« first day (5152 days earlier)      last day (164 days later) » 

01:10
Anyone watching the Trump v Harris debate?
No.
๐Ÿคนโ€โ™‚๏ธ๐ŸŽช๐Ÿคนโ€โ™‚๏ธ
๐Ÿšซ๐Ÿšซ๐Ÿšซ
01:49
๐Ÿ™Š๐Ÿ™‰๐Ÿ™ˆ
๐Ÿ™ˆ๐Ÿ™ˆ๐Ÿ™ˆ
๐Ÿ™‰๐Ÿ™‰๐Ÿ™‰
 
2 hours later…
03:30
@RyderisnotRude. all great apes can in theory get scurvy
The issue with humans is that we changed environment where fruits are no longer in abundance
 
1 hour later…
04:53
Hi everyone,

I'm currently working on a probability problem involving a deck of 52 cards (26 red, 26 black) and Iโ€™ve written a proof that Iโ€™d like some feedback on. Could someone help verify if my proof is correct or provide guidance on how I might strengthen it? Any help would be greatly appreciated!
05:27
Askaway.
Does anyone know about any freely available 3d plotting software where I can define colouring a parametric surface using the parameters themselves? I mean, the colour can be defined as a function of u and v (shown in the WolframScript image below). Wolfram does support it, but I am unable to interact (zoom+panning) with the surface.
 
2 hours later…
08:04
@RyderisnotRude. I posted it here but did not got any answers: math.stackexchange.com/questions/4969735/… .
Would you mind taking a look at it?
08:33
Calculating the Legendre symbol is funny, as getting the correct answer doesn't imply you did everything correctly, it means you made an even number of errors :p
09:29
I'm trying to show that the integral is a complex linear functional on the space of all complex-valued integrable functions, call it $L^1$. We know this space is a vector space because $|af+bg|\leq |a||f|+|b||g|$, and I know the corresponding real integral is a linear functional on the corresponding real vector space. For instance, how do you proceed to show $\int a f=a\int f$ for $a\in\mathbb C,f\in L^1$?
I just get stuck writing out the definitions; $f=u+iv$ and then $\int a f=\int (au+iav)$.
@SohamSaha have you tried geogebra 3d or desmos 3d?
10:16
@SohamSaha Yes, SageMath lets you specify a colour function for a parametric surface in terms of the parameters. See doc.sagemath.org/html/en/reference/plot3d/sage/plot/plot3d/… Here's a short demo:
10:35
@psie I worked it out. It's not hard, but it is a bit of writing.
 
