« first day (3382 days earlier)      last day (1935 days later) » 

00:24
@TedShifrin I forgot to thank you for answering my question about the eisenstein criterion; thank you for clarifying.
 
2 hours later…
02:19
@Semiclassical You think you could help me with some Linearity of Expectation?
Suppose I start typing out random digits (from 0-9 inclusive) on a notepad, where the chance of my typing any certain digit key is equal. I need to find the least number of numbers I'd have to type out before I get at least one occurrence of the string "9845689".
 
2 hours later…
04:55
@LeakyNun Hello
@Jacksoja hallo
I have a smal question about how to factor this
@LeakyNun r^3+s^3+t^3 -3rst
plug it into wolfram alpha
am trying to prove that the arithmetic mean is larger than geometric
so there is no algebraic ways?
have you plugged it into wolfram alpha
04:58
yes about to
slow internet
@LeakyNun okay I see, but my question is if there is a way to factor such things
trial and error
I do not want to rely on program for doing things
okay thanks
it is not like the case of
quadratic and cubic
I mean this is a cubic
you can view it as a cubic in t
@LeakyNun yes I know, i was thinking of more a procedure to factor these things
maybe if there is a course about it
no idea
05:04
okay thank you !
 
4 hours later…
09:16
Hmm...
Can we have an orbit where it oscillate between a point at infinity and some finite point?
 
3 hours later…
11:49
Hi chat
I'm trying to compute the homology of the sphere in $\Bbb R^3$
To do that I'm dividing the cube $[0;1]^3$ in 48 tetrahedrons
By adding vertices in the middle of each edge, face and in the middle of the cube
And the tetrahedrons are the ones given by 3 points on one same face and the point in the middle of the cube
I've showed $H_3 = 0$ and $H_0 = \Bbb Z$
And now I'm trying to prove that $H_2 = 0$ and I don't really know how to go on from there
I need to prove that $\ker d_2 = im d_3$
I realised that to be in the kernel of $d_2$ you need to have a pair number of faces for each edge
But I do not know how to go from there
Never mind I got it
If $(z_n)=((z_{n, k})_k)$ is a Cauchy sequence in $\mathbb R^n$, then for $\epsilon \gt 0$ there is an $N \gt 0$ such that for all $n, m \gt N$ $\sum_{i=k}^{n} |z_{n, k}-z_{n, k}|^2 \lt \epsilon^2$
Then we have $|z_{n, k}-z_{n, k}|^2 \lt \epsilon^2$ for some $N$, but does this $N$ depend on $k$?
12:08
You're using $n$ for two different things which makes it confusing here
Use $d$ for dimension instead, for instance
The $N$ should not depend on $k$ since the sum of squares of modules is always greater than the square of the module of a specific element
Unless I'm misunderstanding what you wrote
@Astyx you can see that he mentions how $\{1, \dots , n\}$ to justify why the convergence is uniform..
I get that if it depends on $i$ there then you can just pick the biggest $N$ but I don't see why it should depend on it, I'm obviously missing something important here
And all the answers here seem to do the same thing: math.stackexchange.com/questions/147446/…
12:34
@FuzzyPixelz That's because they're not talking about the Cauchy criterion but about the existence of a limit
In $\Bbb R$ every Cauchy sequence has a limit
You've proved every coordinate projection of a Cauchy sequence in $\Bbb R^n$ is a Cauchy sequence in $\Bbb R$
But do they all share the same $N$, that's my question ..
Therefor every $x_{n,i}$ has a limit, say $x_i$, ie $\forall \epsilon >0, \exists N \forall n\ge N, |x_i - x_{n,i}|\le \epsilon$
There exists a common $N$ because $\{1,\dots, d\}$ is finite
You can chose the greatest as they claim
So they don't share the same $N$, can please explain it in more detail?
The reasoning gives you for all $1<i<d$ and $\epsilon>0$ an $N_i$ such that $|x_i-x_{n,i}|\le \epsilon$ whenever $n>N_i$
Right ?
Doesn't the "for all $i$" go after the $N$?
12:43
No
It's the same $N$ in the Cauchy criterion
But it doesn't have to be the same $N$ afterwards
Hence why I call it $N_i$ and not $N$
But didn't we just let $m$ go to $\infty$ ?\
Nope
Well I guess you could
But that would still use the fact that you only have a finite number of terms
How so? (I'm sorry for being so dense..)
You have to make sure that $$\sum_i |x_{n,i}-x_{m,i}| \to_{m\to \infty}\sum_i |x_{n,i}-x_i|$$
Which, when $i$ takes an infinite amount of values, is not obvious
Imagine taking $x_{n,i} = (1-1/n)^{i}$
So in either case the proof wouldn't generalize to $l^(\mathbb R)$ for example
12:53
You get $x_{n,i} \to_{n\to \infty}1$
But $\sum_i |x_{n,i}-1| = \infty$
(for $n\ne 0$)
@FuzzyPixelz yup
Alright thank you
No problem
Is it clear now ?
Well... what if we say that the sum is bigger than each term and then let $m \to \infty$
Yeah I guess that could work
13:11
hi
The number of solutions of $\pi(x)^2+\pi(y)^2=8000$ is $\pi(1000)$
13:35
There is an ML Inequality, which tells about the maximum value of a complex line integral, It is of form $ \oint_C f(z) dz \le ML$ Where $M$ satisfies $|f(z)| \le M$ on the curve $C$ and $L$ is length of the curve.
My question is how to find $M$?
13:49
I'm not sure what answer you're looking for
The answer obviously depends on the function f and the curve C
14:07
Does anyone know if this is true:
$x^2-(n^2-3)y^2=-1$ with $3\not|n$ is soluble if and only if there are integers $p,q,r$ so that $n=\frac{3p^2+q^2-r^2\pm 2r^2}{2pr}$
1
A: Why are the minimal solutions of the Pell equation $x^2-(n^2-3)y^2=1$ larger than average when $3 \not | n$?

