« first day (4501 days earlier)      last day (519 days later) » 
00:00 - 21:0021:00 - 00:00

9:06 PM
okay here's a complicated proof: Let $G$ be the group of orientation-preserving symmetries of a regular tetrahedron. The action of $G$ on the set of vertices gives an isomorphism $G \cong A_4$.
The vertices of the regular tetrahedron can be taken to be $(1,1,1),(1,-1,-1),(-1,1,-1),(-1,-1,1)$ this implies that the natural inclusion $G \to SO(3)$ has image contained in $\mathrm{SL}_2(\Bbb Z) \cap SO(3)$. Because they are orthogonal, they preserve the quadratic form $x^2+y^2+z^2$. Reducing mod $3$, we obtain matrices in $\mathrm{SL}_2(\Bbb F_3)$ preserving the quadratic form $x^2+y^2+z^2$, thu
not even a complete proof, just an idea :P
 
@LukasHeger did you check my question ?
 
which one
 
how to prove that the closed interval [a,b] in R is connected?
without referring to topology
I can use the completness axiom of R
and this theorem that says , X a subset of R is connected then for any element z in X in such that a<z<b , then z has to be in X
=)
it seems very trivial but i cant find a way to do this without assuming an interval has to contain no holes
but if i use that then there is nothing to prove
also in my arguments when i tried to prove it i assumed that a<z<b and z is not in (a,b)
but even when i do that, it might that other "holes" in the interval
so i cant even say things like (a,z) is subset of (a,b)
 
surely you can say that (a,z) is a subset of (a,b) if a<z<b
just from the definitions
 
what does connectedness mean without topology
 
9:11 PM
Jack, the standard topology proof depends on the lub axiom.
 
oh right Lukas , i was thinking of something strange
 
Yeah, Thor said what I was going to.
 
I mean the proof should not use techniques from topology
i seen many proof that uses open sets and topological arguments like continuity and things of such
but my idea was to get a contradiction
(a,z) will be bounded so has a SUP
 
I don't think you need continuity
but how do you define connectedness without open sets?
 
that sup call it s , that has to be in the close of (a,z)
 
9:13 PM
what is z?
 
@LukasHeger I like this argument, but Shaun doesn’t know, I presume, that smooth conics are rational.
 
simply as union of seperated sets , non empty
a set is connected if it cannot be written as union of non empty separated sets
 
@Ted I stole the proof idea from this answer: https://math.stackexchange.com/a/2141862/348926
which proves a more nontrivial isomorphism
 
there is no mention of open
a set is connected if it cannot be written as union of non empty separated sets
 
what does separated sets mean?
 
9:14 PM
it means that cl(A) intersection B is empty
and CL (B) intersection with A is also empty
eg (1,2) (2,3) in R
z is a point that satisfies a<z<b and such that z is not in [a,b]
ofcourse this is wrong and should lead to contradiction
 
such a point does not exist...
but that doesn't prove anything
you need to work with the sets A and B to get a useful contradiction
 
a theorem that we have is , a subset of R is connected if a<z<b , then z has to be in the subset
that is why i used this approach
to arrive at contradiction
 
but if you have that theorem, then the result is immediate
 
writing cl(A) refers to topology
whether you phrase this is in terms of closure or open sets does not make a difference conceptually
 
how Lukas?
We are using Rudins book in this class , there is some topology
but it is not that much !
I mean if we assume that intervals has no holes
then it is immediate from that theorem
but I assumed that is what we needed to show in this question
 
9:19 PM
hint: dont use Rudin
 
what is your definition of the closed inteval?
 
haha too late!
 
Rudin has a ton of topology for a beginning analysis course.
 
@TedShifrin Hello. I have been working on Exercise 2.3.1 to 2.3.15 and I am a bit unsure about my solutions to Exercises 2.3.11 and 2.3.13. Should you have the time and inclination to take a look at them, I would appreciate some feedback. My solutions follow.
E.2.3.11. Give an example of a discontinuous function $f:\mathbb{R}\to\mathbb{R}$ having the property that for every $c\in\mathbb{R}$, the level set $f^{-1}(\{c\})$ is closed.
Solution. $f:\mathbb{R}\to\mathbb{R}, f(x)=\mathbbm{1}_{\mathbb{Q}}+x$.
 
@LukasHeger that is exactly what confused me as well Lukas , i am checking the definiton now
 
9:21 PM
E.2.3.13. Prove that if $\mathbf{f}$ is continuous, then the preimage of every closed set is closed.

Proof. Let $C\subset\mathbb{R}^m$ be a closed set, $f:\mathbb{R}^n\to\mathbb{R}^m$ a continuous function and suppose that $f^{-1}(C)$ is not closed. Then $\mathbb{R}^n\setminus f^{-1}(C)$ is not open so there must be some $\mathbf{x}_0\in\mathbb{R}^n\setminus f^{-1}(C)$ such that $B(\mathbf{x}_0,r)\cap \mathbb{R}^n\setminus (\mathbb{R}^n\setminus f^{-1}(C))=B(\mathbf{x}_0,r)\cap f^{-1}(C)\neq\emptyset$ for every $r>0$.
 
it should be [a,b] as in element x in R such that a<= x <= b
 
okay now try and apply the theorem with that definition
 
could it be that simple?
 
yes if you have the theorem it's very direct
 
just saying that a<z<b then z satisfies the criteria of being an element of [a,b]?
 
