« first day (753 days earlier)      last day (4268 days later) » 

user19161
12:00 AM
@Khromonkey What do you mean? You wanted to ask him, he answered your question on main, so the most natural thing to do is to comment on that itself.
 
@Khromonkey Well, that's an interesting deduction!
 
user19161
@PeterTamaroff No, that is a wrong deduction!
 
I already asked you guys like 4 times if you could help me
 
user19161
@Khromonkey Not everyone reads every line of the transcript you know.
 
@Khromonkey I can think about it, but give it some time.
 
12:01 AM
You are planning on continuing ad nauseam?
 
@PeterTamaroff you have $\pm \sum x_i y_i \leq \left| \sum x_i y_i \right| \leq \sum |x_i||y_i| \leq \left( \sum |x_i|^p\right)^{1/p} \cdots$, so the result stating the last inequality states more than just $\sum x_i y _i \leq \left( \sum |x_i|^p\right)^{1/p} \cdots$: stronger inequality = sharper result.
 
@Khromonkey I have my problems myself, too! =)
 
user19161
@PeterTamaroff Yes, like the Mendelson exercises. 8-)
 
@t.b. Oh, OK. I'm not sure what role does $p^{-1}+q^{-1}=1$ play here. Then, I haven't really give it a long thought.
I just downloaded SLENDER. If I type something like "lkshjdgklshjgfklhjsfklghjklHJ"
 
I finished mendelson last week
 
12:03 AM
@PeterTamaroff well, Cauchy-Schwarz is the case $p = q = 2$.
 
It's because I'm having aheart attack.
 
$\frac12 + \frac12 = 1$, right? :)
 
If you have probability spaces you can get rid of the $=$ and get a $\leq$.
 
user19161
@PeterTamaroff Observe that means the sum is the product.
 
@t.b. Yes, yes
@JasperLoy Ok, that's nice.
 
12:07 AM
How can you be sure that at least two pairs with the same D dont share points?
 
@Khromonkey What are talking about?
 
well we know that at least two pairs share D right?
but we need to prove that two of them dont share any elements
 
@Khromonkey What are you talking about?
 
4
Q: combinatoric proof

Khromonkeysuppose $A$ is a set of 16 distinct natural numbers. such that if $p\in A $ then $1\leq p\leq100$ prove there exist numbers $a,b,c$ and $d \in A$ for a, b, c and d distinct such that a+b=c+d

 
@Khromonkey OK.
Have you tried to think what's special about $16$?
 
12:11 AM
Yes, nothing
 
and $1\leq a \leq 100$ iff $a\in A$?
 
@Peter It cannot be an iff, since A only has 16 elements..
 
i dont think it has to the with the fact its a perfect square, tessaract
what do you mean 1≤a≤100 iff a∈A?
 
@FortuonPaendrag Yeah, I just wrote many iffs today =P
@Khromonkey I should have writen $a\in A$ then $a\leq 100$.
 
@PeterTamaroff superfluous f = superfluous t? :P
 
12:14 AM
Yes, that is true, also $a\in A$ then $a\geq 1$
 
(I'll stop now)
 
@t.b. Hahhaa bring the LULZ in!
@Khromonkey That is implicit from saying elements of $A$ are elements of $\bf N$
 
@Khromonkey: Think about this. If the condition that elements of $A$ are less than $100$ is removed, can you find sixteen numbers so that FOR ALL Elements of $A$, $a+b \neq c+d$
 
@FortuonPaendrag yes. powers of 2
 
@PeterTamaroff: if you see Ben can you tell him to ping me, I should be here for a short while, but not for too long.
^Now that's a superfluous thing to say :/
 
leo
12:18 AM
first countable and separable does implies second countable?
and hi
 
@Khromonkey Yeah! So, the bound by 100 forces two of those values to be the same. SO, what is your answer to Brian's first question?
 
120
 
@t.b. Hey
 
or 240 if the order doesnt matter
 
hey there @Ben
 
12:20 AM
@BenjaLim What timing, Benjalim..
 