2 hours later…
12:57
@SoumikMukherjee I've tried Desmos (I sent a feature request for the same too, and they said they will try to implement it in the future) (as of now the best you can get with Desmos3D for algorithmically colouring the surface is something like desmos.com/3d/rarorfr7fi). I've not actually used GeoGebra, but I don't think they support the feature.
@PM2Ring That was exactly what I was looking for. Thanks a lot. (Now I just need to learn how this SageMath works...)
13:18
hi
@SohamSaha Sage is rather huge. It helps if you already know some Python. There's some good introductory stuff here: sagemath.org/tour.html
@PM2Ring Sorry to disturb you again. This works in the sagecell, but doesn't run in the local sagemath that I just downloaded.
Seems like colormaps.plasma is not defined locally. Any idea for why that might be happening?
And yes, Python definitely helps...
13:42
@PM2Ring Strange... Seems like the local version doesn't support all the colormaps... 'colormaps.Blues' works pretty well locally
@SohamSaha Oh. There might be a fix for that. Give me a few minutes. In the meantime, this should print a list of Colormaps:
Yes, got the list with sorted(Colormaps())
The local list is shorter than the one on SageCell()
14:02
@SohamSaha See if this gives you a longer local list:
@PM2Ring Got it now. I was missing the matplotlib.pyplot.cm package. This works locally.
@SohamSaha Oh, good.
@PM2Ring Would it be possible for you to solve a last problem with SageMath?
I have another issue
Regarding the colormaps
It's also possible to build your own colormaps. The details are in the matplotlib docs. But here's a quick demo, with a parametric Klein bottle.
@SohamSaha What's the issue?
Thanks for the link. My issue is: The first line works, but the second doesn't here
But pi is apparently defined, or else (u,0,pi) wouldn't work...
14:17
@SohamSaha Yes, pi is a built-in constant. But it's "magic". ;) It's not a simple floating-point number. It has as many digits as you want, eg sagecell.sagemath.org/…
@PM2Ring That's wonderful... Really.
So n(pi, digits=k) works in my code. But how does (u,0,pi) work?
Why don't we need the n() there?
But the color function needs to output a plain numeric value. So you have to convert any symbolic expressions to a numeric form. The simplest way is with the n() function, which is a short-cut for numerical_approx() sagecell.sagemath.org/?z=eJzLSM0p0MjT5AIADQkCcw==&lang=sage
@SohamSaha Because the actual plotting function is smarter, and can deal with symbolic expressions.
Got it now.
Thanks for introducing me to Sage. It's a really wonderful language
You need to go through the intro tutorial, and learn a bit about Sage types & expressions.
You have any good begineer-to-advanced tutorials in mind?
14:24
That page I linked earlier has good intro material
Ok. Thanks a lot again.
And sorry for disturbing you for so long
Sage lets you create expressions like sqrt(2), which it can manipulate exactly, not as a floating-point approximation. Similarly, with stuff like sin(pi/8). It's very powerful, but it can also be confusing at first.
@SohamSaha Not a problem.
@PM2Ring So its a lot like PARI in that respect I think?
Except that PARI automatically does the n() thing when required, I think.
And it can be annoying when you forget to convert stuff to numeric, and you find that you've accidentally created a huge symbolic expression that takes ages to process. :)
14:28
@SohamSaha It has PARI built-in, which it uses for a lot of stuff.
@PM2Ring Wait, what!
No wonder it took up a lot of space on my PC
You can also call PARI routines directly, if you need to. But mostly its simpler to let Sage handle the messy details.
Also, maxima, sympy, and of course numpy.
And mpmath
So, is the sage console like a (python sagemath library+python interpreter)?
Or is it more like an independent language by itself?
@SohamSaha Pretty much. It runs on ipython
14:34
So it has all the added extras of ipython. Although some of that stuff isn't available in SageMathCell.
But I can use them in the local Jupyter Notebook right?
Sure
But I've never used Jupyter. So don't bother asking me for details about that. ;)
Thanks again, for everything. Would it be ok if I ping you here in the future if I get stuck again?
I just use SageMathCell, mostly on my phone.
@PM2Ring It works with good speed on phones too?
14:37
@SohamSaha Sure. Or in the Python room: chat.stackoverflow.com/rooms/6/python
@PM2Ring Ok
Sage is kind of a dialect of Python, so its kind of on topic in the Python room. But I try not to flood the place with Sage stuff.
But I guess it's more on-topic there than it is in this room. :)
@PM2Ring Yeah, the initial part of our discussion was well-suited here, but not what it turned into :)
@SohamSaha Well, most of the work is being done on the SageMathCell server. Except the GUI stuff is done by JavaScript in your browser.
@PM2Ring Right.
Gotta go now. Bye.
14:43
Bye
 
