« first day (5209 days earlier)      last day (107 days later) » 

00:03
Do multi-foliated systems appear in algebraic topology?
In algebraic topology? No, not to the extent of my knowledge
Perhaps in geometric topology, in some corner
Algebraic topologists don't care much about foliations
00:22
@BenSteffan Ah - that's kind of surprising, because multi-foliated systems can be used to produce examples of spaces with complex homotopy types or cohomological structures, especially if the foliations intersect in ways that yield algebraic varieties. 🧐
we don't care much about algebraic varieties either
and our spaces already pretty much all have complex homotopy type or cohomological structure
well cohomological structure is part of the homotopy type of course but
 
3 hours later…
03:53
@Jakobian obviously ;)
 
5 hours later…
08:36
In metric spaces separable iff 2nd countable iff Lindelöf. In general what would be the hierarchy (if there is any)? I can see that generally 2nd countable implies Lindelöf, and 2nd countable implies separable. Does Lindelöf imply separable?
 
1 hour later…
09:59
@nickbros123 neither
In mathematics, S-space is a regular topological space that is hereditarily separable but is not a Lindelöf space. L-space is a regular topological space that is hereditarily Lindelöf but not separable. A space is separable if it has a countable dense set and hereditarily separable if every subspace is separable. It had been believed for a long time that S-space problem and L-space problem are dual, i.e. if there is an S-space in some model of set theory then there is an L-space in the same model and vice versa – which is not true. It was shown in the early 1980s that the existence of S-space...
Separable space which isn't Lindelof: $\mathbb{N}^\mathfrak{c}$
Compact space which isn't separable: $\omega_1+1$
 
1 hour later…
11:14
Is there a moderator around?
 
1 hour later…
12:34
@robjohn I'm using Chrome on an Android tablet. Up until some months ago, I could just paste in the javascript into my address bar (had to write javascript: in front of it because that was concatenated) but it worked
Now when doing the same, it does a google search
@Claudio I haven't used it before either
13:32
@AlexanderPope i think this question would be a good question for MSE. note that if W is any "polyhedron" := a finite intersection of sets of the form {y in R^n: <y, w_i> >= c_i}, with w_i in R^n and c_i in R, and if p in W is such that <p, w_i> = c_i for at least one i, then your set {p + t(q-p): q in W, t >= 0} will pretty clearly be contained in the intersection of those sets {y in R^n: <y, w_i> >= c_i} for which <p, w_i> = c_i [here < , > is the usual Euclidean inner product].
I hate how the word "polyhedron" means like a dozen different things depending on whom you ask
my guess is that in the case of such p, your set is actually equal to this finite intersection (and is hence closed). this is a symbolic version of what thorgott mentioned earlier about the relevant geometry. for context i might add the paper you were reading where this problem arose, and maybe also the fact that it doesn't hold for general closed convex W (as thorgott noted above).
@Thorgott haha, yeah.
there must be some short and slick proof of this, maybe out of linear algebra using a map A for which W = {y in R^n: Ay >= c} (here A encodes the vectors w_i in the representation above, c encodes the scalars c_i, and >= is componentwise inequality). convex optimization people are all about identifying this kind of thing.
14:11
@copper.hat^
Welcome back to the chatroom @robjohn
14:50
@leslietownes just to haunt us
15:31
we know the theorem (cantor) that $X$ is complete metric space if and only if for every non empty, nested sequence of closed sets $\cdots F_3 \subseteq F_2 \subseteq F_1$ whose $diam(F_n) \to 0$ as $n \to \infty$ we have $\cap_{i=1}^{\infty}F_i=\{x\}$. Can we relax the condition that their diam goes to $0$, add the fact that $X$ is totally bounded, and still conserve the result? I am basically trying to prove the implication "complete+totally bounded implies compact"
what do you want to relax, exactly?
conserve what result? if F_n = [0,1] subset X := [0,1] with the usual topology, then you have that nesting, and the condition that diam(F_n) go to 0 has been "relaxed", and... ?
oh sorry, I want to basically say the intersection is non empty rather than single pint
point
basically get rid of diam-->0
add total bounded ness
and try to get intersection non emptty
3
Q: In what metric spaces is the intersection of nested closed sets non-empty?