@t.b. I couldn't sleep the whole of lasst night
 
@FortuonPaendrag namaste
I was trying to figure out why
I suspect proving it may not be easy
 
leo
someone
 
12:20 AM
@BenjaLim Nhangka
 
@BenjaLim proving what?
 
@Khromonkey I guess this book may be useful for you.
 
what the closed subgroups of $\Bbb{R}^n$ are
 
@Khromonkey The order indeed does not matter, since we are dealing with sets, not ordered pairs.
 
then 240
 
12:22 AM
@leo No, the Sorgenfrey line is first countable, separable (and Lindelöf), but not second countable.
 
15*16
 
@t.b. Will this be helpful?
 
@BenjaLim nah.
 
@Khromonkey But half of those are repeats though
 
man
 
12:23 AM
because of the order?
then yes 120 pairs (a,b) such that a<b
 
leo
@t.b. Thanks :-)
 
@t.b. Hmmm
 
@Khromonkey yeah. But the answer to question 2 is ¿?
 
leo
@t.b. las colas derechas
 
12:26 AM
Yes, more precisely $\leq 99$, as may be the case.
 
@t.b. I think I should look at discrete groups
 
@BenjaLim Look at finitely generated subgroups
@BenjaLim yeah that's a good start :)
 
ok, how does that prove it?
 
@leo I see. I still have a Spanish topology book (way down) on my "to read" pile :)
 
@leo HAHA what??????
 
12:27 AM
@Khromonkey Because there must exist $a,b,c,d$ such that |a-c|=|b-d|, by the pigeonhole principle.
 
leo
@t.b. haha nice
 
but a,b,c,d are all distinct (as brian pointed out)
 
leo
@PeterTamaroff la topología de las colas derechas :-)
it useful to provide many counterexamplesis
 
@Khromonkey No: Read his last comment on the matter.
 
@leo "colas" as in "butts"?? Or "glues"?
 
12:29 AM
He says that somehow we need to exclude that possibility (something he still hasnt done)
 
I have some good Rat Juice.
Maybe I should just get into the whisky and wine business.
 
@leo this one. I think Agustí Roig recommended it somewhere on this site.
 
i have good holy monkey.
 
A gentleman with the eyes of a boiled fish.
@JayeshBadwaik Did that little bastard devour your biscuits and mess up your kitchen?
 
12:32 AM
(I mean vodka)
 
@BenjaLim yeah, that's what I wanted you to prove :)
 
The hard part now is in proving that closed implies discrete
 
Vodka is only good to get maggot.
 
In February I will have drinking contest with Ben here. In number of bottles of vodka.
 
12:32 AM
@BenjaLim that's very hard because it's completely wrong :)
 
I mean closed subgroup @t.b.
 
i i had one yesterday night
still not thinking straight
 
still wrong. $\mathbb{R}$ is a closed subgroup of $\mathbb{R}^2$, right? :) @BenjaLim
 
@BenjaLim That follows from the Hitori-Usk theorem, Ben.
 
leo
@PeterTamaroff as tails I think. It is $\{]a,\infty[:a\in\Bbb R\}$
 
12:33 AM
@t.b. Hmmmm..........
 
@leo Oh XD
 
I always wonder: what did LaTeX do to him that he does this or that to it?
 
So wait I thought I was supposed to prove that every closed subgroup of $\Bbb{R}^n$ is free abelian of finite rank?
 
@BenjaLim well, you know more about the exponential map than just continuity, right?
 
@FortuonPaendrag see how it is not proven yet?
 
12:35 AM
@t.b. He lack all feeling for aesthetics.
 
@JonasTeuwen that's a mean thing to say just because you don't share his feelings for aesthetics :)
 
@t.b. Might be. But nevertheless true.
 
@t.b. Oh I think I know
we can use the fact that it is a local homeomorphism
 
"It's all subjective" and then shove a couple of burgers down your esophagus... damn.
 
@t.b. To prove that the kernel must be discrete
 
12:37 AM
@BenjaLim you know much more than just local homeo...
 
@Khromonkey yes. I do. We have only proven that there exist two distinct sets $\{a,b\}$ and $\{c,d\}$ such that $|a-b|=|c-d|$. This means that without loss of generality $a\neq c$.
 