1 hour later…
Bob
Bob
16:08
Hi
Is anybody around?
@Bob yes, do you have a question?
Bob
Bob
not really
just felling a bit board today
I suspect your math skills are top notch, mine are not
what does it mean to feel board?
Bob
Bob
sorry, spelling error
bored
16:18
my mathematical skills need to improved like anyone elses
Bob
Bob
I am 62 and retired
in comparison to someone I might have areas where my mathematical skills are better, and some are worse
I don't think people whose math skills are really "top notch" hang around in chat rooms like this much
Bob
Bob
from what I have seen, there are a lot of PhD math students here
Bob
Bob
16:20
that has been my experience
or at least my impression
I don't think so
on the site in general, yes, there are some PhD students or one's who left university
Bob
Bob
not sure
I do not come here that much
credentials aren't really that important as much as knowledge is, and the area I care about, which is general topology, I know I have a lot to learn about still
I wouldn't call my skills top-notch then, but I am happy when I am able to help someone
Bob
Bob
I have done some tutoring in math
high school students
some college
like 1st year college
and from I have seen, it is not good
16:24
as in, they didn't want to learn? Or something else?
Bob
Bob
they wanted to learn but they did not want to work
and their background was not so good
for example, I was helping one student take Calculus
I think maybe that's natural. They come to tutoring because they don't know how to learn
Bob
Bob
and she did not know basic triv
she did not know basic trig
part of it is that in high school it is open book exams
and in college it is closed book exams
and they cannot look up the formulas they should have learned in high school
I see, when I was growing up you couldn't just open books at exams. And on the maturity exams, I didn't use the tables with mathematical formulas either, I didn't need them
Bob
Bob
high school teachers are expected to give at least 50% As
if you want 50% of your class to get at least 90 on an exam
and students do not like to memorize
what are you suppose to do?
16:29
I suppose you just do what you can
Bob
Bob
I am in the United States
There was a time when I thought I wanted to be a high school math teacher
now you can bring the formulas in the exams
in general
Bob
Bob
it depends on the teacher but you often can
also, there is a general trend away from rigor and proofs
for example, most schools no longer teach the derivation of the quadratic formula
I don't teach, the only person online who I know that teaches is @XanderHenderson. I feel like I have nothing to really add to the discussion
Bob
Bob
16:34
yes
do students really need to know proofs?
in high school?
I believe they do. Because mathematical thinking is beneficial for general logical thinking
Bob
Bob
In high school
the problem is that if you cover it
is it fair to test on it?
Perhaps not as much "know them" as be able to, perhaps derive them if needed
Bob
Bob
remeber if less than 50% of students get As the teacher is likely to be fired
16:36
during the fifth year of high school, I remember the teacher proved almost all the important theorems of calculus (analysis)
Bob
Bob
Here in the US, we only have 4 years of high school
Here is a true story. A friend of mine was teaching pre-calculus. One of his students
@Bob perhaps it'd be fair to present them with a proof and give them a test to point out some things about the proof, just to see that they understand logical thinking
Bob
Bob
had gotten all As so far in math. The student needed a calculator to find:
2*(-2)
@Jakobian and when the students do poorly you will have a problem.
and they think $\sqrt{(-5)^2}=-5$
Bob
Bob
not sure about that one
16:40
well maybe not that, but some certainly believe that $\sqrt{x^2} = x$ is valid for all $x$
Bob
Bob
I suspect that they can use a calculator and find out that it is not the case
that is, without plugging any particular $x$ they'd probably believe its true
Bob
Bob
I had one student who told me that if a problem required more than 3 lines to answer it, it was too long and he did not want to do it.
He was a high school freshman
nice chatting
bye
good bye
16:59
There's this proposition in Folland:
> Proposition 2.23: If $f\in L^1$, then $\{x\in X:f(x)\neq 0\}$ is $\sigma$-finite.
Let $(X,\mathcal M,\mu)$ be a measure space and recall a set $A\in\mathcal M$ is called $\sigma$-finite for $\mu$ if $A=\bigcup_1^\infty A_j$ where $A_j\in\mathcal M$ and $\mu(A_j)<\infty$ for all $j$. I don't currently see it myself but I know it is true: does it somehow follow from this proposition that if $f\in L^1$, then $f$ is finite $\mu$-a.e. on $X$? I think this is a useful result.
Hi
Perhaps $[|f(x)|> {1 \over n}]$.
That $f$ is finite $\mu$-a.e. means that $\{x\in X: |f(x)|=\infty\}$ has measure zero
17:08
@SineoftheTime Can i let you read what I wrote for "To what does a sequence of functions converge?", could you tell me I need to fix something?
what were you trying to say, @copper.hat? :)
@Pizza sure
Suppose that $f_n(x)$ converges pointwise on $I$, i.e., for every fixed $x \in I$, we have $f_n(x) \to l_x$. Then, there exists a correspondence $x \in I \to \lim_{n \to \infty} f_n(x) = l_x \in \mathbb{R}$. This correspondence is a function due to the uniqueness theorem of limits. We demonstrate that this function is defined as $f: x \in I \to l_x \in \mathbb{R}$, with $l_x = f(x)$. This function is called the **limit function**.