SophieI think this is explained by the fact that $x^2-(n^2-3)y^2=-1$ with $3\not|n$ is soluble if and only if there are integers $p,q,r$ so that $n=\frac{3p^2+q^2-r^2\pm 2r^2}{2pr}$. Proof for the forwards implication: We have $x^2+2y^2\equiv 2 \pmod 3$ which implies $x\equiv 0$ and $y\not\equiv 0$ s...

14:47
@MikeMiller Yeah, but how to find that, any method?
@krauser126 yikes
hi guys, I was thinking, let's say you have a measure space that is not complete, then let $N$ be a measurable set with zero measure, and let $M\subset N$ be a set that is not measurable. Then can I say that if $f:=g+1_{M}$, then $f=g$ a.e.?
mainly because I cannot say that they differ in a zero-measure set, because $M$ is not measurable
maybe start by considering a simpler problem---say, how you'd do the problem if the desired sequence was 2 digits instead of 7. (I don't think the actual set of digits in the exit code matters, just the number of digits)
can I still say that they differ UP to a set with zero-measure?
15:04
@MikeMiller Hey you around?
@AjayMishra To be more precise, you can take $M$ to be $\sup_{z\in C} |f(z)|$.
15:25
If we $(M_i, \mu_{ij})$ is a direct system of modules, and each $M_i$ is isomorphic to $N$, then is $\varinjlim M_i$ isomorphic to $N$?
*the isomorphisms commutes with $\mu_{ij}$ in the obvious way.
What I mean is $\phi_i : M_i \to N$ are isomorphisms and $\phi_i = \phi_j \circ \mu_{ij}$
I am trying to prove this using the universal property of direct limit.
(I just realized it follows very easily from the SES $0 \to M_i \to N \to 0$). But, still I would like to prove it using the universal property.
*ES
15:43
@feynhat if you have a module $K$ and maps $\psi_i:M_i \to K$ such that $\psi_i = \psi_j \circ \mu_{ij}$ for all $i,j$, then define $\psi:N \to K,\psi=\psi_i \circ \phi_i^{-1}$ (for any $i$, it doesn't matter, all will give the same map), then check that this is the unique map $f:N \to K$ such that $\psi_i=f \circ \phi_i$ for all $i$, so $(M,\phi_i)$ satisfies the universal property of the direct limit
Hi all.
@Rscrlli the usual definition of "almost everywhere" is that there is a set of measure zero, s.t. the statement holds on its complement, or equivalently that the set on which it doesn't hold is contained in a set of measure zero, so yes, you can say that
this is also equivalent to the induced outer measure of the set on which it doesn't hold being zero
16:34
> $f : N \to K$
You meant $\psi$, right?
I didn't realize if $\phi_{i}$'s are isomorphisms then so are $\mu_{ij}$'s.
Thanks.
What does the ultraproduct of all finite groups look like
BIG
16:51
thank
@AkivaWeinberger depends on the filter I guess
@feynhat I meant that every map $f:N \to K$ such that $\psi_i = f \circ \phi_i$ for all $i$ is equal to $\psi$
As long as your ultrafilter is not principal it's easy to see that it is an infinite group by Los's theorem, but it actually has cardinality $\mathfrak c$
@AlessandroCodenotti I think every theorem (first-order logic, $(e,\cdot,{}^{-1})$) that's true of that group is true for all sufficiently large finite groups
Not sure though
17:22
Question for you all:
Assume that you have a beta prior Beta$(\alpha,\beta)$ over some unknown probability $p$. I know $p$ but you do not.

