« first day (1941 days earlier)      last day (3078 days later) » 
00:00 - 17:0017:00 - 00:00

5:04 PM
How do you best prepare for the harder things
 
Are rational exponents defined axiomatically for R? Don't think they're defined for Groups. o:
Never thought about it before. lol
 
For positive integers n and m and reals a, there is a unique solution x>0 to x^m=a^n, and we call it a^(n/m). then you can define negative fractional exponents with positive bases. defining fractional exponents with negative bases requires more care.
eventually you can define z^w for complex numbers z and w using the exponential function and picking a branch for the logarithm function
 
once you have rational exponents, you can define it for real exponents using limits as Q is dense in R. then you can analytically extend to C as anon said.
 
user174558
5:29 PM
I actually like Dummit and Foote except for the ring definition, sigh, so I cannot use it.
 
user174558
You know those calculus books have two versions, one with early transcendentals.
 
user174558
I was thinking they should have two versions for algebra books, with different ring definitions, lol.
 
user174558
Bjorn Poonen wrote a very nice article about why rings should have unity.
 
How do you prepare for the harder things?
 
@tired I will look at your answer. I wrote an answer using a simpler auxiliary function.
 
user174558
5:32 PM
Oh well, since Mike is ignoring me, I won't talk to him ever again.
 
user174558
@OverlyExcessive Very vague question.
 
@JasperLoy What I mean is like, algebra and trigonometry.
 
user174558
@OverlyExcessive How to prepare for algebra and trigonometry, you mean that?
 
@JasperLoy Yes I mean that.
 
user174558
@OverlyExcessive Oh I thought you did that already?
 
user174558
5:35 PM
@OverlyExcessive Well, I guess you should look at some high school textbooks you have access to.
 
@JasperLoy I am doing Khan Academy, grade 2 through 7 and now I have started doing some basic algebra but I find it a bit hard.
 
user174558
@OverlyExcessive I don't know anything about Khan Academy, but I am pretty sceptical about such online things generally.
 
user174558
@OverlyExcessive Could you get some nice cheap high school texts in your area? I think that is what you need...
 
@JasperLoy Well, okay what I have covered is basically, fractions, arithmetic, and basic geometry (properties of shapes, area of squares/circles, volumes). 2-step equations, and probably some other things I've forgot to mention.
 
user174558
Ooh, I defecated thrice today, now I feel so good.
 
user174558
5:38 PM
@OverlyExcessive Well it's really hard to know what you have exactly covered just from the description of it. But like I said get some nice cheap books.
 
user174558
@OverlyExcessive For example, you can get the books for O level and A level math from Cambridge University Press.
 
Well I mean high school math, it's based on the American common core
 
user174558
@OverlyExcessive Sorry, I am not American, and even if I were, it is hard to figure out exactly your standard.
 
They are a set of standards that are used nationwide in math education in the U.S I believe.
 
user174558
The books for the IB exam will also be good.
 
5:40 PM
Isn't there something equivalent for the commonwealth?
 
user174558
Yes, A level and IB is what you take at the end of high school, lol.
 
user174558
That is what I have been talking about.
 
user174558
Look under education.cambridge.org for the books, tons of quality books there for school.
 
user174558
You can filter by level and by subject and by exam.
 
I'm gonna look at that, thanks mate
 
user174558
5:43 PM
@robjohn Nothing won at the lucky draw, for two years in a row. =(
 
Is that a good prep for calculus?
 
user174558
@OverlyExcessive The books there will cover algebra, trigonometry and even some calculus. Take your pick and take your time to look.
 
@JasperLoy well, if the odds are less than 50%, I am not really surprised.
 
user174558
@robjohn Well, one in five people got a prize.
 
@JasperLoy so there is a $64\%$ chance that you wouldn't get a prize in two times.
 
user174558
5:47 PM
@robjohn Well well, no need to be so mathematical in daily life, lol
 
@JasperLoy Hey, it's math chat...
 
user174558
@OverlyExcessive You can download Marden's Calculus I, II, III legally if you do a search.
 
user174558
@OverlyExcessive You can also download Kaplan's Calculus and Linear Algebra I, II legally if you do a search.
 
