« first day (938 days earlier)      last day (4381 days later) » 

05:00
Yes, exactly.
and adjoint is $A^\ast = \overline{A}^T $ where $^T$ is the transpose
In R, yes. But in C, you have to take the transpose, and then the conjugate.
i am just making sure I have the right words as my classes where in german and my english is poor, as you already notet
does that make a difference? I thought that commutates
@Sanchez If $M(x)$ is the mertens function, can I conclude that $$1=\int_{0}^1\ln(x)M(\frac{1}{x}) dx=\int_{1}^\infty \ln(x)M(x) dx$$ sense $$\lim_{n\to\infty}\frac{1}{n}\sum_{k=1}^n\ln(\frac{k}{n})M(\frac{n}{k})=1$$?
$M(x)=\sum_{k\leq x}\mu(k)$
@anon
@julien Could you explain what you mean with $A^\ast$ is a polynomial $p(A)$ do you mean the characteristical polynomial?
05:07
@DominicMichaelis I believe jules means that A^* can be expressed as p(A) for some polynomial p().
clearly not the char poly, since that would be 0
Yes it makes a difference. In C, every normal matrix is diagonalizable via a unitary. (and the converse is easily seen to be true). Not true in R.
Yes, anon is right. If A*=p(A)$ clearly $A*A=AA*$. It is interesting that the converse holds.
@Ethan no, the last two integrals are not the same.
why not?
calculus: if $u=1/x$ then $du=-dx/x^2$ so $dx=-u^2du$
@julien yeah sounds interesting
05:10
@Ethan what do you mean "when are they the same"? It's not like they're changing every tuesday.
i will go to university now see you later :)
@anon I don't understand they seem to enclose the same space?
@anon as x->0, in 1/x, 1/x->infinity
Bye Dominic.
@anon Stahp! I'll choke. I'm eating.
05:12
@anon I altered the upper and lower bounds on the integral
@Ethan yes, but you cannot substitute anything for x willy-nilly. for instance, $$\int_0^1xdx\ne\int_0^1x^2dx,$$ even though $x^2$ traces out the same domain for $x\in[0,1]$.
I changed
the upper and lower bounds
you need to review your basic calculus. when you do a u-substitution, the dx may not be the same as du; it will generally instead be dx=(blah)du.
I didn't do u substitution
@Ethan you put 1/x in place for x in the integrand. just because you called the second letter the same as the first does not mean you did not use substitution.
05:14
@Ethan Man, I do not intend to be an asshole, but you should really learn your calculus before trying to get into Analytic Number Theory. Not only because of this ongoing discussion with anon, but because you build up yourself as a math person by first getting into such topics.
notice that we should expect the second integrand ln(x)M(x) to be wildly divergent
@anon not neccisarly the mertens function changes sign offten
@Ethan How do you extend Merten's function to non integer values?
how often is M(x) within [-1/ln(x),1/ln(x)]? very rarely, or I'll eat my shoe
@PeterTamaroff the floor function is implicit when extending counting functions to noninteger values
@anon Oh, that is soooooo boring.
How does one input Merten's function in W|A?
05:21
@anon $$\frac{\pi^2}{6}=\int_{0}^1\frac{\ln(x)}{1-x} dx=\int_{0}^\infty \frac{x}{e^x-1} dx$$
it seems to work here?
That is correct.
@Ethan But you're neglecting what anon said before.
I will explain.
ok
Bah. Too much latex. I will use MSE's preview.
@anon Hahahaha.
05:27
@Ethan i think anon is right (about your first question), the integrands differ by a factor of x^2, not sure about the definite integral
when does a function satisfy $\int_{0}^1 f(x) dx=\lim_{n\to\infty}\frac{1}{n}\sum_{k=1}^nf(\frac{k}{n})$
the key is the negation
(after substitution u=1/x and changing of the limits)
@Ethan whenever the left exists
@Ethan This happens for integrable $f$.
05:28
err
if one of them exists
are they equal?
i don't think they exist
what do you mean?
