« first day (2728 days earlier)      last day (2299 days later) » 

12:00 AM
in this case its just talking about A subset of $\Bbb R $
 
If you put the discrete topology on $A$ the closure of $A$ would be just $A$
@Faust Ahhh okay
 
im still not sure why the answer isnt [1,0]
is that set closed?
 
Yes.
But it's not the smallest closed set containing $A$
 
hmm ok its interior is empty though?
i didnt mean [1,0] lol
 
Look at the complement of $A \cup \{0\}$. Is it open?
@Perturbative lol, all my topology knowledge comes from Rudin :P
 
12:03 AM
@orbit-stabilizer it looks open
 
Okay, what about $A^c$?
 
@Faust If you have $A = \{(0, \frac{1}{n}) \ | \ n \in \mathbb{N}\} \cup \{0\}$, then it's closure in $\mathbb{R}$ would be $[0, 1]$
 
it contains 0
so its neithier open or closed
@Perturbative but thats not what i have i just have the points $A=\{1,\frac{1}{2},\frac{1}{3},...\}$
 
@Perturbative, wait if $A = \{ \frac{1}{n} | n \in \mathbb{N} \}$, then isn't $\bar A= A \cup \{0\}$?
 
ok so its not a simple question
cause at least one of you is incorrect lol
 
12:07 AM
Well, I think it's straight forward with limit points.
No. I think @Perturbative misread your question
 
hmm
 
Yeah I think I did
 
He's looking at the open intervals
 
oh yeah
i belive that
he isnt incorrect but answered a different question
 
What's your defintion of an open set?
 
12:08 AM
last stupid question what is the interior of A ?
 
Just write down $A$ in set builder notation. Like what I did above
 
we dont seem to have a definition of an open set...
 
Set of all $\frac{1}{n}$, $n\in\mathbb{N}$.
what.
oh
 
Okay so $\mathbb{R}$ is a sequential space, so limit points of sequences coincide with topological limit points (I think......someone blow your horns at me if this is wrong), the sequential limit point of $A$ if $\{0\}$ so $A' = \{0\}$, and hence $\bar{A} = A \cup \{0\}$
 
I haven't taken a topology class... is it normal not to define an open set?
 
12:10 AM
yeah no definition of open other than open is in T
 
@Faust That is the definition of an open set :p
 
lmao
 
@Perturbative looks right
 
If $\mathcal{T}$ is a topology, then any $U \in \mathcal{T}$ is an open set
 
@Perturbative what about a set is open iff every point of the set is an interior point?
 
12:12 AM
topology is wierd
 
D:!
Topology is hard and confusing
 
@orbit-stabilizer Urhhhm I haven't thought about that in a while, Rudin likes that definition lol
 
@orbit-stabilizer A set in $\Bbb R$ is open if, for every point in the set, there exists an open interval surrounding the point contained in the set. Right?
 
some of its terms is in complex analysis OMG
 
so im going with the interior of A is the empty set?
 
12:12 AM
nooo
 
WHATTTTTTTTTT
 
(Open intervals are $(a,b)$)
 
isn't it that if it's open then we can find an interior point?
 
Interior points are defined in terms of open sets.
 
points in $\Bbb R $ are closed?
 
12:13 AM
Now, let $\tau$ be the set of all open sets in $\Bbb R$, and consider the ordered pair $(\Bbb R,\tau)$. @orbit-stabilizer
 
@Faust yes
 
The ordered pair $(\Bbb R,\tau)$ is called "a topology".
 
@AkivaWeinberger okay
 
With that in mind, let's define what we mean by a topology
 
puts on learning hat
 
12:14 AM
A topology is an ordered pair $(X,\tau)$ where $\tau$ is a set of subsets of $X$ with a few rules
and note that this will be true for the $\Bbb R$ thing as well
The empty set and $X$ are both in $\tau$
It's closed under finite intersection
and it's closed under arbitrary union.
 
Akiva Munkres to the rescue
 
what what
 
@AkivaWeinberger sounds good. So there are an infinite number of topologies we can put on $\mathbb{R}$.
 
I think that's all of them. Anything that satisfies those rules is called a topology; $X$ is the base space, and $\tau$ is the set of open sets in $X$.
 