@JasperLoy Thanks! Would those be a good start for me? I've done up to and including K-7 like I said, common core.
 
user174558
5:49 PM
@OverlyExcessive You can also buy the above 5 books on amazon at low prices.
 
www.corestandards.org/Math/
 
user174558
@OverlyExcessive I am not familiar with the American standards, but of course do your algebra and trigonometry first. Those books I mentioned review them at the start.
 
@JasperLoy I linked an overview of each of the grade and the subjects covered in the common core.
 
user174558
@OverlyExcessive The books I recommended are thorough but not overwhelming.
 
user174558
@OverlyExcessive Anyone who has done algebra and trigonometry can use the books I mentioned, really.
 
5:52 PM
I...
Have to start with those two then because I have only done pre-algebra and geometry no trig yet...
 
user174558
@OverlyExcessive Yes, so do the algebra and the trigonometry first, some of which is covered in Lang's Basic Mathematics, I believe.
 
user174558
@OverlyExcessive I think Lang's book will be good for you.
 
@JasperLoy Yes Lang's book starts off with algebra actually..
Yes, I think you are right
I think I can do this
 
user174558
@OverlyExcessive Yes, the orange is always right.
 
I must believe, and be persistent, I am spending about 2½ - 3hrs a day practicing, I think that is adequate.
I divide the time between exercises and reading.
 
user174558
5:55 PM
@OverlyExcessive It's more important to understand the concept than to practise.
 
Does practice not make perfect?
 
user174558
@OverlyExcessive Once you understand, you don't need much practice.
 
user174558
@OverlyExcessive Not in all areas of knowledge. For high school math, perfect understanding requires no practice, since all problems are routine.
 
the concepts are so advanced
 
user174558
Oh, anyone who disagrees I won't bother arguing with you.
 
user174558
5:57 PM
@OverlyExcessive They are advanced only because you are not used to them, or had a lousy math teacher your whole life who could not let you see the light.
 
user174558
@OverlyExcessive Some of these math teachers don't even understand the concepts themselves.
 
user174558
@OverlyExcessive Hey, if you find Lang's book too advanced, the Cambridge books go all the way from grade 1 to grade 12, take your pick like I said.
 
user174558
@OverlyExcessive I recommend them because it is the only set I know that goes from 1 to 12.
 
user174558
@OverlyExcessive Also, try the site libgen.io for pdf and djvu copies of books hosted on servers in locations where copyright may not apply.
 
user174558
@OverlyExcessive I am going to bed. Good night, and good luck.
 
6:04 PM
@anon: why are reps of compact groups determined by their characters?
 
Goodnight, @MikeM ... no one got food poisoning?
Hi @anon: Thanks for your interesting group example.
 
not yet
 
hi @TedShifrin!
 
oh, @Balarka has returned as a physicist :)
 
hardly.
 
6:15 PM
No?
 
I attended a couple physics lectures, but nothing which made me think physics is more exciting than math :)
The most interesting lecture was about K-theory and condensed matter physics.
 
Of course, most of my physicist friends don't consider all this mathematical physics stuff physics, anyhow :)
 
Really?
 
@MikeM: I believe the answer is that Schur or something tells you that the irreducible representations of a compact group are orthogonal.
@Balarka, yes.
 
@Ted: Today my reps are even special unitary
 
6:21 PM
no, no, I mistyped. The characters of irreducible reps are orthogonal with respect to the natural inner product.
 
Oh... I guess I should think about the proof again and see why it goes through for compact groups.
 
Plus the theorem that you can decompose any representation as a sum of irreducibles.
 
@TedShifrin Interesting. What do they consider to be physics, then?
 
The stuff that actually has any predictive power about things that happen in testable ranges
 
I am not qualified to answer this, @Balarka. Get back to stuff I know :)
 
6:22 PM
@Ted: Side note. Do you know how many conjugacy classes of elts there are in SU(n)?
 
Surely infinitely many?
 
Oh... I guess that's clear by diagonalizing.
I must be misunderstanding something, as usual :)
 
If you're thinking about characters and conjugacy classes, the 1-1 correspondence may only work in finite groups. I am rusty ...
 
It seems to be cited as a fact.
 
But there can be infinitely many for infinite groups?
 