@DanBrumleve are you talking about the previous integral involving the mertens function?
@Ethan The limit might exist for nonintegrable $f$.
i mean both M(x) and -M(x) are at least O(sqrt(x)) so the second integral probably fluctuates in sign
For example, if you take the indicator function on the rationals, that limit evaluates to $1$.
05:31
@DanBrumleve the sum exists
We want to obtain $\displaystyle\int_0^1\frac{\ln(x)}{x-1}dx=\int_0^\infty\frac{v}{e^v-1}dv$. The data of our substitution is

$$u=\ln x\iff x=e^u \\ du=\frac{dx}{x}\iff dx=e^udu \\ x=0\iff u=-\infty \\ x=1 \iff u=0$$

So we have

$$\int_0^1\frac{\ln(x)}{x-1}dx=\int_{-\infty}^0\frac{u}{e^u-1}(e^udu).$$

Using the substitution $v=-u$, we have

$$\int_{-\infty}^0\frac{u}{e^u-1}(e^udu)=\int_{\infty}^0\frac{-v}{e^{-v}-1}e^{-v}(-dv)=\int_0^\infty\frac{v}{e^v-1}dv.$$
@DanBrumleve $$\sum_{k=1}^n\ln(k/n)M(\frac{n}{k})=\psi(n)-\ln(n)$$
@Ethan What is $m$?
i was referring to what you wrote as $\int_{1}^\infty \ln(x)M(x) dx$
I see no $m$ on the RHS.
05:33
I say explicitly what v is.
@Ethan Ethan, a function satisfies that if, and only if it is integrable over the said domain (i.e. $[0,1]$)
To conclude, Ethan, it just so happens as a fluke that blindly using a substitution and forgetting the differentials works out the same as using valid substitutions. In general, you will not get so lucky, for instance with $\int_0^1xdx\ne\int_0^1x^2dx$ as I already mentioned, or say $1/2=\int_0^1xdx\ne\int_1^\infty\frac{1}{x}dx=\infty$.
@Ethan That is just formal manipulations, if you wish.
@Ethan, have you taken calculus?
05:36
sounds rather evasive
@anon I watched videos on khan academy
@anon What did he say?
@Ethan STAHP REMOVING!
@PeterTamaroff "I don't know isn't that a broad subject"
@anon your second example invalidates my explanation
cool beans
05:38
@anon OK, I have proven the easy part of the lemma.
the primitivity condition using existence of a subset A s.t. blah?
@anon So can I conclude then that $$\int_{0}^1\ln(x)M(\frac{1}{x}) dx=1$$, sense $$\lim_{n\to\infty}\frac{1}{n}\sum_{k=1}^n\ln(k/n)M(\frac{n}{k})=1$$
Yes. But I proved imprimitivity $\implies$ existence, now I'll prove the converse.
@Ethan There is a problem.
What I told you works with proper integrals.
05:42
@Ethan I think so, but it is not immediately clear how to show the integral exists. (One particular type of Riemann sum having a well-defined limit does not in general establish that every possible Riemann sum associated to successively refined partitions necessarily converges or converges to the same value, and this latter stipulation is necessary for the integral to exist, except in a C.P.V. sense)
@anon Yes, as I just told him $\int_0^1 {\bf 1}_{\Bbb Q}$ fails to exist while the limit exists.
@Ethan So really, you cannot.
M(1/x) will of course be discontinuous, but my intuition is that it is not so pathological that there isn't some high-powered analysis theorem that fixes everything
@anon $$\frac{1}{2\pi i}\oint_{c}\frac{1}{x^s\zeta(s)s}ds=M(\frac{1}{x})$$
@Ethan What does that mean?
it's a contour integral
05:44
@anon Yeah, I know.
mellin transform, in particular
I was wondering if Ethan could explain.
I don't understand it, but someone else might
(Not that I know anything about contour integrals)
Why do you type it here if you don't know what it means?
