« first day (2507 days earlier)      last day (2810 days later) » 

15:00
I'm sorry.
Nothing to apologise for :P
I want to relativistically punch something
Punch a light beam :P
15:01
Punching something at that speed will probably break whatever you're punching. Your arm, too.
+ They never punch back
Part of what's going on here is that, in relativity, one combines the scalar energy and the vector momentum into a combined energy-momentum 4-vector
namely, $p=(E/c,\vec{p})$.
Correct. Even the very definition of a "straight" line gets bent out of shape near the speed of light. A straight line is defined as the path that light travels.
"A straight line is defined as the path that light travels."
a straight line is just a pair of points really
15:09
Never let a physicist define something, I guess.
If you do a rotation, then $|\vec{p}|$ is left unchanged.
@user685272 That's more general relativity than special, right?
If you do a boost (something that changes the speed of your reference frame) this isn't true. But it remains true that $p^2=(E/c)^2-(\vec{p})^2$ is left unchanged by such a transformation.
15:11
And also means that, once inside a black hole, the center is in a straight line in all directions…
This is analogous to the fact that while two observers need not agree as to the time $\Delta t$ and the distance $|\Delta \vec{x}|$ between two events
Special relativity is a special case of general relativity @AkivaWeinberger ie without accelerated frames of reference
@Semiclassical Ah, so you make $(E/c\cdot i,p_x,p_y,p_z)$ the vector? (And then ignore the $i$ probably)
the $i$ shouldn't be there, no.
that was a typo.
the two observers will agree on the space-time interval $\sqrt{(c\Delta t)^2-|\Delta \vec{x}|^2}$.
For some reason the arrow isn't appearing over the $x$ for me
Test: $\vec x$
15:14
Weird.
Yeah, it's offset to the right
$\bar x$ $\leftarrow$ That's fine
o.O
looks fine to me
O.o
me too
$\overset{\to}{x}$ I wonder if this looks weird
15:16
It looks just like \vec for me, which also looks fine...
As a test of this, suppose I consider two observers. One is in a rocketship moving from the earth at a speed $u$.
The other is at rest on earth.
Maybe it's your browser's javascript handler :/
(We usually say our relativistic things are "spaceships", as if that makes it any more plausible :P )
And we like to stuff people in those spaceships to get "paradoxes"
Better hope there's two medics on board, then
(Obvious pun is obvious)
15:19
Some time after the rocket launches, the person on earth sends two pulses of light with a time $T$ in between.
But by the time the second gets to where the ship was the ship has moved?
Hmm, this isn't quite what I want.
I want both of the relevant events to occur on earth at the same position
They're the same time to the light beam
I think I'll change my setup a bit.
Some time after the rocket launches, the person on earth tosses a ball in the air and catches it at the same height in time T.
From their perspective, there was no separation in space between the two events. So $\Delta t=T, \Delta x=0$, and the spacetime interval is $cT$.
From the perspective of the person on the spaceship, though, Earth would have moved a certain distance $a$ farther away in the time that the ball rose and fell.
Spacetime interval is $\sqrt{(c\Delta t)^2-|\Delta\vec x|^2}$, yeah?
15:24
Right.
And for that observer one should have $\Delta x = u \Delta t$ where $u$ is the speed of the craft.
So they'd have a spacetime interval of $\sqrt{(c \Delta t)^2-(u \Delta t)^2}$.
For these to match, we need $(c^2-u^2) (\Delta t)^2 = c^2 T^2\implies \Delta t = \dfrac{T}{\sqrt{1-u^2/c^2}}$
i.e. the time for the ball in flight for the moving observer would be dilated by a factor $\gamma = 1/\sqrt{1-u^2/c^2}$.
So time-dilation for the moving observer is equivalent to the space-time interval for the two observers being identical.
In more formal terms, this spacetime interval is a Lorentz invariant.
The point for the 4-momentum is that $\sqrt{(E/c)^2-|\vec{p}|^2}$ is also a Lorentz invariant.
Semiclassical do you know of any good tutorials for convolution, I dont know what to do with the bounds of integration here $f * f * f$ where $f$ is the characteristic function of the unit inverval
i imagine i do it in two steps
tbh the main thing I know about convolution is "it's nice after Fourier transforming"
15:35
But one should still be able to write down the integral.
For instance, we have $(f* g)(x)=\int_{-\infty}^\infty f(x')g(x-x')\,dx'$
well here I'm just doing it over the whole real luine
which isnt that much because the support is so small
hmm, you're right.
So if $f(x)=g(x)=\chi_{[0,1]}$
ya
$= h(x)$
Then that's $(f*f)(x)=\int_{0}^1 f(x-x')\,dx'$
oh good call
didnt see that
15:39
Alternatively, $(f*f)(x)=\int_{x-1}^{x}f(x')\,dx'$
I think that's equivalent.
I want $x-x'=1$ and $x-x'=0$ as the endpoints.
@SteamyRoot So I downloaded SumatraPDF for my new OS :P
So yeah, $x'=x-1$ and $x'=x$
@BalarkaSen Ha :P (What OS?)
windows 10
If $x-1>1$ i.e. $x>2$ then $f(x')=0$ over the region of integration. Same with $x<0$
So we can assume $0<x<2$.
15:41
i'll buy that
@BalarkaSen Repent, ye sinner, and install TempleOS!
Since $f(x')=1$ wherever it's supported, we should basically be looking for the overlap between $[0,1]$ and $[x-1,x]$
so it's just ... $\chi[0,1]\cdot t$?