6:28 PM
@TedShifrin haha, ok. I have made a sin. During the time I was in the seminar, prof taught and made me do exercises from ch 1 Guillemin-Pollack. Of course, I had to take the inv. func. thm as granted all the time. But I am getting back to calc right now. Should I start on chapter 6 after learning Lagrange multipliers and revising Gram-Schmidt?
 
The key should be the compactness of the domain.
Lest all of representation theory go bust.
 
@Balarka: Since you know about annihilators, there's a better way to look at Lagrange multipliers.
 
Annihilators as in module theory?
 
or linear algebra, yes
 
I am curious. Can you tell me/give me a reference?
 
6:33 PM
Well, it's the same proof I did in the text with perps, but really annihilator is more natural. The point is that if $Dg_1(a),\dots, Dg_m(a)$ are linearly independent (linear maps), and f has a critical point at $a$ on $g=0$, then $Df(a)$ annhilates precisely the same vectors that all the $Dg_i(a)$ do, and so it follows that $Df(a)=\sum \lambda_i Dg_i(a)$ for some scalars $\lambda_i$. Prove it.
 
after knowing it exists with some work you could probably devise the right formulation as yourself
or that
 
There's also a lie in there, I think.
 
r9m
@DanielFischer is there a way to get the functional equation of $\displaystyle F(z) = \int_0^1 \frac{\log^2 (1-x)}{z - x}\,dx$, defined on $z \in \mathbb{C}\setminus [0,\infty)$?
 
Sorry, was away. Thanks, let me read it.
 
@Balarka: The word "precisely" is clearly wrong. Fix it. :)
 
r9m
6:49 PM
BBL
 
7:07 PM
I've just been thinking... Let $S$ be a set. Given a class of elements, such that each element of this class is also in $S$, is this class necessarily a subset of $S$ ?
 
@Jake1234 Unless I am misunderstanding something, it looks so.
 
I'm not entirely sure, how would show this?
 
@r9m Pretty sure that there is a way, if the function has a nontrivial functional equation. Do you happen to know what the functional equation is? That might help proving it.
 
Hi!!! I have a question... Can a closed set contain an open set?
 
@evinda Sure. Even nonempty ones. $[0,1]$ contains the open sets $(0,1),\, (1/3,1/2)$ and a lot of others.
 
7:23 PM
@DanielFischer Ah, I see... If $X$ is a metric space and $K \subset X$ then with $\overline{K}$ we symbolize the smallest closed set that contains K.

So for example if we consider (0,1) as K then $\overline{K}=[0,1]$, right?
 
@evinda Right.
 
@DanielF: If I perturb an operator by a compact one, that preserves invertibility, but it should otherwise change the spectrum, yes?
Oh no, that's not even true.
 
@MikeMiller I was going to ask whether "preserves invertibility" was an extra assumption.
 
@DanielFischer Nice... Thank you!!!
 
@r9m I get $-2\operatorname{Li}_3\left(\frac1{1-z}\right)$
 
7:29 PM
@DanielFischer Could I also ask you something else?

We have that $l^2(\mathbb{N})=\{ (x_n) \in \mathbb{R}^{\mathbb{N}}: \sum_{n=1}^{\infty} |x_n|^2< +\infty\}$.

I want to find in $l^2(\mathbb{N})$ a subspace $Y$ and a $x \in l^2(\mathbb{N})$ such that $d(x,Y)$ is not attained.

Could we pick $Y=\{ x \in l^2(\mathbb{N}) | \exists n \in \mathbb{N} \text{ such that } x_j=0, \forall j>n \}$ ?
 
@DanielFischer: what if I said it was a really, really compact operator?
 
@MikeMiller Even if the range is one-dimensional you can lose invertibility. Smallness of norm of the perturbation however preserves invertibility even if the perturbation is non-compact.
 