05:45
because others might
@anon Does $M(x)$ approach zero fast enough when $x\to \infty$?
@PeterTamaroff M(x) does not even approach zero ...
@Peter, even on RH you only get $M(x) = O(x^{1/2 + \epsilon})$
It is easy to show that M(x) is nonzero for infinitely many values of x.
@Sanchez they are equivalent in fact
05:47
@anon Forget I asked.
also one can use the summatory liouville function instead of M if one prefers
no
@anon One thing.
The LEMMA reads "either $\text{blah}$ or $\text{blah}$" so I should treat those cases separately, yes?
@Dan, sure.
That is, we can have $gA=A$ or $gA\cap A=\varnothing$.
05:49
@PeterTamaroff this is inside a quantifier on g, don't forget
L(x) = O(x^(1/2+\epsilon)) is my favorite way to look at RH is all :)
So, $\forall g\in G~\big(\,(gA=A)\wedge(gA\cap A=\varnothing)\,\big)$
@anon The claim is $\exists A$ such that $\forall g\in G$, $gA=A$ xor $gA\cap A=\varnothing$.
oh, right, xor
well, it's not like both can be true :)
@anon Hehe, I was hoping there was some slick idea where we could track the cases in parallel =P
05:51
Anyway. So, I take it you've shown that if the action is imprimitive then such an $A$ exists?
Yes.
It is not hard, as you said.
For the converse, show that the partition of G into left cosets of A is stabilized by the action. First of all you have to show the left cosets of A are all disjoint and their union is all of G.
Note that we are not assuming A is a subgroup, even though we are speaking of its left cosets.
@anon Yes.
@anon But cosets wrt to the action, so that they partition $S$, right?
oh. is S our overgroup? I thought it was G.
No, $S$ is the set and $G$ acts on $S$.
I thought we were consdiering partitions of $S$. Let me think about what you're saying.
Don't hint me further.
05:55
oops, I am thinking of A as a subset of G. derpola cola. but my line of reasoning still works.
my reason for thinking so is that the summatory Liouville function translates more naturally to a uniformly-distributed binary sequence than the ternary-valued Mertens function.
@anon Oh, OK. So $A\subsetneq S$, and we want cosets wrt to the action, yes?
or translates, yes
@DanBrumleve you don't think the mobius function is a bit more 'natural'
@DanBrumleve the mertens function evaluated at n, also has a nice exponential sum over a farey sequence of order n, $$M(n)=\sum_{q\in F_n}e^{2\pi i q}$$
depends on the context i guess
05:59
I think farey sequences are 'natural' in some sense
i like to think of the moebius/liouville functions as analogies for "randomness".
i that sense a binary equidistributed value is more convenient than a ternary one where one value has probability 6/pi^2
Im pretty sure that formulas useless for computations tho, lol, since a farey sequence typically has around $\frac{3}{\pi^2}n^2$ terms
next person that uses sense instead of since has to sit in the corner
I wouldn't doubt it :)
Hm.. I might have to try to find a meaningful sentence in which sense can be interchanged with since.
How about when you are simply referring to the words "sense" and "since" themselves? Gasp!
06:05
since we're on the subject in exactly what sense is a computable set diophantine is it just since a program can be encoded as a diophantine equation in some sense or is it in some other sense?
@anon Indeed. In one history, I am to sit in the corner. In the other, I will remain sitting elsewhere.
@Sanchez Did you have a counterexample in mind for math.stackexchange.com/questions/254455/… this question? I was reading it again and it seems like if you assume the $u_i$ are the distinct roots of $f$ it may be true.
@peoplepower You feline, you.
@anon Could you reword your hint? Recall: $G$ is assumed to act transitively on a set $S$ of more than one element and we're assuming a proper subset $A$ exists with $gA=A$ (then comes the other hyp)
You mixing up $G$ and $S$ got me confused.
@PeterTamaroff We're assuming there exists a proper subset A such that for all $g\in G$, either $gA=A$ or $gA\cap A=\varnothing$.