I remember that X and empty set are in the topology
 
12:15 AM
So, essentially, in general, "open" is kinda not really defined, it's just an element of $\tau$ which satisfies a few rules
 
@AkivaWeinberger Yeah that's all, just could be confusing to say "closed" under finite intersection etc. when there's another topological meaning for closed
 
So we could have this topology on $\mathbb{R}$: empty set, R, and $\{1\}$.
 
so the interior of $A=\{1,\frac{1}{2},\frac{1}{3},...\} $is the empty set? @AkivaWeinberger
 
noooooooooo
 
@Faust Yes
 
12:16 AM
@orbit-stabilizer You could have $\mathbb{R}$ and $\emptyset$
 
what!
oh the interior.
 
@orbit-stabilizer You could, though it's a fairly boring one.
 
set of all interior points.
oops.
my bad
 
indiscreete topology is ftw @AkivaWeinberger
 
What does it mean geometrically if I have that particular topology?
 
12:17 AM
$1$ is isolated from the rest of the points, which are all infinitely close to each other
 
it means there is no neighbourhood around any point that doesnt contain everything
 
(You said $\emptyset$, $\Bbb R$, and $\{1\}$, right?)
 
Yeah.
It feels really strange though, because $\mathbb{R}$ is an ordered field.
So, you already have this notion of nearness.
 
Yeah, for your topology you just threw that all away
 
lol
 
12:18 AM
We lose the ordering on R when we define that topology?
 
hell yeah ;)
 
With just those three open sets?
 
Nah there's an order topology on R too
 
With those sets you lose em
 
12:19 AM
@orbit-stabilizer If we relabel the elements of our topology, we don't consider it a "new" topology
 
why is topology confusing? I keep getting my definitions mixed up oh lawd
 
It's isomorphic
Technically the word is "homeomorphic"
 
that's a bijection too isn't it?
 
(Not to be confused with "homomorphic" from group theory)
@usukidoll Yeah
 
@AkivaWeinberger whats the diffrence!
 
12:20 AM
Yeah, like from manifolds.
 
hom-e-omorphic, hom-omorphic @Faust
 
@usukidoll You have to take into account continuity too
 
That e
2
 
there's endomorphism and monomorphism too :S :S
 
Ayy, that was in the first few pages of Aluffi
 
12:20 AM
Homomorphisms are just functions, homeomorphisms are bijections as well
 
@AkivaWeinberger not what i meantee
 
I like Chapter 0
 
Personally, I'm a fan of Chapter 5.
 
@orbit-stabilizer It turns out that a function from $\Bbb R$ to $\Bbb R$ is continuous iff the preimage of an open set is open.
 
The plot really picks up after 5
 
12:21 AM
(Exercise: Why can't say say that the image of an open set is open?)
 
For a continuous map?
 
Yeah
Hint: Take $f(x)=x^2$
 
last stupid question the interior of $\Bbb Q $ is the empty set as well then ?
 
Is the image of an open set always open?
 
Or, just map everything to some constant
 
12:22 AM
the problem with time is that $\pi_1(S^1) = \Bbb Z$
 
@Faust Yes
 
so there must be a discontinuity
 
@AkivaWeinberger that works, right?
 
@orbit-stabilizer Yeah
 
@LeakyNun Where did the the fundamental group come from?
 
12:22 AM
I was thinking of how the image of $(-1,1)$ under the squaring map is $[0,1)$, which is not open
However, preimages of open sets will be open
 
Yeah, I remember that from Rudin last semester.
 
Thus, we can define continuous functions between any two topological spaces $(X,\tau)$ and $(Y,\upsilon)$ by saying a function $f:X\to Y$ is continuous iff for any $U\in\upsilon$, $f^{-1}(U)$ is in $\tau$
 
@AkivaWeinberger i think you may be assuming that your map is continous
 
@Faust Yes
 
@AkivaWeinberger for the love of god, $(Y,\sigma)$
 
12:23 AM
oh ok then i agree ;)
 
Two topological sets are homeomorphic if there is a bijection between them such that (a) it's continuous and (b) its inverse is continuous.
 
So back to my question, we lose the order on R when we define a different topology?
 
(This is equivalent to saying you can relabel the elements.)
 