MelodyI know that in a complete metric space, if $F_1\supseteq F_2\supseteq...$ are closed, and $\text{diam}(F_n)\to0,$ then $\bigcap\{F_n\}\not=\emptyset.$ John B. Conway claims falsely that even if $\text{diam}(F_N)\not=0,$ but is still bounded for all $n$, then $\bigcap\{F_n\}\not=\emptyset.$ I say ...

my approach to trying to prove that every open cover has a finite subcover is as follows: from total boundedness, we can show the space must be lindelof, so we can restrict our view to countable cover. then (for sake of contradiction) say this does not have a finite subcover, and somehow construct a nested sequence of closed sets, try to get some contradiction
@leslietownes the top answer there is using what I am trying to prove lol
15:40
42
Q: Totally bounded, complete $\implies$ compact

Emir Show that a totally bounded complete metric space $X$ is compact. I can use the fact that sequentially compact $\Leftrightarrow$ compact. Attempt: Complete $\implies$ every Cauchy sequence converges. Totally bounded $\implies$ $\forall\epsilon>0$, $X$ can be covered by a finite number of b...

is that relevant to your interests
when you 'add the fact that X is totally bounded' are you conserving X being complete
yes yes
@nickbros123 If every monotone sequence of non-empty closed sets has non-empty intersection, then every countable family of closed sets with finite intersection property has non-empty intersection, which should be equivalent to countable compactness, which is equivalent to compactness for metric spaces
i was waiting for one of our general topology weirdos to weigh in on this :)
I think there's just two, maybe 3
unless you count algebraic topologists
which is still an insane amount for how many people visit this chat, and how unpopular general topology is
anyway the take away from this is that by removing the diameter condition and assuming the intersection is non-empty instead of a singleton, we end up with compactness and not completeness like you wanted, Nick
@Jakobian i'm not sure that general topology even exists outside of this chat
15:52
yeah somewhere around this ballpark, I waas almost converging to the idea of taking an infinite subset E of X, and taking $\varepsilon=1$. We would get finite $1-$balls in $E$, so infinite poins of $E$ are in one of these 1 balls. I pick x_1 in this ball. then within the 1-ball I take a smaller ball size, maybe $1/2$ and use the fact that this 1-ball is totally bounded-> find another 1/2 ball containing infinite points and x_2 in this 1/2 ball. keep continuing for $1/2^n$ sized balls

I think finally we will get a cauchy sequence which will converge from completeness, this giving us limit p
@leslietownes I'm not sure either. On mathoverflow the same person which answers me, K. P. Hart, is the same person that contributes to Encyclopedia of General Topology. Which is insane to me
the other thing to keep in mind, jakobian, is that the world only has like four or five dozen people in it
"everyone else" is just text on some kind of device
@leslietownes its actually 8
8 literally? or 8 dozen?
no, like literally 8
8.025 person to be precise
15:58
actually my initial approach to showing total boundedness + completeness ==> compact was to take a countable cover that had no finite subcover ($\{G_n: n \in N\}$) and then say $x_1 \in G_1 ^C$, $x_2 \in G_1^C \cap G_2^C$, $x_3 \in G_1^C \cap G_2^C \cap G_3^C$ and so on. This would form a nested sequence of non empty sets. If I was able to show that this intersection was non empty, (via completeness + total boundedness) then I would have shown $\cap_{n=1}^{\infty}(G_i)\neq \emptyset$