@t.b. That's all I know.
Well it's a surjective group homomorphism now
 
@BenjaLim well, it's a local diffeo
 
because the tie algebra is abelian
@t.b. Ok we have not covered that in class yet....
Not so deep into the differential geometry
 
leo
@t.b. ugly
 
12:38 AM
@t.b. I have a proof
 
good, shoot!
(not you, leo :))
 
@FortuonPaendrag yes but what i required is $b\neq c$ right?
 
Discreteness is equivalent to proving that there is an open neighbourhood $U$ about every point $x$ in the kernel such that $U \cap \{x\} = \{x\}$
 
@BenjaLim yes.
 
@Khr Yeah, there must be someway to exclude this situation.. Im thinking..
 
12:40 AM
Hmmm my proof only works if I take $x$ to be the zero matrix....
 
?
 
Ok let me shoot anyway
Suppose for all $U$ about $x$, we have that $U \cap \{x\}$ containing some other point that $x$
choose $U$ small enough so that there is $V$ about the identity in $G$ such that $U$ is homeomorphic to $V$ @t.b.
 
How could I work out the proof for Holder?
 
@PeterTamaroff You know Young's inequality $ab \leq a^p / p + b^q/q$?
 
Hmm I made a mistake somewhere
 
12:43 AM
I hate mathoverflow
 
@Khromonkey I have another idea. We should try to show that there is an arithmetic progression of length $4$ in the set $A$. If that is true. It would be an easy proof then. What about MO?
 
@t.b. No. Darn! There are so many inequalities out there! We need more socialists in maths to work things the right way.
3
 
@t.b. But if I can get discreteness I can get countable immediately
 
@Khromonkey Are you discussing that 16 numbers less than 100 non-unique difference thing?
 
leo
@PeterTamaroff :-)
 
12:44 AM
yes
 
that the kernel as a set must be countable. This comes from the fact that $\Bbb{R}^n$ has a countable basis
 
@Khromonkey why are you in such a hurry?
 
Because my best friend managed to solve it in half an hour and I just cant
 
heh :)
 
si hubiera justicia en el mundo debiera tenerse, para funciones integrables,
$$\lim_{n\to \infty}\int_0^1 f_n dx=\int_0^1 f dx$$
Topology book FTW
 
12:46 AM
@PeterTamaroff well, you know that Young is just AM-GM, right? Now call the anarchists in :)
 
@t.b. Given any ball about $x$, we know that all the points it contains cannot all be in the kernel
because the kernel would then be open
 
I think the solution has to do with looking at pairs ((a,c),(b,d))
 
and then it would be the whole of the lie algebra, contradicting the exponential map not being the zero map
 
Gerry Myerson is stricking me as either bitter or ill tempered.
 
leo
@PeterTamaroff then the world would be a very nice place
 
12:48 AM
@BenjaLim well, you got a neighborhood around the identity where you are a homeomorphism. Conclude.
 
@Khromonkey Khro, Are you on Project Euler?
 
yes
 
@t.b. Yes I am trying to untangle the knots now
 
leo
can someone help in here? (I'm just starting differentiable manifolds)
 
@BenjaLim you look like a kitten trying to do that :)
 
12:49 AM
huhuhuh?
 
leo
@t.b. :D
@PeterTamaroff it must be signs's table or table of signs (tabla de signos)?
 
@BenjaLim like so
 
why do you say I'm like a kitten?
 
@t.b. Aww!
 
leo
@t.b. does AM-GM implies Young's inequality?
 
12:52 AM
@leo table of signs
 
leo
@PeterTamaroff thanks
 
you never use "'s" for objects
 
leo
@PeterTamaroff solo para nombres propios?
 
@leo more for the "living"
 
12:54 AM
@BenjaLim well, your "untangling knots" sort of gave me that picture. No offense intended.
 
@t.b. i guess it is not weird to say "the dog's food"
 
Sigh, I always get tied up in knots
 
leo
@t.b. I always proved it by using a function, find its minimun, etc
 
@BenjaLim because you bring in the lulz?
@t.b. now, how can I prove Young's?
 