Using the expansion
$$
\begin{align}
\frac1{z-x}
&=\frac1{z-1}\frac1{1+(1-x)/(z-1)}\\
&=\frac1{z-1}-\frac{1-x}{(z-1)^2}+\frac{(1-x)^2}{(z-1)^3}+\dots
\end{align}
$$
we get
$$
\begin{align}
\int_0^1\frac{\log^2(1-x)}{z-x}\,\mathrm{d}x
&=\int_0^1\frac{\log^2(1-x)}{z-x}\,\mathrm{d}x\\
&=\sum_{k=0}^\infty\frac{(-1)^k}{(z-1)^{k+1}}\int_0^1\log^2(1-x)(1-x)^k\,\mathrm{d}x\\
&=\sum_{k=0}^\infty\frac{(-1)^k}{(z-1)^{k+1}}\int_0^1\log^2(x)x^k\,\mathrm{d}x\\
&=\sum_{k=0}^\infty\frac{(-1)^k}{(z-1)^{k+1}}\int_0^\infty t^2e^{-(k+1)t}\,\mathrm{d}t\\
 
@evinda Yes, we could. And since we're in a Hilbert space, it must be something along those lines (you need a non-closed subspace).
 
@DanielFischer In order to show that this Y is not closed , we pick a sequence $\overline{x_m}$ in Y such that $\overline{x_m} \to x$ and we want to show that $x \notin Y$.
$\overline{x_m}$ will be of the form $(x_{m1}, x_{m2}, \dots, x_{mn},0,0, \dots,0)$ and we will have that $\sum_{j=1}^{\infty} |x_{mj}|^2< +\infty$.
But then don't we have that $|x_{mi}-x| \leq ||x_m-x||_2= \left( \sum_{j=1}^n |x_{mj}-x_j|^2\right)^{\frac{1}{2}} \to 0$ ?
 
7:42 PM
Okay so I need a kind soul to explan this to me.. $\sqrt{50} - \sqrt{32} \ne 1$. I don't understand why I'm wrong. I've simplified the expression and I get $ 5(\sqrt{2}) - 4(\sqrt{2})$ apparently the answer is $1(\sqrt{2})$.
Wouldn't $\sqrt{2} - \sqrt{2} = 0$ ?
 
@evinda I don't understand what you're trying to do there. For that particular $Y$, you should know that it is dense, and given any $x\in l^2$, there's a pretty obvious sequence in $Y$ converging to $x$.
@OverlyExcessive You multiply. $a\cdot x - b\cdot x = (a-b)\cdot x$. That $x - x = 0$ plays no role.
 
@DanielFischer I tried to prove that Y is not closed. Doesn't this hold?
 
@DanielFischer Please, can you explain why? Even though $\sqrt{2}$ is irrational if it is subtracted from itself it should still be 0? No?
 
@evinda But how do you try to show it? You need to show that there is some $x$ that belongs to $\overline{Y}$ but not to $Y$.
@OverlyExcessive But you never subtract it from itself here. You have $5\cdot \sqrt{2} - 4\cdot\sqrt{2}$.
 
@DanielFischer Right I see it now, thanks.
 
7:47 PM
You subtract the factors of $\sqrt{2}$.
 
@DanielFischer How can we find such a x?
 
@DanielFischer Is this possible because the order of evauluation doesn't matter with multiplication?
 
@evinda Look at the definition of $Y$. That tells you when an $x$ does not belong to $Y$. Since $Y$ is dense, in this example that is everything you need.
@OverlyExcessive I don't see how order of evaluation could have anything to do with it. It's the distributivity of multiplication over addition, $x\cdot (y+z) = x\cdot y + x\cdot z$.
 
@DanielFischer Ah yes, thanks again.
 
@DanielF: I suspect the perturbation is a contraction, but I should probably, you know, check.
 
7:54 PM
@MikeMiller Or you can make a heroic assumption ;)
 
"In fact, we may as well assume the perturbation is zero."
 
@DanielFischer Is the square root of $n$ equal to the square root of all prime factors of $n$ multiplied?
 
consider sqrt(4*3) versus sqrt(2)*sqrt(3). backtracking, even n itself is not the product of the prime factors of n, unless you count them "with multiplicity"
 
@OverlyExcessive If you take the proper powers. If $p^3\mid n$ (and $p^4 \nmid n$), then you must have $\sqrt{p}^3$ in the product.
 
@anon Is it not a fact that all composite numbers are the product of their prime factors?
@DanielFischer Sorry I don't understand, explain like I'm 12? :)
 
7:58 PM
@DanielFischer So we consider a sequence of the form $(x_{m1},x_{m2}, \dots, x_{mn},0, \dots,0)$ and want that $\sum_{j=1}^n x=+\infty$ ?
 