@orbit-stabilizer There's a thing called the order topology on $\mathbb{R}$
 
@orbit-stabilizer Sure. You just have an uncountable set of elements, without a topology
 
12:25 AM
Huh. I thought a bijection was enough to relabel
 
@orbit-stabilizer $[0,1]$ and $(0,1)$ have a bijection between them, but they're not homeomorphic.
(They both have continuum many elements)
Any such bijection must be discontinuous.
 
@AkivaWeinberger thank you for the great explanations
 
@Perturbative sure... but the order on $\mathbb{R}$ doesn't need a topology. We define a relation on $\mathbb{R}$.
 
@orbit-stabilizer I would object to this statement on the basis that you can't even obtain the order on R with the standard topology
 
@LeakyNun relations?
 
12:26 AM
Right, yeah, you have two possible orders
The regular one and the backwards one
They'd both define the same topology
 
that isn't even what I mean
I mean that you can't discern any order from the topology
 
So the order relation gives rise to the natural topology
 
I'm off to bed
Night everyone
 
but you don't obtain any order just by looking at the topology
so you don't "lose" the order when you move to a different topology
 
12:27 AM
Oh, I see what you're saying now.
 
@orbit-stabilizer To define the order topology, you define $(a,b)$ to be $\{x:a<x<b\}$, and define $(-\infty,b)$ and $(a,\infty)$ similarly, and then call those "open intervals",
 
@LeakyNun So... we can still have order on R with a different topology?
 
and then say a set is open if for every point in it you can find an open interval around it that is in the set.
 
@orbit-stabilizer the topology does not determine the order on R
 
@AkivaWeinberger right, and like you said, it'd work backwards as well
@LeakyNun that's not what I am saying.
I'm asking if there's any conflict between the two.
 
12:29 AM
there is no conflict
 
@orbit-stabilizer I guess, but then that new topology won't be the order topology
 
just don't expect $(1,2)$ to be open
 
Incidentally, there's also the "standard topology" for $\Bbb R^n$, which essentially replaces open intervals with open balls
 
Okay, so geometrically then. Is it still correct to view that topology as 1 being isolated and every other point being infinitely close to one another
wait, shouldn't it be infinitely far away?
 
Yeah, essentially, that's the best way to view it
Nah, it just needs to be an isolated point
 
12:30 AM
{1} is dense
 
No, the other points, shouldn't they be far away from each other
 
lol
 
@LeakyNun Oh, wait
You might be right
Dammit
 
every point is close to 1 (informally)
 
12:31 AM
But 1 is isolated from the other points as well
I don't know how to visualize this
 
btw why 1, why not 0
 
I dunno
You might have noticed that the definition of topology is a little... permissive @orbit-stabilizer
 
How is it dense? My definition of dense is that every point of the superset is a point in the subset or is a limit point of the subset.
 
@orbit-stabilizer I don't try to visualize open sets
 
g
 
12:32 AM
@orbit-stabilizer yes, that's right
 
@orbit-stabilizer A set is dense if every nonempty open set intersects it
 
Okay, using that definition then.. How is $\{1\}$ dense?
 
(Limits can be defined in terms of open sets)
 
@orbit-stabilizer define limit point
 
@AkivaWeinberger I see it using that definitoin
 
12:33 AM
@orbit-stabilizer This happens to be what's called a non-Hausdorff topology, and
it turns out, in those, limits of sequences are not unique.
 
@AkivaWeinberger no, don't do limit points via limits
 
@LeakyNun x is a limit point of X if every open set containing x has a point not equal to x in X.
 
In fact, the limit of the sequence $(1,1,1,\dots)$ in this topology is every point in $\Bbb R$
 
@orbit-stabilizer give me an open set containing 2
 
$\mathbb{R}$
 
12:34 AM
@orbit-stabilizer Mhm. And thus every real is a limit point of $\{1\}$ in that topology
 
@orbit-stabilizer does it contain 1?
 
exactly
 
very interesting.
 