which is contradiction
jakobian: can't wait to see who the 9th is going to be
@Jakobian by pigeon hole princple this chat must have multiple accounts that are the same people
@nickbros123 wait, which one of us is talking?
@nickbros123 well, you are right that if you assume $X$ is totally bounded and complete, and show that by taking a sequence of closed sets $F_n$ which would be monotone and prove that $\bigcap_n F_n \neq \emptyset$ then that's enough to see that $X$ is compact
Depending on which order things are proven, you might want to have some results which hold only for metric spaces
without anything concrete there's not much to say
yeah I have to prove totally bounded + complete implies nested closed sequence has non empty intersection. Thats the black box in my attempt :)
I thought of a different method though, showing that totally bounded + complete implies limit point compact, which was more easier
for me
16:11
How about this, given sequence $(x_n)$ pick closed set $F_1$ with $\text{diam}(F_1)\leq 2^{-1}$ and $F_1\cap \{x_n : n\in\mathbb{N}\}$ is infinite. And so on by induction, by decomposing $F_1$ into finite amount of closed sets of diameter $\leq 2^{-2}$ and so on
Thus you will obtain a sequence $(F_n)$ such that it consists of non-empty closed sets of decreasing diameter and $F_k\cap \{x_n : n\in\mathbb{N}\}$ is infinite
Since $X$ is complete, now we have there must exist $x\in \bigcap_n F_n$ and you want to prove that $x$ is in fact the limit of a subsequence of $(x_n)$
well, for this approach lets assume $(x_n)$ consists of infinite amount of distinct elements, otherwise this is pretty clear that this sequence has a convergent subsequence and the construction doesn't work then
or we can just start with an infinite set i think. Limit point compact says infinite subset has a limit point, which is same thing as saying any sequence has a convergent subsequence
Just pick $n_k$ such that $n_1 < n_2 < ...$ and $x_{n_1}\in F_1$, $x_{n_2}\in F_2$ and so on
then $x_{n_k}\to x$
@Jakobian this is given by total boundedness and that $(x_n)$ has infinite amount of distinct elements
yeah I understood the technique
yeah total boundedness gives you a decomposition of $X$ into finite amount of sets of diameter $\leq \varepsilon$, which you can take closures of since it doesn't change the diameter
@nickbros123 This was my approach, kinda similar
anyways thanks
I wouldnt have gone into these things, my friend asked me for a proof of heine-borel. I was going to show him the classic proof technique of "splitting the n-cell into 1/2^n parts" trick, but then I thought why not go a different approach: since any bounded set in $R^n$ is totally bounded, and any closed subset of complete set is complete, we get in 2 lines that closed + bounded = compact
16:22
Why not just prove that $[0, 1]^n$ is compact
cuz whats the fun in that
I don't know, all those elementary facts about metric spaces, I'd rather try and prove equivalence of all of them and be over it than try to look for all the different possible approaches
yeah i would basically just paste stuff from MSE or textbooks in response to all of this
which eliminates the fun of doing it yourself but is a whole lot faster
time being money, etc
I count time in the amount of general topology theorems I can put out
well I, for one, am for the first time proving that complete + totally bounded iff compact, and second, college is for wasting time anways, and third, I dont think im wasting time
apart from typing this obviously, that is technically wasting time
16:28
I didn't say you are wasting time, there is a perspective from which just existing is wasting time
no its fine its just a question of the goal, if the goal is to get through it then yes, go through it
but if its like, how do i prove this, divorced from any perceived value in personally going through it, i would say, paste from MSE or textbook
Its just that I'd rather you discuss something more interesting than reproving the overdone theorems
fair enough
even if you were to reprove overdone theorems in a more general setting - that sounds interesting
this is something that sometimes gets people downvoted on main, if you are asking something that is well known and in a lot of books it is helpful to acknowledge that and not hide it behind, i'm generalizing some unknown thing about diameters
had this begun with 'i am personally trying to prove that in a metric space, completeness + totally boundedness is equivalent to compactness" its the same conversation but slightly different perspective
16:31
I am this one guy who doesn't see excitement in the world anymore and by world I mean questions that undergraduates come up with
i just like knowing context if there's context, i vastly prefer any of this stuff to disguised attempts to prove the riemann hypothesis
By the way this is my way of humour
we spell that "humor" where i come from
sound of racking shotgun
we spuell thuat "humour" whuere I cuome fruom
I should stop laughing at my own jokes, they're not that good at all
@Jakobian what about this one: classify all open subsets of $\mathbb{R}^n$ on which any $C^1$-function with bounded gradient is Lipschitz.
16:36
too much differentiability
i'm still not convinced that thorgott and jakobian are different representatives of the 8 human beings