9:23 PM
why 0 and 1?
and I think you need to take all a',b' in [a,b] and all z such that a'<z<b'
 
was thinking of unit I
 
Your counterexample is discontinuous everywhere, no? There are far simpler examples (like ones we can graph easily), but it seems fine.
 
the theorem is for all a and b in a set, right?
 
yes
 
so you can't just take only the endpoints
 
9:25 PM
no I would assume not
if we are going to use the theorem directly it does follow
I dont see though why this question was asked
but if we use my first approach
would that get us somewhere?
 
what was your first approach?
 
For 13 can’t you do a direct proof?
 
if we proceed by contradiction , assume that a<z<b and such that z is not in [a,b]
what contradiction can we arrive at?
 
that doesn't lead anywhere
that just proves that all z with a<z<b are in [a,b]
 
cant it lead to sup of (a,z) cannot be in [a,b]?
 
9:27 PM
but that's true by the definition of [a,b]
 
you are right
I was thinking that against all odds there exist some magical number that is not in R
 
And where in the universe would that be?
 
that is somehow not in that interval and has the property that is larger than a and less than b
 
a complex one I pressume
 
to arrive at maybe the supremum axiom is false?
 
9:29 PM
With ordering?
 
there is a direct prove using the supremum axiom, but that's not it
 
what is it?
because i litterly spend 2 days on this now
but it was useful
since I did read more in the book and redone the theorems
 
@TedShifrin Thanks. Regarding E.2.3.11 I know that the function is discontinuous everywhere but I proved that it is discontinuous at $0$ to show that it satisfies the requirements in the text of the Exercise (to be discontinuous at least at one point). For E.2.3.13 I am attempting a direct proof right now.
 
if you have two separated set A and B whose union are [a,b], then you can take points a in A, b in B such that (wlog) 0<a<b<1 and then consider the sup of the set {x in A | x<b}
that leads to a contradiction
 
okay thank you !
I did try something similar but then somehow in the middle
i ask my self, if we are allowed to assume interval properties
why are we doing this
haha
@LukasHeger i dotn think you meant to write 0 and 1 there
but the idea is clear
 
9:34 PM
I mean yeah the endpoints
 
I meant there are some x in A, y in B such that a<x<y<b
 
Yes!
but that would simply breaks down
to the defintion of interval
 
what's your definition of closure, by the way?
 
it is the set union its limit points
there is some similarity with topology courses
but we cant use baire theorem
or basis for topology
and things like that
 
9:38 PM
I would be surprised if the baire category theorem is useful for doing stuff in rudin
but maybe not that surprised
 
haha but the funny thing is
that connectedness of [a,b]
also follows from its convexity
since convex subsets of R^k are connected
 
are you sure that's not circular?
how do you show that convex sets are connected without using that intervals are connected?
 
that is a good question !
a missspoke
i think the only proper way to do this is what you wrote last
is to do it via the supremum axiom
that bbd subsets of R has sup , since that is the only axiom we have on R
the rest are theorems
 
@LukasHeger a baby case is an exercise in the topology chapter.
 
I don't know Rudin's book. It's just that I think of applications of Baire as more of a functional analysis thing
presumably that's wrong and there are also real analysis applications
 
9:47 PM
@TedShifrin Direct proof of Exercise 2.3.13.
Proof. Let $C\subset\mathbb{R}^m$ be a closed set, $f:\mathbb{R}^n\to\mathbb{R}^m$ a continuous function and $\{\mathbf{x}_k\}_{k\in\mathbb{N}}$ a sequence of elements in $\mathbf{f}^{-1}(C)$ which converges to some $\mathbf{a}\in\mathbb{R}^n$.
Then $\{\mathbf{f}(\mathbf{x}_k)\}_{k\in\mathbb{N}}$ is a sequence of elements in $C$ and since $\mathbf{f}$ is a continuous function we have $\mathbf{f}(\mathbf{x}_k)\xrightarrow{k\to\infty}\mathbf{f}(\mathbf{a})$ by Proposition (3.6). Now $\mathbf{f}(\mathbf{a})$ belongs to $C$ since it is
 
Right. Or even use Prop. 2.1 :)
@LukasHeger Like how discontinuous can the pointwise limit of a sequence of continuous functions be?
 
Thanks
 
@TedShifrin right, you get continuous on a dense $G_{\delta}$ set I think
comeager even
 
10:03 PM
i think you could also use to to show that something is not connected
as in the proof of Q with the usual toplogy
and some other applications
Rudin has some exercices on Baire category theorem
which I skipped!
 
10:18 PM
the good applications of Baire are "embeddings are dense among all maps X -> Y if <constraint on dimension>" type results
 
10:54 PM
@LukasHeger Right, the discontinuities form a set of first category. Corollary: If $f$ is differentiable, the discontinuities of $f'$ is a set of first category. See — basic analysis :)
 
11:04 PM
Hi! Is 0.(9) different from 1 in the hyperreals?
 
11:21 PM
No.
 
@Ted Is that because the reals are included in the hyperreals? Or are they completely different constructions anyway?
 
Yes, the standard reals form a subset.
 
11:42 PM
I see, thank you!
 
00:00 - 21:0021:00 - 00:00

« first day (4501 days earlier)      last day (519 days later) »