Why is it a bad idea to think of a 'divisor' as a polygon?
 
@OverlyExcessive does 12 equal 2 times 3? I mean, the prime factors of 12 are 2 and 3...
 
@anon No.. But 12 = 2^2 * 3
 
@bolbteppa why is it a good idea?
 
@anon because that is the historical origin and motivation for the idea
 
8:00 PM
@OverlyExcessive right. so is sqrt(12) equal to sqrt(2)sqrt(3)? no, it's sqrt(2^2)sqrt(3).
 
@anon But doesn't that mean that the square root of a number is equal to the product of all prime factors?
 
@bolbteppa I assume you mean representing it as a d-gon inside an n-gon? It feels unlikely that's the original motivation for the idea of one integer dividing into another, but sure you can do that, for positive naturals. won't work in general rings ofc.
@OverlyExcessive you mean the product of the square roots of its prime factors. do you think 4 is a prime factor of 12?
 
@anon Okay, you're right but that's semantics -- you know what I meant
 
if you mean 12 is the product of 2 and 3, which is what you're actually typing, then no. if you mean numbers are products of their prime power factors, then yes. and of course sqrt(ab)=sqrt(a)sqrt(b) generalizes to any number of things being multiplied together, it has nothing to do with prime numbers really.
 
@evinda Sorry, you lost me there. What is that supposed to accomplish? (And $\sum_{j = 1}^n x = +\infty$ can't be.)
 
8:03 PM
@anon sorry haha I mean in the algebraic geometry sense en.wikipedia.org/wiki/Divisor_%28algebraic_geometry%29
 
@anon I know that I was typing that but you know that I didn't MEAN that, which is different.
 
What do we define as $\overline{Y}$ in this case? @DanielFischer
 
@OverlyExcessive I didn't know that you didn't mean that
 
@anon What I said was that 12 is the product of all it's prime factors and this is still true, $2 * 2 * 3 = 12$. I don't know where you got that I said$ 2* 3 = 12$ from.
 
the product of 12's prime factors is 6, not 12. you have to say something to the effect of "counted with multiplicity" in order to mean what you want to mean.
 
8:07 PM
@evinda $\overline{Y}$ is the closure of $Y$. You picked $Y$ (as the subspace of sequences with only finitely many nonzero terms). That determines $\overline{Y}$.
 
@anon Okay, I understand, so it's a matter of semantics then.
 
@bolbteppa well, a polygon has vertices counted with multiplicity 0 or 1, so no <0 or >1, and it has edges which a divisor doesn't seem to have
 
I was just wondering if the same principle applied to radicals but I understand it does then.
 
@OverlyExcessive $$n = \prod_{k = 1}^m p_k^{a_k} \implies \sqrt{n} = \prod_{k = 1}^m \sqrt{p_k}^{a_k}$$
 
@OverlyExcessive "it's a matter of semantics" often implies that the intended meaning is something that should have been understood anyway. but when mathematicians refer to "the prime factors of a number," they almost always don't mean with multiplicity, and when they do, they actually say that.
@DanielFischer does overly know what \prod is?
 
8:10 PM
No.. I
 
@anon I don't know. Apparently not.
 
I asked him to explain like I was twelve, I suppose this is his idea of a joke ..
 
@OverlyExcessive No, sorry, just inattentiveness.
 
@anon how would someone think about the idea of a divisor if not as just a trick to talk about polygons from an algebraic function perspective?
 
@DanielFischer No worries, it's Friday after all :)
 
8:12 PM
@bolbteppa the number of minutes I've thought about them is probably less than an hour, but it feels like they are formal linear combinations of points meant to generalize the idea of rational functions being the only (whatevers) on the Riemann sphere
 
So the closure of Y contains the sequences the limit of which have only finitely many nonzero terms ? Or have I understood it wrong? @DanielFischer
 
@evinda No, the closure of $Y$ is all of $l^2(\mathbb{N})$. $Y$ is dense in $l^2(\mathbb{N})$.
 