I dive into uniform spaces and topological spaces, but I never dive into anything differentiable
@leslietownes this was just my attempt at proving that thing :) my original question was a precursor question to this
imo it's a fun question, if I'm allowed toot my own horn (the question is from my undergraduate self)
@nickbros123 its all good, i'm just happy not to see famous open problems buried in the woodwork of what you're asking :)
16:40
@Thorgott depending on what you mean by $C^1$-function, shouldn't that just always be the case because of mean value theorem?
the mean value theorem only works if you have convenient line segments in your open sets
a sufficient condition is quasi-convexity (there is a constant $C$ s.t. any two points can be connected by a path of length at most $C$ times their distance)
do you mean functions from $U\to \mathbb{R}^n$ where $U$ is this open subset
Actually why not $U\to\mathbb{R}$, should be equivalent
right, those are all that matters
16:46
Noticed that the wikipedia page for mean value theorem has an error
> In particular, when the partial derivatives of $f$ are bounded, $f$ is Lipschitz continuous
This needs something like convexity of $G$
yeah, that's a bit sloppy
quasi-convexity suffices, as I mentioned, but I actually don't know of any connected open satisfying the property that isn't quasi-convex
16:58
its fixed now
if not me then who
and its just maybe a minute to fix that
Is there a well-known theorem that tells one how to show path connectedness of intersection sets of high dimensional manifolds that intersect nicely?
obviously randomly intersecting will not yield path connectedness which is why I'm only interested a high degree of regularity for the intersections
17:24
I proved this:
Does it help with
Or is that independant from it?
WITCHCRAFT!
@XanderHenderson lmao
Long time no see, Xander
(Honestly, I feel like this was something that I needed as part of my masters work; and it came up again when I took ergodic theory in my phd program).
17:26
@XanderHenderson Are you talking about the problem or to Modular?
@ILikeMathematics I don't know what you mean by "Modular". I am talking about the approximability of irrational numbers.
I meant the person who wrote something above this, ModularMindset
Hrm... I think that the way that you have phrased the first problem, it is even kind of a trivial consequence of density.
@XanderHenderson Yeah that might be the case, though since we haven't covered density or even any terminology like that in this course, I had to prove it 'elementarily'
Sure.
What class is it?
17:29
It's an analysis course
Ah. That makes sense.
But yeah, the second problem intuitively seems completely obvious
Indeed, the second problem you posted essentially asserts that the set $\{m\tau + n \mid m,n\in\mathbb{Z}\}$ is dense in $\mathbb{R}$.
(more or less---I'm still in a bit of a brain fog; I've been ill this week)
@XanderHenderson Do you have an idea on how we could prove it 'in elementary terms'? I guess we just need to construct one number both in I and this set
Using only elementary tools? Off the top of my head? No idea.
But I'm not braining very good.
17:33
I think an Ansatz could be using the previous problem, setting $\varepsilon = b$
Then we know there exist integers $m, n$ with $0 < m \tau + n < b$
all of those type of results strike me as hitting just above the dividing line between elementary and non elementary
@leslietownes Yeah, that's kind of my thouht.
Now I think if $a > 0$, we are done?
@ILikeMathematics $b$ isn't guaranteed to be positive...
Yeah you're right
17:34
like, are you worrying about that? congratulations on your admission to graduate school
haha
But if you can make $m\tau + n$ small, then you can apply the Archimedean property.
Make $0 < m\tau + n < |b-a|/2$ (or something like that).
Yeah we can do that
In what way is this useful though?
Are you familiar with the Archimedean property?
it's really best for everybody if we don't ask that question.
@XanderHenderson Yes
17:40
@ILikeMathematics So... make $m\tau+n$ small (the first problem says you can), then multiply it by something big enough to make $k(m\tau+n)$ land in $(a,b)$.
We can find an $n \in \mathbb N$ with $n (m \tau + n) > |b - a|/2$
@XanderHenderson Yeah that sounds very reasonable, thanks a lot!
@XanderHenderson Could we also pick epsilon = 1/N for some big enough N?
Since your idea seems to just rely on the Archmedean property
@ILikeMathematics But... why? The point is that you want to make $m\tau + n$ small enough that you can ensure that some multiple of it lands in $(a,b)$.
Picking an arbitrary $N$ isn't really going to help---you want to estimate it in terms of $|b-a|$.
The idea is that if $(k-1)(m\tau + n) < a$ and $k(m\tau + n) > a$, then $k(m\tau+n) < b$.
(assuming that $a, b > 0$).
(the case when $a\le 0$ and $b > 0$ is just applying problem 1, and the case when $a,b<0$ is the same as $a,b>0$, but with the signs reversed).
I am likely going to regret asking this here but as this is a chat a discussion can be made about it before it get's deleted
What do you guys think of the approach I took here for the conjecture:
https://math.stackexchange.com/questions/4995324/collatz-conjecture-convergence-proof

