x+y;-y+z]$
 
So that sends $x$ to $x - z$, not $x - y$.
 
I see... What is the matrix of a linear map if not the one whose columns are the images of the standard basis vectors?
 
You should have gotten [x - y, y - z, z - x], right? But you didn't. So the matrix is wrong.
 
I understand. I was just sure I remembered from linear algebra how the matrix of a linear map is defined..
 
Hey everyone!
 
6:04 PM
@BalarkaSen I think that's also what's written here mathinsight.org/matrices_linear_transformations
 
@Calc That is true. But image of [1, 0, 0] by the linear map is [1, 0, -1]. So the first column should be that, not the row!
 
Ahh!!! I finally see what I did wrong
I looked at things as if $x$ itself is the first basis vector, not just a number
Thanks for explaining everything
 
No problem
 
6:24 PM
Hi @Daminark
 
@TedShifrin can I bother you for a minute ??
 
What's up?
 
Is the weak operator topology of a Hilbert space sequential? With this I mean is the closure of a set the set of all (sequential) limit points of this set?
 
I'm asked to graph the range of vectorial functions
But my book doesnt say a lot about it
 
@Maks: Like drawing parametric curves or surfaces?
 
6:31 PM
And I found (on the internet) that you can discover how the range graph of a parametric function looks like by plugin a couple values
And joining them
Isnt there a better way to do it ?
 
Well, that's not too helpful. How to join depends on what sort of function you have.
 
Because maybe for some functions I need to get like 20 or 30 numbers to get an accurate graph
 
Ugh.
What kinds of functions are you trying to graph?
 
$r(t) = (t,-t,2t)$
$r(t) = (sen(t), 3, cos(t))$
$r(t) = (t^2,t,2)$
And a couple more
 
So you should "know" that the first is a line, the second is a circle, and the third is a parabola.
 
6:34 PM
Ok, that helps me
 
Note that in the second two, one component is constant, so you're in some plane parallel to a coordinate plane.
 
Now, how do you know that ?
What procedure do you use to discover it ?
 
For the first, you have $t(1,-1,2)$, so all scalar multiples of a fixed vector. That's a line through the origin.
The answer to your question is experience and recognition.
You know how to parametrize a standard circle in the $xy$-plane. How?
 
So we ended up going over the midterm in class
 
@TedShifrin Its just a line parallel to (1,-1,2) ?
 
6:36 PM
Through the origin, @Maks.
Hi @Daminark.
 
@TedShifrin Well but there has to be a technique to discover them
Like some kind of procedure with steps
Or not ?
 
How's it going?
 
No, recognition of familiar things, @Maks. You should recognize $y=x^2$ as $(t,t^2,0)$.
You should know how to parametrize a circle of radius $a$ centered at the origin.
 
... the thing is I dont know
:(
 
@TedShifrin Yup, that's what Semi and I proved (after a few false starts).
 
6:39 PM
Well, figure it out.
 
The mapping is nice.
 
DogAteMy: Best thing to do is make a linear change of coordinates on $\Bbb R^4$ so that it is $S^1$ in one $\Bbb R^2$ cross with $S^1$ in another. Did you do that?
 
$\begin{bmatrix}\cos a/2\cos b/2&\cos a/2\sin b/2\\ \sin a/2\cos b/2&\sin a/2\sin b/2\end{bmatrix}$, if I write it out explicitly, I think.
 
Oh, way too hard.
 
For $a,b\in[0,2\pi)$.
@TedShifrin I mean, the $/2$s aren't strictly speaking necessary.
 
6:42 PM
That's not in $\Bbb R^4$, so I'm not sure what that means.
 
It's in $S^1\times S^1$
I mean,
 
No?
 
I'm mapping $S^1\times S^1\mapsto\Bbb R^{2\times 2}$
 
I'm suggesting you look at $ad-bc$ as the dot product of $(a,b)$ and $(d,-c)$.
Oh, I see what you're doing. I'm being more extrinsic, working in $\Bbb R^4$.
 
@TedShifrin But this is the first time I'm writing it out explicitly.
 
6:43 PM
:)
 
I'm thinking of the topology/geometry in $\Bbb R^4$, DogAteMy. The so-called "link" of the cone point.
 
Is the "link" what $X$ is to $CX$?
 
In general, it's the intersection with a small sphere centered at the singular point.
It needn't be literally a cone.
 
You can generalize my approach nicely, though. Consider $x_1x_6-x_2x_5+x_3x_4=0$ intersected with $S^5$. Figure out what surface that is. [This is very important in geometry/topology ...]
(This may actually be an exercise later in my book. I've forgotten.)
 
6:46 PM
OK. So we have $(a,b)\perp(d,-c)$, and $\|a,b\|^2+\|c,d\|^2=1$?
 
So you have $x,y\in\Bbb R^2$ with $\|x\|^2+\|y\|^2 = 1$ and $x\cdot y=0$. Yes.
Prove that that is linearly diffeomorphic to $S^1\times S^1$.
 
@TedShifrin do you mean here that it is homotopy equivalent to a certain kind of neighbourhood around the singularity? Or homeomorphic?
 
That dot product is handy, since it means $\Vert x+y \Vert^2 = \Vert x-y \Vert^2=1$ as well.
 
Sssshhhh @Semiclassic
 
Pft, all I'm noticing is the algebra.
 
6:49 PM
Remind me never to tell a joke with you in the room, @Semiclassic.
 
So I can map it to $(\pm\|x\|,\pm\|y\|)\in S^1$, where the signs are the same or different depending on the orientation of $x$ and $y$ in the plane
 
Harrumph.
 
Well, that's not unique.
 