Now assume that you draw $n$ samples are drawn from a Bernoulli distribution with parameter $p$, that is, $x_i\sim$Bernoulli$(p)$, $i=1,\ldots,n$. The samples are ***hidden from me***, I only know that $n$ samples were drawn.

I know that your posterior is a beta distribution, but since I don't know the samples (only the counts) I don't know what your posterior is. What is my best guess of your posterior on the probability parameter $p$?
17:37
@BalarkaSen what have you been up to?
@AkivaWeinberger Sure
0
Q: Riemann Hypothesis is about all arithmetic functions, not just $1(n) = 1$.

Shine On You Crazy Diamond$$ f(s) = \sum_{n \in \Bbb{N}} \dfrac{\phi(n)}{n^s} = 0 \\ g(s) = \sum_{n \in \Bbb{N}} \dfrac{\psi(n)}{n^s} = 0 \implies f(s) - g(s) = \sum_{n\in \Bbb{N}}\dfrac{\phi(n) - \psi(n)}{n^s} = 0 $$ Thus, for a fixed element $s \in \Bbb{C}$, the set of all coefficient functions $\psi : \Bbb{N}\to \Bbb{...

What do you guys think on this one?
17:54
@BalarkaSen sort of
@AjayMishra I don't know what you're expecting to get as an answer. This is like asking "Is there a method to find the derivative of a function"
Yeah and it's gonna depend on the function
18:10
@MatheinBoulomenos Okay.
Imagine being a Radon-Nikodym derivative.
Would you be content with yourself?
Ever?
 
3 hours later…
21:36
Is it just me or is the arXiv offline?
It seems to be
Are there other ways of computing homology than thinking in terms of simplicial complexes ?
I'm sure there's plenty of tools and tricks, Mayer-Vietoris comes to my mind
I'll look into that, thanks
21:52
Okay, I've got a good one for you now
what if you sum over $\Bbb{Q}^{\times}$
instead of $\Bbb{N}$ in a Dirichlet series for Riemann's zeta function?
0
Q: Dirchlet series over $\Bbb{Q}^{\times}$ instead of $N$.

Shine On You Crazy Diamond$$ f(s) = \sum_{q \in \Bbb{Q}^{\times}} q^{s} $$ Then the set of elements $G_s = \{q^{s} : q \in \Bbb{Q} \}$ is a subgroup of $\Bbb{C}^{\times}$. $|q^{s}| = \sqrt{ q^{s} q^{\overline{s}}}= \sqrt{q^{\overline{s} + s}} = q^{\text{Re}(s)}$, so that we can say that the series converges absolutely wh...

22:13
@ShineOnYouCrazyDiamond what is QQ^X supposed to be?
22:44
usually that would be the units in $\mathbb{Q}$

« first day (3382 days earlier)      last day (1935 days later) »