(if it is ok with the mod I post this here)
@XanderHenderson Oh, and |b - a|/2 ensures that we don't get > b aswell, it will only be 'halfway there' when going from k - 1 to k
Thanks!
That's the idea.
18:21
@AlessandroCodenotti do you know if image of a finite-to-one zero-dimensional space is zero-dimensional (looking for positive results, not counter-examples)
problem is in literature such maps are mainly treated for metrizable spaces
18:55
I just need it to be open I think
no wait. I need the isometry to be open (you guys probably don't know what I'm talking about by isometry here)
that's never happening
19:17
I need some result of the form, $f:X\to Y$ map between compact Hausdorff spaces, $\dim X = 0$, then $\dim Y = 0$
I'm going to try to search in Engelking's Theory of Dimensions, Finite and Infinite
19:29
I found a result which says this would work if my map $f$ would be open
19:40
@think_meaning_builds whoa, did someone turn on the convex-light?
@psie excuse the delay, will look at your question shortly.
hi, sanity check. When we define a discrete random varible, it's not necessary that we have a probability function right?
We just need the sample space and the event set
@SineoftheTime probability function?
A random variable is a measurable function of some kind. But to say its discrete you still need to have a probability measure on the domain
I meant probability measure
Can't I define a r.v. as a measurable function from $\Omega$ to $\Bbb R$ say?
and I don't use the fact that I have a probability measure
> But to say its discrete you still need to have a probability measure on the domain
If I say it's discete when $X(\Omega)$ is a discrete set in $\Bbb R$, do I need to have a probability measure?
19:56
You could do that and it would be a meaningful definition, but rather than discrete you should let the image be countable
@ILikeMathematics Are you familiar with the pigeonhole principle? Take a look at
In number theory, Dirichlet's theorem on Diophantine approximation, also called Dirichlet's approximation theorem, states that for any real numbers α {\displaystyle \alpha } and N {\displaystyle N} , with 1 ≤ N {\displaystyle 1\leq N} , there exist integers p {\displaystyle p} and q {\displaystyle q} such that 1 ≤ q...
If $X:\Omega\to \mathbb{R}$ is a discrete random variable with how we use it, then there exists a random variable $Y:\Omega\to\mathbb{R}$ such that $X = Y$ a.e. and $Y$ has countable image
and conversely such random variable $Y$ will always be a discrete random variable, so we're not losing anything with this definition
Makes sense. When I've taken probability, I remember that we did not need a probability measure when defining discrete r.v. whereas the definition of r.v. was: $X: \Omega \to \Bbb R$ is a continuous r.v. (with density $f$) if its distribution function is: $F(x)=p(X\le x)=\int_{-\infty}^x f(t)dt$
So we need a probability measure
@copper.hat no hurries and worries :) I think I understand now that the completion of the pushforward of a measure by measurable isomorphism is the pushforward of the completion of the measure. That's clear.
$\Omega$ has a probability measure always
And we can define the distribution function for both discrete and continuous random values
And also for those that are neither discrete nor continuous
20:07
@PM2Ring That's the result I had in mind---I just wasn't sure how "elementary" it was vis-a-vis whatever @ILikeMathematics was looking for.
@psie how do you define the completion of a measure? via outer measure, appending null set subsets, other?
i won eu60 from Paddy Power for betting on the winning president
@VladimirLysikov yes, I know that. I my doubt was: do we really need $p$ on $\Omega$ if we define a discrete r.v. as a measurable function from $\Omega to $\Bbb R$ s.t. $X(\Omega)$ is a discrete set?
@SineoftheTime I guess technically not
But for me it would be weird to call this a random variable
I would just call this a measurable function
@copper.hat the completion is $\overline{\mu}(E\cup F)=\mu(E)$ where $E\cup F\in\overline{\mathcal M}$ and $F$ is a subset of (usually) a Borel null set.
What is unclear in my question, and Ramiro has been very helpful, is how to obtain $$\int_{\mathbb{R}^n} f(x) \,dx = \int_0^\infty \int_{S^{n-1}} f(rx') r^{n-1} \,d\sigma (x') dr.\tag{2.50}$$When we work with the incomplete measure $\rho\times\sigma=m_\ast$, then formula (2.50) for $f=\chi_E$ reduces simply to $\rho\times\sigma(E)=m_\ast(E)$ (where recall, $m_\ast(E)=m(\Phi^{-1}(E))$ and $m$ is Lebesgue measure).
When we have $\overline{\rho\times\sigma}=\overline{m_\ast}$ however, I am stuck at how this becomes formula (2.50).
Ramiro has been alluding to the complete version of Fubini-Tonelli, and I think using that theorem, we'd have $$\int_{(0,\infty)\times S^{n-1}}f\,dm_\ast=\int_0^\infty\int_{S^{n-1}}f(r,x')\,d\sigma(x')d\rho(r).$$
@VladimirLysikov I see
thanks
20:38
@XanderHenderson Fair enough. IMHO, the pigeonhole principle isn't that hard to understand, and that Wikipedia article is fairly readable. (As I'm sure you've noticed, many Wiki mathematics articles gradually become overly complicated over time, making them almost impenetrable to new students).
@PM2Ring Honestly, I couldn't remember how the proof went. Once you say the words "pigeonhole principle", I think it clicks pretty quickly. But I am really not braining well today.
i think of the pigeonhole principle as meaning that some value is $\ge $ the average value.
@psie $\overline{m}_*$ is a pushforward measure just because $\lambda$ (Lebesgue) is.
The completion of $m_\ast$ is indeed the pushforward of the complete Lebesgue measure under $\Phi$, the measurable isomorphism.
so, are you asking why $\int f dm_* = \int f d \overline{m}_*$ for Borel $f$?
(may take me some time to catch up, am slowly returning to 'math mode'.)
I wish you had the book :) I suppose you don't?
it is claimed that the integral of any integrable Borel function $f$ can be written as $$\int_{\mathbb{R}^n} f(x) \,dx = \int_0^\infty \int_{S^{n-1}} f(rx') r^{n-1} \,d\sigma (x') dr,$$ where $\sigma$ is a surface measure on the unit sphere. Taking the product of this surface measure with the measure $\rho(E)=\int_E r^{n-1}\,dr$, we obtain $m_\ast$.
Now, the formula $\int_{\mathbb{R}^n} f(x) \,dx = \int_0^\infty \int_{S^{n-1}} f(rx') r^{n-1} \,d\sigma (x') dr$ is just a restatement of $\rho\times\sigma=m_\ast$ when $f$ is an indicator function.
My question is; does the formula hold for Lebesgue measurable functions?
Folland says that we need to consider the completion of $\rho\times\sigma$.
20:57
if $f$ is Lebesgue measureable then there is some $\tilde{f}$ Borel measureable that equals $f$ ae.
correct
@copper.hat and the a.e. is with respect to the complete measure
so $f=\tilde{f}$ $\overline{\mu}$-a.e.
21:12
I guess one could simply argue like that. I will try to write down an answer.
i need a few minutes, i may have a copy of Folland somewhere
@Jakobian Well this is just false
There is a 2-to-1 surjection $2^\Bbb N\to[0,1]$ I think
@psie found it. will take me a little longer. i think i understand your concerns.
I can only give results that bound the dimension of the domain in terms of those of the fibers and the codomain
@copper.hat cool :) take your time
22:17
@AlessandroCodenotti yeah that's a good example
1
A: Spaces with every compactification $0$-dimensional which aren't locally compact