@leo sure, the basic inequalities look more like a space-filling curve than a segment
 
leo
12:55 AM
I like the proof given in Zygmund & Wheeden, it is a kind of geometric proof
 
@BenjaLim homomorphism and local homeo is already enough.
 
@t.b. If I know that the kernel if isomorphic to $\Bbb{Z}^k$ for some $k$, then $\Bbb{R}^n/\Bbb{Z}^k \cong \Bbb{R}^k/\Bbb{Z}^k \oplus \Bbb{R}^{n-k}$ finishing the problem yes?
 
yes.
 
@t.b. One day, my first abstract algebra teacher got tired of saying the entire word "homomorphism" and said "These maps are clearly Homo and these are..." and he realized what he had said. Needless to say, the rest of the lecture was incredibly awkward.
 
I can imagine... That's why I always use an addtional "mo".
(I try not to abbreviate anything in lectures).
 
1:00 AM
@FortuonPaendrag at least he doesn't say "Guys, Anal is not that hard." to freshman Real Analysis students.
 
LET ME GET A COFFEE
 
@BenjaLim I DON'T LET YOU. STAND PUT.
 
@PeterTamaroff But Analysis is already brief though :)
 
@t.b. This problem is good. It means that equating a torus and an abelian, compact, connected matrix lie group are not different from each other.
 
I once sat in on a friends abstract algebra class, and the prof abbreviated finite abelian groups as FAGs.
 
1:02 AM
@BenjaLim I should go to bed now... You can try this if you're completely stuck.
 
@t.b. Thanks man
@t.b. Have a good sleep
I think I can finish it from here.
 
@BenjaLim yeah, it's a nice problem. :)
Good night y'all!
 
Gute Nacht, @t.b.
 
Good night.
 
@FortuonPaendrag How can I work out a proof of $xy\leq {x^p \over p}+{y^q\over q}$ given that $p+q=pq$?
I mean, reason it out
 
1:09 AM
@Peter Young's inequality. I do not remember off the top of my head. @Jonas might be more knowledgeable in this area.
 
Just draw it.
 
Ahhh. @JonasTeuwen I think t.b. did that in an answer to a question once? Or was it you?
 
Both I guess.
 
Thank you. :)
 
user19161
@JonasTeuwen Sure, but after that one still needs to write down the proof more rigorously.
 
1:10 AM
@JonasTeuwen Draw it?
 
user19161
@PeterTamaroff The graph of some function.
 
I "solved" a problem on project Euler, but I feel I've cheated, I just used Mathematica: Select[Divisors[600851475143], PrimeQ]
 
@JasperLoy I don't wanna be drawing no bivariate functions, dawg.
@FortuonPaendrag Spoiler alert?
 
yes, I just realized.@peter
 
user19161
@PeterTamaroff I can't remember what to draw, but it is really intuitive dawg.
 
1:13 AM
@Peter Think of $xy$ as area of a rectangle in the plane.
 
@FortuonPaendrag Yes, I did think about that. I can't get the importance of $p+q=pq$
 
Now, what does the term $\frac{x^p}{p}$ remind you of?
It's a little vague, but think calculus.
 
@FortuonPaendrag $\int x^{p-1} dx$
 
user19161
@PeterTamaroff Yes!
 
user19161
Now try to compare some areas.
 
user19161
1:20 AM
You can read it in wikipedia actually, the pic is shown there too @peter.
 
Anyway, I must be off. Adios, gentlemen.
 
user19161
@FortuonPaendrag Bye!
 
user19161
Hey @ben, are you taking the several variables course this sem?
 
@JasperLoy When is a set of numbers a weigh?
 
user19161
@PeterTamaroff Weight just means you give more weight to some numbers like if you take weighted mean, I don't know what you mean.
 
1:27 AM
@JasperLoy It seems there isno axiom, just positivity... just asking.
 
1:46 AM
Good evening all
 
leo
@PeterTamaroff are you free to some english review?
 