@orbit-stabilizer Here's something a bit more sane
Say you have $\Bbb R$ in the usual topology
 
12:35 AM
Rudin talks about neighbourhoods instead of open sets sometimes. Though they're the same thing I assume
 
Define an equivalence relation by $a\sim b$ iff $a-b$ is an integer. Essentially, it's equivalence "mod 1".
It turns out that there's something called a "quotient topology"
 
to help your visualization: a point which is dense is called a "generic point"
 
Ooh, we did this in group theory, we defined a group like this
 
We can "quotient" $\Bbb R$ by $\sim$ to get a new space. Can you guess what that space would look like?
 
Yeah!
It's a coset space
 
12:36 AM
there's also quotienting by group action of $\Bbb Z$ on $\Bbb R$, which is a bit less arbitrary than doing the equivalence relation
 
But geometrically @orbit-stabilizer
 
$[0,1]$ basically
 
@orbit-stabilizer no
 
@orbit-stabilizer Not quite, since $0\sim 1$
You need to "glue" those together
 
$[0,1)$
 
12:37 AM
no
 
@orbit-stabilizer Hint: This is a subset of the plane
 
the resulting space is not homeomorphic to $[0,1)$
 
not of the line
 
Hmm, We want it to wrap around? Why is that?
 
because you need to glue 0 and 1
 
12:38 AM
The circle?
Why?
 
or, imagine you're a point travelling on $\Bbb R$
 
you get wrapped back every time you reach an integer
 
Because in $\Bbb R$, the sequence $(0.9,0.99,0.999,\dots)$ approaches $1$, right? @orbit-stabilizer
 
12:38 AM
Oh, I see. We want it to be continuous?
 
So in $\Bbb R/\sim$, the sequence $([0.9],[0.99],\dots)$ should approach $[1]=[0]$
 
or, imagine having a point at every integer, and viewing the set of point as a single object that moves together
everytime you move right by 1 unit, you're essentially back to where you started with (there is no discernible difference)
 
So we get a circle. We can define a homeomorphism from $\Bbb R/\sim$ to the unit circle by sending $[x]$ to $(\cos(2\pi x),\sin(2\pi x))$
(note that this only depends on the equivalence class on $x$)
 
So, $\mathbb{R}/ \sim \cong mathbb{R} \ (2\pi \mathbb{R})$
 
12:40 AM
Similarly, if we take $[0,1]$ and quotient it by the relation $0\sim1$ we also get a circle
@orbit-stabilizer \sim
 
I give up
isomorphic to r mod 2pir
 
Group theoretically, it'd be mod $2\pi\Bbb Z$, I guess
but also just mod $\Bbb Z$
 
Hmmm. You're right.
2pi would make it a unit circle though, right?
 
Sure. But in topology, we don't care about size
 
@orbit-stabilizer do you wanna know how you quotient a topological space by a group action?
 
12:42 AM
All circles are homeomorphic
 
So, size doesn't matter?
 
(that's related to your name also)
 
Yeah. All circles are homeomorphic, and we only care about things up to homeomorphism.
 
@LeakyNun sure
It was a joke :P
 
even $(0,1)$ is homeomorphic to $\Bbb R$
 
12:42 AM
But, yeah gotcha
 
In fact, a circle is homeomorphic to (the boundary of a) square
 
though the former is obviously much "smaller" than $\Bbb R$
 
We don't care about corners
Now, here's something weird: Take $[0,1]$, and partition it into $\{0\}$ and $(0,1]$.
And consider the equivalence relation whose equivalence classes are just those two sets
and quotient $[0,1]$ by it. Essentially, we're collapsing $(0,1]$ down to a point.
We're left with a space with just two points.
 
@orbit-stabilizer Let $(X,\tau)$ be a topology. Let $G$ be a group that acts on $X$. Then, define the equivalence relation $x \sim y \equiv ([x]=[y])$ where $[x]$ denotes the orbit of $x$. Then, the quotient space is $X/\sim$
 
12:45 AM
This is the SierpiƄski space.
 
so $\Bbb Z$ acts on $\Bbb R$ by $(n,r) \mapsto n+r$
 
@AkivaWeinberger cool! I've heard that name before
 
Essentially, you end up with $X=\{0,1\}$ where the open sets are just $\emptyset$, $\{1\}$, and $\{0,1\}$.
 