AnonymousLet $\omega$ denote the natural numbers and let $N$ be a countably infinite discrete subset of $\beta \omega \setminus \omega$. If $X = \beta \omega \setminus N$, then $X$ is not locally compact, and every compactification of $X$ is zero-dimensional. That $X$ is not locally compact at any point...

someone gave me this answer to my post and I was thinking about how could I salvage it
It seems like $X$ might have only zero-dimensional compactifications, but the proof doesn't show it
Here its obvious that we have a function $f:\text{cl}_{\beta\omega}N \to \text{cl}_{\beta\omega}N\setminus N \cup K\setminus X$
with $N$ mapping to $K\setminus X$
and that existence of such function induces a compactification $K$ of $X$
and that $\text{cl}_{\beta\omega} N\cong \beta\omega$
so we can assume $N = \omega$
and the question becomes, suppose $f:\beta\omega\to K$ is a continuous surjection such that $f\restriction_{\beta\omega\setminus\omega}$ is a homeomorphism onto its image, is $K$ zero-dimensional?
in principle we are only modifying the natural numbers... it sounds like it should be
and $f[\omega]\cap f[\beta\omega\setminus\omega] = \emptyset$ of course
22:46
its weird that even taking such quotient is possible, maybe I can prove $|f^{-1}(f(n))| = 1$ for this
I have a notation question
@ModularMindset Read the room topic. "Just ask; don't ask to ask."
I didn't ask to ask
no that's definitely possible, a surjection $\omega\to\omega$ with finite fibers induces a surjection $\beta\omega\to\beta\omega$ which maps remainder to remainder
are there even countable spaces which aren't strongly zero-dimensional
maybe those are the only examples
no wait I didn't prove it maps remainder homeomorphically
but if I were to prove it must have one-point fibers then it would have only one compactification at which point $\beta X\setminus X$ is a one-point set, so that's not possible either
@Jakobian maybe this does induce a homeomorphism of $\beta\omega\setminus \omega$...
23:06
Here's the notation in question:

$$
\mathcal{V}_{\mathcal{I}_{\Gamma}} := \sum_{i=1}^4 \mathbf{v}_i |_{\mathcal{I}_{\Gamma}}.
$$