Why do we deduce from the fact that $Y$ is dense in $l^2(\mathbb{N})$ that the closure of $Y$ is all of $l^2(\mathbb{N})$? Could you explain it further to me? @DanielFischer
 
do you know what dense means?
 
8:18 PM
wasn't a pun
 
@evinda It's the definition of "dense".
 
Haha it certainly seemed like one :P
 
@anon this is of course incoherent because the number of minutes you've thought about them is a unitless construct, and hence cannot be measured in hours
 
ofc
 
A set A is called dense if its closure is X. @anon
 
8:21 PM
Y is dense in l^2(N) means the closure of Y is all of l^2(N), right
 
@DanielFischer @anon But how do we deduce that Y is dense?
 
suppose $(x_1,x_2,\cdots)$ is any sequence in $\ell^2(\Bbb N)$. form a sequence of elements of $Y$ that converge to $x$. (note elements of $Y$ are of course themselves sequences)
 
We can only do this if $x_{n+1}=x_{n+2}= \dots=0$. Or am I wrong? @anon
 
you are wrong
 
Could you explain me why? @anon
 
8:26 PM
I am not going to do the problem for you.
Find a sequence of elements of Y that converge to (x_i).
 
The first elements could be also x_i, or not? @anon
 
you seem like you're on the right track. expand on your thoughts...
 
@anon I have thought the following:

If the sequence of $\ell^2(\Bbb N)$ is of the form $x=(x_1, x_2, \dots, x_n, 0, \dots,0)$ then the sequence of elements of $Y$ that converges to $x$ is exactly the same, x.

But if x is of the form $(x_1, x_2, \dots, x_n, x_{n+1}, \dots, x_m, \dots)$ (not all of the $x_i$ are equal to 0) then we cannot find a sequence of elements of $Y$ that converges to $x$.
 
false
stare at your first sentence long enough and you'll think of a sequence of elts in Y that converge to x
 
@anon Is every element of set a set in ZFC?
 
8:36 PM
you'd have to ask a set theorist that
iunno
 
hm
 
Can we take for example also x and let n tend to $+\infty$ at the case where not all x_j for j>n are zero? @anon
 
@Jake1234 In ZFC, there are only sets. There's also ZFC + Atoms, or ZFCU (Urelemente), in which things are considered that aren't sets.
 
@evinda right, (x1,...,x_n,0,0,...) tends to (x1,x2,...) as n->inf
(prove this)
 
If I were to prove that some "thing" that contains a sequence in $\mathbb{N}$ and nothing else is a set, and concretely a subset of $\mathbb{N}$, because it is an element of the power set, and thus a set?
*would that be right?
 
8:43 PM
wut
 
:D
I'm trying to see, whether everything that contains only elements from some set, is necessarily a subset in ZFC.
 
in what context do we prove "things" are sets? I am not a set-theorist, but I have not seen that kind of thing before. as opposed to, say, proving putative maps are well-defined in abstract algebra or topology.
what do you mean by "thing" or "everything"?
 
@anon So do we prove it as follows?
Let $\epsilon>0$. We want to show that $\exists n_0 \in \mathbb{N}$ such that $\forall n \geq n_0$:

$|(x_1, x_2, \dots, x_n,0,0, \dots, 0)-(x_1, x_2, \dots, x_{n}, x_{n+1}, \dots)|< \epsilon$.
 
well, you want to say $\displaystyle\lim_{n\to\infty}\sum_{k=n+1}^\infty x_n^2=0$ if $\displaystyle\sum_{k=1}^\infty x_k^2<\infty$. whether you feel that requires an epsilon-delta proof is up to you.
 
I know the question I'm writing doesn't really make sense. By thing, I mean something more general than a set. There's thing that can be defined, but they don't exist as ZFC sets, right?
I dunno, I'd maybe say "thing" could be a class, but I'm not sure whether it couldn't be more general than that.
Yeah I think class isn't general enough, there it follows immedietly... class of sets that has some property, and where all the elements are from some set, is a set - it's the set I get if I put that property constraint on the set.. I think.
 
9:02 PM
@anon We prove that $\lim_{n\to\infty}\sum_{k=n+1}^\infty x_n^2=0$ if $\sum_{k=1}^\infty x_k^2<\infty$ as follows, right?