06:08
Oh, me dumb dumb.
Moving on.
I want you to show that $\{gA:g\in G\}$ is a partition of $S$ that is stabilized by the action.
of $G$????
yes
for the first part, you must show that it is a partition of S. this is the easier part.
I'm confused. The defintion of primitivity talks about partitions of $S$.
next person that uses $G$ instead of $S$ has to sit in the corner
06:10
bob saget
@anon Soooooooooooooo?
"bob saget" = curse word everyone used in middle school.
I have edited my previous statements so that they are accurate.
@anon That is a comedian, apparently.
also true
How is "bob saget" a curse word?
06:13
"tourettes guy"
Still, nothing.
Bells be silent.
just nevermind it
@JSchlather, I'm not sure what I was thinking back then, but from the context I think I did assume all the $u_i$ are distinct.
Booooooooo!
@anon I wanna know your initial.
06:17
@anon :sad:
"there is a face beneath this mask, but it's not me. I'm no more that face than I am the muscles beneath it, or the bones beneath them"
@anon "There is an idea of a Patrick Bateman; some kind of abstraction. But there is no real me: only an entity, something illusory."
The full quote is better " There is an idea of a Patrick Bateman; some kind of abstraction. But there is no real me: only an entity, something illusory. And though I can hide my cold gaze, and you can shake my hand and feel flesh gripping yours and maybe you can even sense our lifestyles are probably comparable... I simply am not there. "
@Sanchez If I have two functions $\psi(x),f(x,y)$ such that $$\psi(s)=\int_{1}^\infty f(x,s) dx$$ for all s>a, and both $\lim_{s\to a}\psi(s)$ exists and $\lim_{s\to a}f(x,s)$ exists for all x, can I conclude that $$\psi(a)=\int_{1}^\infty f(x,a) dx$$
:8287798 I am not sure I understand the question. Something about free money?
heh :-)
you didn't
wait, maybe you did
06:22
Craigslist :)
@OrangeHarvester FUCK YEAH MAN. You're my new god.
it is a very old one and doesn't give my name though. not to mention I mentioned the ad in chat.
But I shall respect anon's privacy.
@Sanchez Yeah, I was trying to do something simple like $(x-\sqrt[p]{t})^p$. But I couldn't get it to work.
06:23
@PeterTamaroff :-) I have not found much else though.
@anon Oh, OK... I'll call you then and you'll never know what hit you.
@anon you are correct.
@OrangeHarvester I am no longer in reno, btw
@anon I know, I found your more recent one. The one where you are currently.
I don't have one for Omaha though, heh.
06:25
@anon I can show you if you want.
sure
@anon I shall use transitivity, by which for all $s\in S$ and $a\in A$ there exists $g\in G$ such that $ga=s$.
@anon check your mail. :-)
@OrangeHarvester I am now very concerned, since I did not post that ad.
@anon My sympathies.
06:27
@anon Ohh. I thought you did. I got your mail from chat sometime ago btw.
@OrangeHarvester It is a replica of what I posted in the reno ad. However the only way it would be reposted for the omaha location is if someone who knows me posted it.
@anon I see. So someone in your friends group I suppose?
(I did not post it just to be clear.)
@OrangeHarvester It's within possibility. How many other places have you found that particular ad?
@anon only one.
@OrangeHarvester How did you find it?
06:30
it doesn't give a date for when it was posted
@anon How the heck does the author expect someone to "observe" such a claim?
@anon If you have a limit outside of an integral, can you interchange them, and take the limit on the inside?
@Ethan Not always.
@Ethan under some conditions
@PeterTamaroff searched for email.
06:31
And it really depends on what kind of limit you're looking at.
@anon what conditions?
@Ethan Can you be more specific? What are you looking at?
uniform convergence
Well, but maybe he's looking at say $$\lim_{x\to a}\int_D f(x,s)ds$$
$$\lim_{s->a} \int_{1}^\infty f(x,s) dx = \int_{1}^\infty \lim_{s->a} f(x,s) dx$$ and $\lim_{s\to a} f(x,s)$ exists, and is defined for all x
06:32
Spot on! I deserve a cookie!