This is meant to convey vector fields on surfaces being added up precisely along a lower dim. strata for example where the surfaces tranversely intersect (1 dim. strata).
$\mathcal V$ is a vector bundle on $\mathcal I_{\Gamma}$
the vertical line $|$ means restricted to
A typical basis set of $\beta\omega\setminus \omega$ is of the form $A' = \overline{A}\setminus \omega$ where $A\subseteq \omega$. Then $f(A') = \overline{f(A)}\setminus \omega = f(A)'$, I am pretty sure. For $f(A)' = A'$ one just needs for $f(A)\setminus A$ and $A\setminus f(A)$ to be finite. So for example if $f$ modifies only a finite amount of points of $\omega$ then $f$ restricted to $\beta\omega\setminus\omega$ should be the identity
this is fairly common notation
My question is: Can this notation be made better?
I still want to express the same underlying concept but maybe there are too many symbols?
$\mathbf v_i$ are vector fields.
I think a more advanced mathematician might re-express the idea in terms of the direct sum of the tangent bundles, instead of only looking at specific sections of the tangent bundle.
$$
\mathcal V_{\Gamma} = \bigoplus_{i=1}^4 TS_i |_{\gamma},
$$
where $TS_i |_{\gamma}$ denotes the restriction of the tangent bundle of $S_i$ to the loop $\gamma$. This sum represents the addition of tangent vectors from each intersecting surface along $\gamma$.
Correct me if I'm wrong but this second formulation involving the direct sum seems to capture the main idea a bit better in terms of notation
23:23
I am pretty sure I got it actually
@ModularMindset No, but you didn't ask a question. You announced that you had a question to ask. You don't need to do that. Just ask.
@ModularMindset The point of notation is to communicate. If it is fairly common notation, then you are likely to be able to best communicate your ideas using that notation. Don't rock the boat.
Using non-standard notation is generally a bad idea.
Suppose I have a closed differential 1-form $\omega$ defined in an open subset of $A \subseteq \mathbb{R}^3$ (which is of course of class $C^1(A)$. My professor said that $\omega$ is locally exact, which means I can take $x_0 \in A$ and a ball $B_r(\mathbf{x}_0)$ s.t. $\omega$ is exact in it, since the ball is simply connected.
I was wondering: can I extend this to the whole open set just by considering a series of balls, which in the end fully cover $A$
That doesn't seem to work.
if I add the assumption that $A$ has no holes in it
23:28
I don't really know the theory you are working in, but you can't extend simply connectedness in this way.
well if $A$ is simply connected you can, yes
An annulus is not simply connected, but every point in an annulus has a simply connected neighborhood.
Oh... well, if $A$ is simply connected...
In $\Bbb R^3$ an open set can have a hole and be simply connected though
wait, now that I think about it
although picking local preimages under $d$ and trying to arrange for them to glue together on all of $A$ doesn't strike me as a great way to go about it
@SineoftheTime the hole doesn't matter
23:30
yeah
@BenSteffan The hole is the best part!
no hole = simply connected works in $\Bbb R^2$
You get a much higher ratio of glaze / toppings to dough!
@XanderHenderson I agree, but your 1-forms don't see it :)
2-forms though
Honestly, all this stuff about $1$-forms and $2$-forms and urmom-forms is beyond my ken. Not something I ever studied.
23:32
@BenSteffan yeah that's what I was thinking: since the 1-form is exact in each ball, I'd be able to find a potential $U$ in every ball, but then I'd need to glue everything together
well your professor argued that $\omega$ is exact over a ball since the ball is simply-connected
@SineoftheTime yeah, I was thinking in $\mathbb{R}^2$ indeed
so you just need to exchange the ball for $A$
I wonder if this locality result has some physical implications
@BenSteffan yeah it was indeed a stupid idea
maybe it's not relevant to you, but on non connected sets it's not true that two potentials differ by a constant
23:37
non connected sets...
that means also a non connected set, whose single components are connected?
components are connected by definition
that's why they're called connected components
0
A: Spaces with every compactification $0$-dimensional which aren't locally compact

JakobianThis is to finish flawed argument of Anonymous that $X$ has only zero-dimensional compactifications. I'll prove that $K$ is totally disconnected, which is equivalent since $K$ is compact. Let $S\subseteq K$ be connected and $|S|\geq 2$ where $K$ is a compactification of $X$. Since $\beta\omega\se...