For each $n,m$ with $m>n$ we have:

$\sum_{k=n+1}^m x_k^2= \sum_{k=1}^m x_k^2- \sum_{k=1}^n x_k^2$

We fix a $n$ and take the limit as $m \to +\infty$.

$\sum_{k=n+1}^{\infty} x_k^2= \sum_{k=1}^{\infty} x_k^2- \sum_{k=1}^n x_k^2$

Now we let $n \to +\infty$ and we have:

$\lim_{n \to +\infty}\sum_{k=n+1}^{\infty} x_k^2= \sum_{k=1}^{\infty} x_k^2- \sum_{k=1}^{+\infty} x_k^2=0$
 
yes
first half with m was unnecessary
just the last line with n->inf was fine
 
9:20 PM
@anon So do we say the following?


Let $x=(x_1, x_2, \dots, x_n, \dots) \in l^2(\mathbb{N})$.

So, we have that $\sum_{k=1}^{\infty} x_k^2<+\infty$. This implies that $\lim_{n \to +\infty} \sum_{k=n+1}^{\infty} x_k^2=0$.

So if we let $n \to +\infty$ then the element of Y, $(x_1, x_2, \dots, x_n, 0, \dots)$ tends to x.
Thus the set Y is dense.
 
yes
 
So in general if we want to prove that a subset of a set is dense we pick a sequence of the subset and want to prove that it tends to any sequence of the set, right? @anon
 
if you want to prove a subset of a space is dense you pick an element of the space and exhibit a sequence of elements of the subset that converges to it
 
A ok. So in order to show that Y is not closed we want to find a sequence $(\overline{x_m})$ of Y for which it holds that $\overline{x_m} \to x$ and there is no $n \in \mathbb{N}$ such that x_i=0, for all $i \geq n$ right? @anon
 
not sure what you're trying to say with your symbols
you should show some sequence of elts of Y converge to something not in Y
(it's enough to observe Y is dense but not all of l^2, which automatically means there are things in l^2 that things in Y converge to but is are not in Y)
 
9:35 PM
@anon Why if we show that l^2 is not dense do we deduce that there are things in l^2 that things in Y converge to but are not in Y ?
 
I never said anything about "l^2 is not dense" (which is nonsense)
I said Y is dense in l^2 but Y is not all of l^2
 
A ok... But how could we find such a sequence? @anon
 
"Y is dense in l^2" means everything in l^2 is converged to from things in Y, but the fact that "Y is not all of l^2" means some of those things converged to from Y are not in Y
@evinda you already have
56 mins ago, by anon
@evinda right, (x1,...,x_n,0,0,...) tends to (x1,x2,...) as n->inf
 
A ok.. I think that I got it... I will rethink about it.

Now in order to show that d(x,Y) is not attained for some $x \in l^2(\mathbb{N})$, could we pick for example $x=\left( 1, \frac{1}{2}, \frac{1}{3}, \dots\right)$ ? @anon
 
mmhmm
 
9:40 PM
How can we pick a x so that the distance is not attained? @anon
 
I said yes, you can pick x=(1,1/2,1/3,...)
 
@anon Oh sorry...
@anon So we want to calculate $\inf \{||x-y||: y \in Y\}$.

If $y \in Y$, then it is of the form $(x_1, x_2, \dots, x_n,0, \dots,0)$.
Then $||x-y||=||\left( 1, \frac{1}{2}, \frac{1}{3}, \frac{1}{4}, \dots \right)-(x_1, x_2, x_3, \dots, x_n,0, \dots,0)||$.

But how can we find the infimum ? Do we take as the first $x_1, \dots, x_n$ the first n elements of x respectively?
 
you have things in Y that converge to x
that means the distance between those things in Y and x converges to 0...
 
So you mean that we pick $x=\left( 1, \frac{1}{2}, \frac{1}{3}, \dots\right)$ and say that since Y is dense all the elements of Y convergence to x, and thus $||x-y|| \to 0$ for all y in Y? @anon
 
 
1 hour later…
11:05 PM
@anon I finally found what I was asking about math.stackexchange.com/questions/1521632/…
 
00:00 - 17:0017:00 - 00:00

« first day (1941 days earlier)      last day (3078 days later) »