@Sanchez I posted an answer after deciding it was true.
@JSchlather, I actually think you are right - not sure what I was thinking.
@Ethan Let me think about it.
@Sanchez can u help me with ^, conditions that allow me to interchange the limit and integrand
@Ethan, sorry, I don't want to think now.
06:39
@anon Is it possible to have a definite integral that doesn't exist but doesn't grow with out bound either, like the sum $\sum_{n=0}^\infty (-1)^n$
@anon Done with that. Moving on.
hmm, I can think of one other person who could have posted it, who doesn't know me, or at least only met me once
I will be contacting the administrators anyway for further information
@Ethan that sum does diverge
but consider eg $\int_0^\infty \sin(x)dx$
ahhh
nice
not thinking lol
@Ethan There is a limiting value of functions which aren't Riemann integrable--that's why we move to Lebesgue integration. If you're considering the Lebesgue integral you have many, many more tools open to you.
@Ethan Divergence means literally not convergence.
There is no inbetween.
06:42
oh
Although some people like to use oscillation, as a sub category of divergence.
are there any general techniques that can be used to show a definite integral exists?
squeezing
Well, there are depending on your integral.
06:43
@Ethan You mean just a general integral, or do you mean a limit of integrals/integral of limits?
I can bound the integral between a sum?
11 mins ago, by Ethan
$$\lim_{s->a} \int_{1}^\infty f(x,s) dx = \int_{1}^\infty \lim_{s->a} f(x,s) dx$$ and $\lim_{s\to a} f(x,s)$ exists, and is defined for all x
@Ethan Under some monotonicity conditions.
Under what conditions
is that true
@Ethan If you pass to proving this for all sequences you have the entirety of Royden open as a possibility
wut
06:45
So, you're trying to prove something about a limit involving the continuous variable s
yes
move instead to trying to prove that you can switch for arbitrary sequences convering to a
converging*
err
let me just tell u the specific problem
In particular if $f(x)$ is any positive function that is monotone decreasing, then $$\sum_{k=2}^n f(k)\leq \int_1^n f(x) dx\leq \sum_{k=1}^{n-1} f(k)$$
06:48
Thus $\sum f(k)$ exists $\iff$ the integral does.
@anon
eh?
What we proved in the lemma is that if such $A$ exists then $S=\bigcup_{g\in G}gA$, right?
I have a Mellin transform for a function $\phi(s)=\int_{1}^\infty\frac{f(x)}{x^{s+1}}dx$, valid for all Re(s)>1 , and, $$\frac{d}{ds}\phi(s)=\int_{1}^\infty \frac{-\ln(x)f(x)}{x^{s+1}}dx$$ both $$\lim_{s\to 1} \frac{d}{ds}\phi(s)$$ exists, and $$\lim_{s\to 1} \frac{-\ln(x)f(x)}{x^{s+1}}=\frac{-\ln(x)f(x)}{x^2}$$ exists and is defined for all $x\in [1,\infty]$, can I say that $$\lim_{s\to 1}\frac{d}{ds}\phi(s)=\int_{1}^\infty \frac{-\ln(x)f(x)}{x^{2}}dx$$
@Ethan Don't you want $\int_{\geq 1} x^{s-1} f(x)dx$?
@anon Nevermind.
that doesn't make any sense. if the action is transitive then the union of translates is the whole space.
welp
06:56
@anon OK, let me write down what the author says.
Because he actually proves that observation and I just didn't notice...
I'll just take a picture. Too much writing.
Uploading from phone to Dropbox. Might take a while...
I lost 2 reputation, and I don't see why
@DominicMichaelis You either got downvoted or unaccepted a question.
what does the reputation tab on your profile say Dominic?
Or maybe you're mixing up meta and main rep?
07:03
normally I see -2 and then the answer or something like that
no i don't have meta reputation
Yes, you do...