I mean, if you cover you open set with balls, and two of them are disjoint, you can't find a potential just adding a constant
Update: I've fixed the argument and proved that indeed, the space $\beta\omega\setminus N$ where $N\subseteq \beta\omega\setminus\omega$ is an infinite countable discrete set has only zero-dimensional compactifications
but maybe this is irrelevant
23:40
So its my first example of such space which isn't locally compact
@SineoftheTime Oh you mean I can't glue the potentials together, yeah that was what I tried to say before but you just gave me the answer hahah
@SineoftheTime so my idea was flawed to begin with :p I'd have to avoid having disjoint balls
so basically the best idea is to prove that the open set I'm working with is simply connected, which is pretty much impossible hahaha
why is it impossible? :(
because I need to show that every path can be contracted continuously to a point
which probably is like 3-4 years of a math degree I suppose ahahah
well that's always true
Path? or loop?
23:46
you need to show that each loop can be contracted to a constant one
I meant closed loop on the set
@Jakobian correction: I didn't fix the argument as it was unsalvageable, I merely wrote my own using similar ideas
@Claudio It generally shows up towards the end of an introductory topology class, which is often taken in the third or fourth year of undergraduate coursework at American institutions.
@BenSteffan wdym, this differs from the idea my professor gave me? What is a constant loop??
a loop is a map $S^1 \to X$. A constant loop is one whose image is a single point
23:47
@Claudio A loop is a function from an interval to your space.
Is the open set you're working with convex?
(With the additional constraint that it maps the endpoints of the interval to the same point).
A constant loop sends everything in the interval to a single point.
Now I'm glad that I didn't waste my time with trying to prove that every space with this property must be locally compact
@BenSteffan's definition is equivalent, but cleaner.
23:48
I see, again she also said that this definition won't be of any use anyways for us, we just work with intuition
then what are you worrying about :^)
Like Jewish cooking!
Just use feelings!
(and then add some more butter and/or schmalz)
@BenSteffan it seemed scary and interesting ahhaha
for instance, she said that even showing that a ball in $R^3$ is simply connected is difficult
I disagree
I, too, disagree
23:51
@Claudio I guess... for very expansive definitions of "difficult"...
showing that a ball is simply connected is not very hard
showing that $\mathbb{R}^3 \setminus \{0\}$ is simply connected is, at your level
Maybe I got something wrong I don't remember exactly, she seemed traumatized when talking about simple-connectedness in general :)
But... like, a ball is convex. Convexity makes things easy.
I see that's why @SineoftheTime asked me that
convex sets are easy, and sets that are homeomorphic to ones you know to be simply connected
for instance, the ball is also homeomorphic to $\mathbb{R}^3$
23:53
Yeah, but $\mathbb{R}^3$ is convex. :P
there's a result on equivalence between closed and exact forms when the open set is a star domain, maybe you can use that
@XanderHenderson true, that's why it's contractible
Yer mum is contractible!
homotopically speaking my mum is copy of $S^1$
and so are you
@BenSteffan I have more holes than that!
23:56
I'll reask her tomorrow, maybe it wasn't a ball
Maybe a sphere?
what, did you get your ears pierced in your rebel years @XanderHenderson
That is a bit less trivial. Should still be doable.
does that involve some sort of stereographich projection thing to show???
maybe it was a horse, since physicists approximate horses with spheres
23:57
@XanderHenderson "A bit less trivial" might be underselling it
@XanderHenderson it might have been a sphere
I don't think it's easy at all
@BenSteffan No, but I've got nostrils and ears and some other holes, too.
@BenSteffan I like understatement.
it seems like my observation that any countable subset of $\beta\omega$ is $C^*$-embedded isn't just trivial in my argument as I thought the first time, so I do use it on the last line when claiming that $\beta\omega\setminus \omega$ is zero-dimensional
you think you have a proof and then you realize that there are surjective loops $S^1 \to S^2$ and then everything is painful
@XanderHenderson these aren't true holes, topologically speaking
but of course we're simplifying things here :)
23:59
@BenSteffan Sure they are.

« first day (5209 days earlier)      last day (107 days later) »