Reminder: Don't tell a joke with @Semi in the room
 
@s.harp: Generally, homotopy type.
 
6:49 PM
But it uniquely maps to a point in $S^1/(x\sim-x)$, which luckily is another $S^1$.
 
What's not unique, DogAteMy?
 
Double harrumph!
 
Um, no, that isn't another $S^2$.
 
@TedShifrin Are we not working with simplicial complexes? Then it should be homeomorphism type.
 
Oh so funny thing that happened today in class
 
6:50 PM
I mean $S^1$ everywhere! Sorry!
 
And real algebraic varieties are triangulable.
 
Homeomorphism type of whom?
 
The link is well-defined up to PL homeomorphism
 
Oh, no, DogAteMy. I want a linear map. Yours is yuck.
 
Our professor wanted to use quotient spaces in order to prove that matrices over $\mathbb{C}$ were upper triangular (we only did the $3\times 3$ case on the test, which allowed you to dodge that kind of argument)
 
6:51 PM
:(
 
@MikeM: I thought s.harp was asking if the actual variety was homeomorphic to the link. I dunno what's going on.
 
something nice a friend asked me about the other day: show that the configuration space of two unordered points on the circle is the closed Mobius band, without cut-and-paste arguments
 
Now, it was in Hoffman and Kunze which we did last quarter
Using conductors and annihilators
 
@Daminark: You certainly can do an elementary induction argument. That's an exercise I always assign.
 
And Schlag was ranting on about it, it was hilarious
 
6:51 PM
(In my free time I'm watching these lectures on stratified spaces youtube.com/watch?v=ga7W5TXN4kU , which is why I was interested in your definition of link )
 
Huh? You don't need that.
 
@TedShifrin I thought he was asking if the link was well-defined up to homeomorphism.
 
@MikeM: I'm lost ... too many discussions.
 
@s.harp Stratified spaces are more general and you get less control over the link of a stratum.
 
"Don't give me the book Souganidis told you to learn from last quarter and start talking about conductors, I want linear algebra, not trains!"
 
6:52 PM
In a simplicial complex, the link of a simplex is well-defined up to PL homeomorphism.
 
Right, for small enough spheres, definitely up to homeomorphism/diffeomorphism.
 
Oh. $(x-y,x+y)$, then @TedShifrin
 
Uh huh, DogAteMy.
 
Hello could someone help me find the third degree taylor polynomial of xe^y-y+1=0 near (0,1)? (asking for a friend, he can't write here)
 
We were all cracking up about it for quite a while
 
6:53 PM
Which means $\begin{bmatrix}a&b\\c&d\end{bmatrix}\mapsto \big\langle(a-d,b+c),(a+d,b-c)\big\rangle$?
$\in S^1\times S^1$
 
@Daminark: I'm inclined to agree. I can do a totally elementary argument if I only know what an eigenvector is.
 
And then a neighborhood of the simplex is literally homeomorphic to the cone on the link. That fails for stratified spaces.
 
Um, I guess, DogAteMy.
 
Do you work by quotients? That's how we did it on the homework
Or well, perhaps not
Just stack 0s on top
 
And then I'm done. For unit singular matrices, anyway.
 
6:55 PM
You can, @Daminark, but you need not.
Right, DogAteMy.
But it's homogeneous, so therefore a cone.
 
What's up
 
Something to the effect of, you necessarily have an eigenvalue/eigenvector because FTA
 
Homogeneous meaning I can multiply by scalars and still be in the set? @TedShifrin
 
@user379685: I don't understand your question. Should the =0 be missing?
 
Now take the bottom-right $n-1\times n-1$ submatrix
 
6:56 PM
Yes, DogAteMy.
 
Got it.
 
Create a basis for $\mathbb{C}^{n-1}$
In which that submatrix is upper triangular
 
Right ... and you get the upper row automatically.
 
Now stack 0s on each vector
 
@TedShifrin it should be there
 
6:56 PM
Along with the original eigenvector
Should give you what you want
 
@Daminark: You can't stack 0's, but you know the first column is 0 other than the diagonal.
 
@TedShifrin Less trivial but still true is is that the homeomorphism type of a neighborhood of any singular pt in an algebraic variety is homeomorphic to the cone on the link, then.
 
You get garbage in the top row, but who cares?
 
No I mean, if your vector in $\mathbb{C}^{d-1}$ is $(v_1,\ldots, v_{d-1})$
 
Right, @MikeM.
Oh, on those vectors, yes, @Daminark.
 
6:57 PM
@TedShifrin Earlier I was thinking that for unit 3x3 singular matrices you get something horrible like $\Bbb{PR^2\times PR^5}$
 
Ugh, where is my brian today.
 
Consider the vector $(0,v_1,\ldots,v_{d-1})$
Yeah
 
(I noticed the misspelling in the above as I was typing it, but I left it in because it seems appropriate.)
 
@Semiclassical That's condescending
to him
 
That's much harder, DogAteMy, because you now have rank 1 to worry about, not just rank 0.
 
6:58 PM
Yeah. Is PR2xPR5 anywhere close to the answer?
 
Pah, my brian should be working harder.
 
Also, I need to leave soon.
 
I actually don't know, DogAteMy. I'd have to ponder. The higher-dimensional question I gave you is of a different nature.
 
slacking off on a monday? pshaw.
 
All right. I'mma go now. Bye!
 
6:59 PM
You would need to find the answer inductively by stratifying by how singular the matrices are.
Find the homeomorphism type of rank 1, then do rank 2.
 
later @akiva
 
@user379685: So just use first and second partial derivatives and write out the Taylor polynomial.
 

« first day (2378 days earlier)      last day (2637 days later) »