@DominicMichaelis Did you downvote anyone?
@DominicMichaelis If you downvoted twice, you lost 2 rep
meta rep = main rep, unless we're talking about SO.
@anon may be he lost an edit for a question which was deleted. @DominicMichaelis Try checking the show deleted questions checkbox.
07:05
@anon The author assumes the existence of this $A$ set and then says let $B$ be the complement of $\bigcup_{g\in G} gA$ in $S$.
@PeterTamaroff Meta does not have a rep iirc. Your site rep = your meta rep.
@OrangeHarvester Well, I mean the update lag might make em vary.
okay.
@Ethereal SO has things differently?
meta rep is somewhat delayed I think
@OrangeHarvester Yeah, on meta.SO and SO you have different reps.
@anon that's right
07:07
well i try to be productive see you later
@anon That's the proof.
He says that the partition is actually the cosets $gA$s plus $B$
evidently the author was ultraderping. just ignore the B and carry out the rest of the proof.
Must. Not. Ragequit. And. Watch. Bear. Grylls. Eat. A. Zebra.
Must. Not.
@anon Sure? I am losing faith in Jacobson, man.
yup
Bill Dubuque recommended it.
@anon Just out of curiosity, I might ask on main. Not that I am not convinced, though.
But maybe someone can tell us what this guy was thinking.
Maybe Bill himself.
07:14
My guess is J momentarily forgot transitivity was present.
Oh, prolly. Because then it would also work. I mean, all what he says although repetitive is true, since $B=\varnothing$.
right
Perhaps he is not using the hypotheses of the theorem at first, since that result does fit in greater generality.
that also makes sense
@peoplepower That seems to be what anon said. Could you expand on that? What generality?
07:24
pp is saying that the proof shows the existence of such a subset A implies that the action is imprimitive, even without first assuming the action is transitive.
@peoplepower You're saying what anon is saying?
@PeterTamaroff My point is that it is probably not a matter of forgetting. He is providing a more general statement outside the confines of transitive actions. Notice that he states the hypothesis "$G$ acts imprimitively on a set $S$" rather than on the set $S$.
@peoplepower Oh, but in the theorem's enunciation he also uses "a set".
@PeterTamaroff It is not the same $S$.
But I do get your point, he disregards transitivity first.
07:28
The reason why he does not state that as a separate lemma is perhaps that Jacobson does not find it interesting by itself, or it might just be a matter of style.
@peoplepower I see.
Faith in Jacobson restored.
lyj
lyj
does anyone know if it would be appropriate to ask about a specific reu on stackexchange, and if not, where it would be appropriate to do so (and actually receive an answer)?
lyj
lyj
research experience for undergraduate
*s
you want to talk to the academic people at uni about that most likely
lyj
lyj
07:32
i'm not sure most professors know that much about reus though
07:46
@anon Nice. 13 exercises and then come the Sylow Theorems.
It is almost 5 am
I wont sleep today.
@lyj which REU?
@peoplepower You gone?
@lyj ?
@PeterTamaroff Still here.
@peoplepower What's your time?
07:55
3am
I see. It is 5am, so I discard sleeping. It will mess up my biological clock...
I'll solve some problems
Like showing the conjugacy class of $\gamma=(12\dots n)$ consists of $(n-1)!$ elements.
@PeterTamaroff gstab(x)g^(-1)=stab(gx)...apply that and the fact that gamma{1,...,n}={1,...n}, and orbit-stabilizer
better exercise: show that the conjugacy classes are precisely the permutations of given cycle types, in any symmetric group (arbitrary cardinal size)
@anon do you know what you want to do?
I want to sleep. Rephrase?
07:59
@anon Are you having trouble with that?
@anon with respect to the only thing I would be asking you about, since you are just a username on a mathwebsite :)
I am a simple man. I want to do math. If I can get something that pays me to do something mathy, I will probably like it. I am not committed to any particular career path.

« first day (938 days earlier)      last day (4381 days later) »