@leo OK. I trade it for an explanation of the last inductive step here in "Proof 1: Finite form." =P
@mixedmath Good... night =)
 
;p
 
@leo Namely the part of the $i=2$ to $n+1$ summation.
 
leo
@PeterTamaroff okay :-), see here
 
1:49 AM
@leo Now, what do you need me to reviedw?
 
leo
@mixedmath hi :-)
 
hiya
 
@leo Very minor corrections done.
 
leo
@PeterTamaroff thanks :-)
 
@leo When writing "Let's (not lets)" remember it is an abbreviation for "Let us" =)
 
leo
1:54 AM
I see
 
"Lets" is "allows", as in "My teacher lets me play during class."
 
leo
:O
@PeterTamaroff you got the previous parts?
 
@leo Yeah.
@leo What's so surprising? XD
 
leo
@PeterTamaroff the surprising is:
5 mins ago, by Peter Tamaroff
"Lets" is "allows", as in "My teacher lets me play during class."
 
@leo Haha childhood dreams
 
leo
2:06 AM
we have:
 
@leo I have one silly question.
 
leo
@PeterTamaroff me estas preguntando por qué vale la desigualdad?
en ese paso?
 
Como estamos seguros que $$\sum_{i=2}^{n+1} \frac{\lambda_i}{1-\lambda_1} x_i$$ esta en el dominio de $\varphi$? (Para poder seguir a la siguente desigualdad) La desigualdad sigue por convexidad, pero esa suma tiene que estar en el dominio de $\varphi$)
En realidad no entiendo como aplican la hipotesis inductiva.
 
leo
correcto
bueno esta parte empieza "For a real convex function $\varphi$..."
 
@leo Que queres decir?
 
leo
2:13 AM
@PeterTamaroff si el dominio de $\varphi$ es todo $\Bbb R$ (como parece que se asume) entonces no hay problema
bueno, eso es lo que yo interpreto de "real convex function $\varphi$..."
 
@leo Ahh, bien. Yo pensaba que $\varphi$ era de la forma $\varphi: D\to \Bbb R$, con $D$ arbitrario pero $x_i \in D$ para cada $i$.
 
leo
@PeterTamaroff En ese caso creo que es suficiente que $D$ sea convexo
 
@leo Ahhh, claro. Bueno, dejando de lado ese detalle, como se aplica la hipotesis inductiva?
Es claro que $$\sum_{i=2}^{n+1} \frac{\lambda_i}{1-\lambda_1} =1$$
Ahh, duh, creo que entiendo.
No entendia porque pensaba que la hipotesis estaba mas restringida.
 
leo
ok
 
Es interesante como uno entiende cuando discute las cosas =P!
 
leo
2:18 AM
y en el paso:
se usa el caso $n=2$
 
@leo Claro
 
leo
o sea que la función es convexa
 
@leo La hipotesis es simplemente que para $n$ numeros la hipotesis es cierta. Como la suma de $i=2$ a $n+1$ es de $n$ numeros y la suma es $1$, la induccion demustra el teorema, si?
 
leo
@PeterTamaroff así es!
con lo del dominio, dado $D\subseteq \Bbb R$ convexo, si $x_1,\ldots,x_n\in D$ y $\sum_{i=1}^n \lambda_i$ con $\lambda_i\in [0,1]$ entonces $\sum_{i=1}^{n}\lambda_ix_i\in D$
 
@leo hehehe bien, eso era el detallín!
@leo Sip.
 
leo
2:22 AM
y la prueba es por inducción
 
@leo ahora quiero probar la desigualdad de Young.
Que sigue de esta
 
leo
ok
creo que no te dije mentiras :-)
 
@leo Hhmm?
 
leo
@PeterTamaroff estaba verificando, parece que todo está bien :-)
 
@leo Goooooooooooooooooooood. Viste "Todopoderoso?"
 
leo
2:34 AM
@PeterTamaroff Cuál? hay una nueva?
 
leo
2:53 AM
@PeterTamaroff :-)
it is
 
@JasperLoy Me taking dude!
 

« first day (753 days earlier)      last day (4268 days later) »