**Definition**: $f_n$ converges pointwise to $f$ on $I$ if and only if:
@psie this should follow from $\{1/n \leq |f|\}$ having finite measure
psie there are a lot of ways of seeing that thing you are asking about. one approach that suggests itself is in terms of what you were discussing the other day. if f is not finite mu-a.e. on X, at least one of the sets {f = +oo}, {f = -oo} has nonzero measure. but then one of int f+, int f- would be infinite (because int_X f+ >= int_{f = +oo} f_{+} and similarly for int_X f-) and f would not be integrable
17:16
@Pizza looks good
to see this, $\mu(1/n \leq |f|) = \int_{1/n\leq |f|} 1 \leq n\cdot \int |f| < \infty$
minor lemmas in there being things like "if P has nonzero measure then int_P +oo = +oo" and "if P subset X and g is nonnegative then int_X g >= int_P g"
thus we have [...]. Is it relevant?
psie there is a lot of what i might call 'path dependence' in this area where depending on what you've just done, or are willing to use as a black box, some results will have really straightforward, right-out-of-a-black-box explanations, and others will require some level of nuance. helps to keep in mind if you study out of more than one resource on the subject
@SineoftheTime mm no
17:19
It's not wrong tho
one thing you can say is that even if $(f_n)$ are continuous, it's not guaranted that $f$ is continuous
e.g. one book going into some iterative construction with sequences of subsets and epsilon/2^n in it to prove X isn't a sign that "the proof of X" "needs" that kind of argument, it just might be that a low tech, high nuance argument is all that a particular author in a particular place in exposition can do
ok ๐Ÿ‘ yeah, I'm not sure from what result it is the most "efficient" way of deducing what I want, but the above proposition appears quite early on in the section where $L^1$ is introduced, so I thought it somehow should follow from this
similarly once you have a lot of the basic properties of the integral (e.g. monotone or dominated convergence theorem) you will find that its pretty common for people to use those properties as their black boxes in arguments, or in mentally organizing what explains what, instead of going back to 'first principles' about measure
i think there's more variability in how books choose to organize this stuff than there is in how books approach a lot of other analysis adjacent things
@psie I suppose that once you can write $\{f\neq 0\} = \bigcup_n A_n$ where $\mu(A_n) < \infty$, then $\mu(\{|f| = \infty\}\cap A_n) = \mu(\bigcap_m \{|f| > m\}\cap A_n) = \lim_{m\to\infty} \mu(\{|f| > m\}\cap A_n)\leq \lim_{m\to\infty} m^{-1}\int |f| = 0$ for all $n$.
from which it follows $\mu(|f| = \infty) = 0$
@SineoftheTime Should I add it to the end of the definition?
17:24
no
it's not part of the definition
what you wrote is perfeclty fine, you can keep it as it is
ok :)
actually like, we don't even use sigma-finiteness anywhere
$\mu(|f| = \infty)\leq \lim_{m\to\infty}\mu(|f| > m)\leq \lim_{m\to\infty} m^{-1}\int |f| = 0$
@Pizza other doubts?
@Jakobian ok, thanks
I didn't quite understand this
$f_n$ converges uniformly to $f$ on $I$ $\Leftrightarrow \lim_{n \to +\infty}\left(\sup_{x\in I} |f_n(x) - f(x)|\right) = 0$, meaning:

$\forall \epsilon > 0, \ \exists N_\epsilon \ \text{such that} \ \forall n > N_\epsilon \ \forall x \in I, \ |f_n(x) - f(x)| < \epsilon.$
how do i find that sup?
17:39
psie asked this a while ago
let me see if I can find it
Aug 10 at 10:01, by psie
I struggle with understanding one direction in the equivalence between uniform convergence and convergence in sup-norm. In particular, why does $$f_n\to f\text{ uniformly on }E\subset\mathbb R\implies \sup_{x\in E}|f_n(x)-f(x)|\to 0\text{ as }n\to\infty?\tag1$$The other direction is immediate since $\sup_{x\in E}|\ldots|$ is an upper bound to $|f_n(x)-f(x)|$ for every $x$ (and using squeeze theorem), however, the implication in $(1)$ I don't see. Is this clear to someone?
@Pizza you can click on the permalink and read the conversation
Wowow I'm really starting to get into abstract algebra
what topic are you studying?
group theory
@SineoftheTime oh ok, I'll try to do an exercise
@Jakobian Boo!
17:57
@leslietownes you write int_X f+>=int_{f=+oo} f+ . Now, how would you show int_{f=+oo} f+ is infinite?
int_{f=+oo} f+ is just int_X f^+ chi_{f=+oo} and by definition, this is the supremum over all integrals of simple functions phi such that 0<=phi<= f^+ chi_{f=+oo}. I guess we can simply take phi=r chi_{f=+oo} for some real r.
no way leslie convinced you to type oo instead of infty
@XanderHenderson $\text{(hangel)}^{-1}$ Gabriel
@SineoftheTime I follow leslie on lesliegram. I just copy what leslie does.
psie: yes you could take e.g. the family of 'constant' simple functions c 1_{f = +oo}, c >= 0, to compute/realize that supremum
right, makes sense ๐Ÿ‘ although c can not equal zero, otherwise we get 0*oo=0
18:10
it's fine if you get that, it just isn't particularly relevant to the supremum
:)
the supremum of [0, +oo) and (0, +oo) are the same thing
ok, true
@SineoftheTime $f_n(x) = \begin{cases}\text{a if} \ \ \ 0 < x \leq \frac{1}{n} \\ \text{b if} \ \ \frac{1}{n} < x \leq 1 \ \ \ \text{or} \ \ x = 0\end{cases}$

the limit function is $f(x) = b \ \ \ \forall x \in [0,1]$ Then $f_n \to f(x) = b$ punctually in $[0,1]$. But $\text{sup}_{x\in[0,1]} |f_n(x) - f(x)| = |b-a| \ \ \text{not} \to 0.$ Convergence is not uniform
to speed up the computation of the supremum just consider the values c = 2^{2^{2^n}}}, n = 1, 2, 3, ...
save mental neuron time
using tricks like that i can compute dozens of suprema before i hit the gym at 3:50 am every day #grindset
why it is not |b-b| given that the sup in [0,1] is 1?
@Pizza I'm not following you
the sup of what is $1$ in $[0,1]$
18:19
@SineoftheTime given that $x \in [0,1]$ the sup is $b$, because if $x = 1$ the sequence of functions is $b$
do you mean $\sup |f_n-f|=b$ ?
i mean $\sup |f_n-f|=|b-b|$
did you draw $|f_n-f|$?
I don't understand why |b-a| is there
it seems that you're implying that the sup is achieved at $x=1$
18:22
@SineoftheTime I only have the drawing of $f_n(x)$
can you draw $|f_n-f|$?
@SineoftheTime yes
why this should be correct?
this is true if $x\mapsto |f_n(x)-f(x)|$ is increasing
@SineoftheTime from $0 < x \leq \frac{1}{n} \ \ f_n(x) - f(x) = |a-b|$. If $\frac{1}{n} < x \leq 1 \ \ \text{or} \ \ x = 0 \ \ \ f_n(x) - f(x) = 0$
so if the sup is the largest value $|a-b| > |b-b|$ , so the sup is $|a-b|$?
yes
here the sup is a max
19:03
@psie If $f$ is $L^1$ then must be finite ae. If not, then $\int |f|$ would be infinite.
@copper.hat ok ๐Ÿ‘
20:05
Hi
Does anyone have any strategies for focusing on studying?
20:34
Decide what you want to achieve in small chunks and then do it.
20:59
@Gian'sPizzeria stop coming here, for a start :^)
at least during study times
> DCT: Let $\{f_n\}$ be a sequence in $L^1$ such that (a) $f_n\to f$ a.e., and (b) there exists a nonnegative $g\in L^1$ such that $|f_n|\leq g$ a.e. for all $n$. Then $f\in L^1$ and $\int f=\lim\int f_n$.
> Proof: $f$ is measurable (perhaps after redefinition on a null set) by Prop. 2.11 and 2.12 ...
I'm a bit confused by Folland's proof here of why $f$ is measurable. If we deal with $L^1(\overline{\mu})$, then $f$ is measurable from Prop. 2.11 (end of story...Prop. 2.11 just says $f$ is measurable if $f_n\to f$ a.e. and $(f_n)$ are measurable), but if we deal with $L^1(\mu)$, then how does Prop. 2.12 imply $f$ is measurable?
Here's Prop. 2.12:
> Prop. 2.12: Let $(X, \mathcal M,\mu)$ be a measure space and let $(X,\overline{\mathcal M} ,\overline \mu)$ be its completion. If $f$ is an $\overline{\mathcal M}$-measurable function on $X$, there is an $\mathcal M$-measurable function $g$ such that $f=g \ \overline{\mu}$-almost everywhere.
Prop. 2.11 assumes of course the measure is complete.
@psie Folland is not claiming that $f$ is measurable
uhh
I read a "Proof: $f$ is measurable [...]" there
21:14
the [...] is important you know
Again, Folland is not claiming that $f$ is measurable, but that we can find an $f_0$ such that $f = f_0$ a.e. and $f_0$ is measurable
what function is he claiming is measurable?
ah ok
thats what the remark in the brackets means
why is he saying "perhaps" though?
it just means that "it might be necessary to"
ok
@Jakobian when might that be the case?
21:18
@psie whenever $f$ is not measurable
I am not certain why Folland claims that $f\in L^1$, perhaps he means that $f\in L^1(\overline{\mu})$ here
where $\overline{\mu}$ is the completion of $\mu$
one has to carefully see what is he proving
yeah, he says prior to this theorem that $L^1(\overline{\mu})$ is "identified" with $L^1(\mu)$
or that we shall "identify these spaces", whatever that means
I see, so he essentially doesn't care
27
Q: Measurability of an a.e. pointwise limit of measurable functions.