@LeakyNun so we're looking at the partition that arises naturally from group actions
 
yes
 
12:46 AM
Just like how, in $[0,1]$, the closure of $(0,1]$ is $[0,1]$... in the SierpiƄski space, the closure of $\{1\}$ is $\{0,1\}$ (the entire space).
 
@orbit-stabilizer exercise: what do you get if you quotient your topology $(\Bbb R, \{\{\}, \Bbb R, \{1\}\})$ by $\Bbb Z$?
 
In $[0,1]$, if we have a sequence $(a_0,a_1,a_2,\dots)$ such that all the $a_i$ are in $(0,1]$, we don't know where the limit will end up; it could be in $(0,1]$ and it could be in $\{0\}$.
 
@LeakyNun i don't get how that example is related to topology, seems completely group-theoretic. We look at quotient groups in group theory too
 
In the SierpiƄski space, the constant sequence $(1,1,1,\dots)$ will have two limits, $0$ and $1$.
 
@AkivaWeinberger that is really weird.
non-unique limits
 
12:48 AM
hey
i finished my integral demo thing
 
If we define the function $f:[0,1]\to S$ ($S$ being the SierpiƄski space) by $f(0)=0$ and $f($anything else${})=1$, then that $f$ is continuous.
 
does anyone wanna see it
 
@AkivaWeinberger did you learn this from Munkres?
 
@MeowMix sure
 
12:49 AM
click partition to add an element to the partitioning set
 
@AkivaWeinberger oh, where then?
 
as you partition more, the integrable function's error will approach 0
 
@orbit-stabilizer let $S^3 := \{(z,w) \in \Bbb C^2 ~:~ \|z\|^2 + \|w\|^2 = 1\}$. Then, $S^1 := \{z \in \Bbb C ~:~ \|z\| = 1\}$ acts on $S^3$ via $(\lambda,(z,w)) \mapsto (\lambda z, \lambda w)$.
 
$S^3$ isn't really a group, which makes this notion of quotienting a topological space by a group useful
 
12:50 AM
@MeowMix you hardcoded the date?
 
$S^3$ is the 4D equivalent of a sphere
@MeowMix Did you solve the tetrahedron thing?
 
ofc
 
$S^3$ is the 3-dimensional sphere sitting in $\Bbb C^2 \cong \Bbb R^4$
 
fuck i forgot, sorry Akiva
 
@AkivaWeinberger that does look like a nice little book!
 
12:52 AM
Well, in the same sense that a regular sphere is intrinsically 2-dimensional, $S^3$ is 3-dimensional. But it sits inside 4-space.
 
bye
 
have fun exploring topology @orbit-stabilizer and don't forget my exercise
 
S^3 is totally a group.
 
@LeakyNun bye! I'll make sure to look
 
12:53 AM
Right, it's SU(3), right?
Or however you call it
 
SU(2)
 
Okay, wow that's enough topology right now for me. Back to $\mathbb{C}$.
 
Well it's the universal cover of SO(3)
and the unit quaternions, which is probably what you were getting at
 
If I have $|z-z_0| < c$, then is $|z_0 - \frac{z_0}{c}|$ a lower bound for $|z|$?
 
12:55 AM
You have $c-z_0\le z\le c+z_0$
(as well as, equivalently $c-z\le z_0\le c+z$. It's nice how it's so symmetric)
 
You mean the modulus of z?
 
Nah I meant $z$
 
How are you getting order for a complex number?
 
Oh, these are complex. Sorry.
 
Yeah, I'm trying to find a lower bound for $|z|$ for an $\epsilon - \delta$ proof
 
12:57 AM
In any case, no. Take $c=100$ and $z_0=1$
Then $z=0$ satisfies that, but $|z_0-\frac{z_0}c|=0.99$
 
You're right
So it should be: $|z_0 - c \cdot z_0|$
 
Is it true that $\int p(x) f(x) \le \int f(x)$ where $f(x) > 0 $ for all $x$ and $p(x)$ is a probability density so $\int p(x)= 1$?
 
Yeah, since $p(x)f(x)\le f(x)$ for all $x$ @Monolite
 
but p(x) can be greater than one
 
I thought you said it was a probability density
 
12:59 AM
Uh what.
 
Aren't those strictly between $0$ and $1$?
 
only in the discrete case
 

« first day (2728 days earlier)      last day (2299 days later) »