user145993Suppose that $(f_n)_n$ is a sequence of measurable functions on a set $E$ and that $f_n \to f$ a.e.on $E$. Does this imply that $f$ is measurable? I know that pointwise limit of measurable function is measurable. But here we only have convergence a.e. So I got confused.

here's a post which shows that a.e. pointwise limit of measurable functions need not be measurable
@BenSteffan but for example if I read something and I still don't understand it, should I continue?
Or I have to stay on that thing until I learn it
it depends
you don't have to get caught up on every little thing
if you think the thing is important, spend some time trying to understand it
the fact that a.e. pointwise limit of measurable functions need not be measurable, just shows that I had some gaps when it comes to my measure theory knowledge
21:25
if you think it isn't, balance how much material you have/want to cover and how much time you have, and if the balance is unfavourable move on
because I didn't know that, I thought it's always measurable
there's also value in giving material a superficial reading to come back to it later and study it in detail
for instance for making exactly these kinds of judgements: what's important and what isn't
@Jakobian yeah, completion is a key assumption. You said you were not certain why $f\in L^1$. What were you uncertain about?
as with everything, studying is full of tradeoffs you have to make
@psie of course the measurability of $f$. But as Folland identifies $L^1(\mu)$ with $L^1(\overline{\mu})$, there is nothing to worry about
21:27
ok, I see
@Gian'sPizzeria don't study
even though all we can say is $f\in L^1(\overline{\mu})$, we can still find another representative for $f$, say $f_0$, for which $f_0\in L^1(\mu)$. And that's how the theorem should be understood
@SineoftheTime meditate on the text until your third eye opens and you gain direct access to the secrets of the universe :^)
@Jakobian yeah exactly (that's Prop. 2.12 above)
@BenSteffan keep the books open at night so you can absorb the information while sleeping
21:29
@Gian'sPizzeria I don't know
ah the ol' book-under-pillow trick
overthinking studying is also not good
don't study; buy a gun, drive to the author's home and have them explain the material to you at gunpoint
(this is a joke etc.)
just try to do it, and if you can't, well, then I suppose you need to come back later
remind yourself that you are a living being and everything that is happening in your life can influence your studying
so we can go further and make the question "how do I set up my life so that I don't have such hard time studying". And I have no idea
"People work faster if threatened with a gun"- Sun Tzu
21:35
most people will try to tell you how to study efficiently
but they won't tell you how to have the will to study, or energy for it
hydration, satiation, regular exercise, this will all help you to study better I believe
Thanks very much for the advice!
@Jakobian I'm also reading some lecture notes that are based on Folland's text. I highlighted a sentence here which confirms our suspicion; that the measure needs to be complete (otherwise Folland would have needed to assume $f$ is measurable).
@psie yes. I suppose it's not a bad thing that you check those things in other sources
22:31
Thinking some more about how Folland has worded the dominated convergence theorem, it is really confusing (especially after reading peek-a-boo's answer).
> $f$ is measurable (perhaps after redefinition on a null set)
It is 1) unclear if he deals with a complete measure or not and 2) if he assumes $f$ is measurable or not.
I would just ignore the whole paragraph. Unless they show you the proof, you'll spend too long on that one remark.
Bml
Bml
@Semiclassical I have made progress in the work of finding the roots and have been able to achieve the same results as you. However, I am still left with one last question regarding the root you discarded. When you are free, could we talk about it?
22:49
@psie why would it be confusing?
@psie I thought we went over this. Neither
@Jakobian because peek-a-boo now speaks of $f\in L^1(\mu)$ in their answer
@Jakobian clearly he has to assume the measure is complete
@psie and?
@psie Did you miss the whole conversation we had
@Jakobian I thought we agreed that all we can say is that $f\in L^1(\overline{\mu})$. Only then can we find an $f_0\in L^1(\mu)$ such that $f_0=f$ a.e., but not without the assumption $f\in L^1(\overline{\mu})$.
@psie Sure. But Peek-a-boo assumed $f$ is measurable
ok, so I guess Folland does not assume $f$ is measurable. But he must assume $(f_n)\subset L^1(\overline{\mu})$.
22:57
If you assume that $f$ is $\mu$-measurable, then of course there would be no question about its measurability
@psie No. Folland assumes $(f_n)\subseteq L^1(\mu)$
@Jakobian indeed
@Jakobian that makes no sense, then its limit might not be measurable
The conclusion is that the a.e. pointwise limit $f$ is in $L^1(\overline{\mu})$
@psie it does make sense
Folland now says that we can replace $f\in L^1(\overline{\mu})$ with some $f_0\in L^1(\mu)$
While $f$ might not be in $L^1(\mu)$, it represents an element of $L^1(\mu)$
this is why Folland said that he identifies $L^1(\overline{\mu})$ and $L^1(\mu)$
@Jakobian ok, but the link you sent earlier, didn't that show how if $(f_n)\subseteq L^1(\mu)$ and $f_n$ pointwise a.e. to some $f$, then $f$ might not be measurable because $\mu$ is not complete?
@psie that's true. And isn't in contradiction with anything I'm saying
@Jakobian ok, so you're saying that the conclusion is that the a.e. pointwise limit $f$ is in $L^1(\overline{\mu})$, i.e. $\overline{\mu}$-measurable. Does that not imply it is $\mu$-measurable?
23:08
The conclusion is that if $(f_n)$ converges a.e. to $f$, then we can find measurable $h$ with $f = h$ a.e. and then the statement of the theorem will hold for $h$
that is, $(f_n)$ converges to $h$, $h$ is in $L^1$
in essence, $f$ and $h$ represent the same equivalence class in $L^1(\mu)$
@psie Of course that doesn't imply it's $\mu$-measurable
If $f$ is any function, then $f$ will represent some element of $L^1(\mu)$ if there exists $h\in L^1(\mu)$ such that $f = h$ a.e.
Then you could identify $f$ with $[f] = \{h\in \mathcal{L}^1(\mu) : f = h\text{ a. e.}\}$
where I use $\mathcal{L}^1$ to mean functions, and $L^1$ to mean equivalence classes
What Folland is saying, in essence, is that the equivalence class $[f]$ is meaningful here (i.e. the set is non-empty), and the statement holds if we plug any of its elements in place of $f$
The statement is only confusing if you consider it literally - which it isn't meant to be considered
ok
And since we went over this like 5 times, I'd imagine you shouldn't be confused about it anymore
well...
I do have one final question.
would $(f_n)\subseteq L^1(\overline{\mu})$ be too strong of an assumption or not make sense at all?
You could just assume $\mu$ is complete at that point
true
23:22
If $\mu$ is complete, then $f$ is $\mu$-measurable, literally speaking
and the statement of the theorem holds as is written
23:49
@SoumikMukherjee 8.5/9 :D

« first day (5152 days earlier)      last day (164 days later) »