Not quite. If I imagine changing $x$ smoothly so that $[x-1,x]$ slides along the $x$-axis
Then I'll first start with no overlap. But once $x=0$ they'll begin to overlap and that will increase until $x=1$.
so it's a rectangle that overlaps with ... a rectangle?
15:43
At that point, it'll start to fall off again as the right side of $[x-1,x]$ leaves $[0,1]$
@SteamyRoot Really?
or a square in this case
so it's $t$ for part of the time and $-t$ for another part of the time?
It's decreasing like that after, yeah.
@BalarkaSen Heh, no. Unless you're as crazy as the guy who programmed that, which I'm pretty sure you're not.
But it should still have overlap = 1 when $t=1$.
So if I'm reading that right you'd have $(f * f)(x)=x$ for $0<x<1$ and $(f * f)(x)=1-x$ for $1<x<2$.
15:45
I am not super familiar with TempleOS but it basically looked like a useless thing.
so it's an isosceles triangle with altitude 1 at 1 and corners at 0 and 2?
Right. A triangle map
with a religious theme
I think the way to check this is to assume $0<x<1.$
Well, in a certain way it's impressive in the sense that it's a working operating system.
15:46
In that case, you've got $x-1<0<x<1$.
So the overlap is just $x$.
By contrast, if $1<x<2$, then $0<x-1<1<x$.
So the overlap is now $1-(x-1)=2-x$. So I was a bit wrong.
okay I understandc this geometrically
It matches what you said, though.
but dont see how it comes from the integrand
Yeah, it's a bit of a pain.
But, it can do pretty much nothing. No network support because it'd take too many lines of code to implement a proper browser.
15:47
If $1<x<2$, then $\int_{x-1}^xf(x')\,dx$...
@SteamyRoot lol
There is the God-Approved™ 640x480 16-bit display, though!
Ah, okay. We can split that up into $\int_{x-1}^1+\int_1^x$.
Where the fact that $1<x<2$ implies that $0<x-1<1<x$.
15:49
which is the support
Oh, and if you go to university next year, check if your uni has an agreement with Microsoft Imagine (formerly dreamspark)
Hello everyone
If you're a windows fan, you may be able to get free copies of Office or Windows 10 or whatever there
Right. The second integral contains only points outside of $[0,1]$, so that goes away.
I like the look of Windows 10, but I dunno what it's good at. It seems fancy.
15:50
For the first, f(x')=1 for all points in $[x-1,1]\subset [0,1]$.
Windows 10 is like windows 7 imo
So therefore that's $\int_{x-1}^1 1\,dx = 1-(x-1)=2-x$.
All schools in Canada offer free windows upgrades and Microsoft word
So $(f * f)(x)=2-x$ for $1<x<2$.
My uni offers Microsoft Word downloads as well.
@SteamyRoot Thanks! I didn't know the existence of that thing.
15:51
I forget about windows updates.
@BalarkaSen Well, the "delayed start" for many programs should give way better boot times than whatever you had before.
so thats not what I said
the altitude is at $x = 0$?
its a right triangle?
No, it is at $x=1$.
That's only for $1<x<2$.
Convolution is essentially a continuous version of multiplying generating series
@Semiclassical How can you forget about them now that they decide to shut down your computer without giving you a chance to abort? :P
15:52
@SteamyRoot I forget if my uni offers them :P
Akiva we're doing generating functions starting today
so I dont know that yet :P
@Akiva Yeah it's like Cauchy? product. $c_n = \sum_k a_k b_{n-k}$
Probability course
For $0<x<1$, one instead has $\int_{x-1}^x f(x')\,dx'=(\int_{x-1}^0+\int_0^{x})f(x')\,dx'$
@Semiclassical Oh, in that way.
15:53
Yeah Cauchy
has anyone here ever measured the intensity of a light bulb that is partly "shielded" by the cover of the lamp?
Hey akiva how goes you
Gutentag sha
like this one
hi @Dodsy
how are ya
@SteamyRoot I see
Good thanks!
15:54
I had 7 before.
@Dodsy Guten shag
Where this is sensible since $x-1<0<x<1$ under these assumptiosn.
The first integral vanishes due to the lack of support, and the second just gives $x-0=x$.
So $(f * f)(x)=x$ for $0<x<1$, $(f * f)(x)=2-x$ for $1<x<2$, and it vanishes outside this interval.
@BalarkaSen Compared to 7 the boot should be a lot faster. The strain on your CPU and RAM may also have increased.
The graph of that is indeed an isosceles triangle with base 2 and height 1.
Anyway, it's definitely a good idea to google for what settings you can/should disable. By default, you're pretty much sending Microsoft everything you do.
15:56
which, heh, means it's an isosceles right triangle :)
But that's $(f*f)(x)$. Agreed? @GFauxPas
(even if you don't care about them knowing, it still takes processing power and network bandwidth)
Ah, good idea.
yeah sure
My laptop is doing weird things :/
At this point, it may be worth generalizing somewhat. Suppose we want $(f * \chi_{[0,1]})(x)$ for a generic $f(x)$.
15:58
It says there's a problem with the wifi adapter
That'll be:
@Dodsy Open device manager, see if things work, maybe try reinstalling or updating the driver
$$(f * \chi_{[0,1]})(x) = \int_{-\infty}^\infty f(x')\chi_{[0,1]}(x-x')\,dx'=\int_{x-1}^x f(x')\,dx'$$
interesting
So we'll always be integrating $f(x')$ over a specific interval $[x-1,x]$
Now, suppose we take $f(x)=(\chi_{[0,1]} * \chi_{[0,1]})(x)$ i.e. the one we just had.
16:01
I need a haircut
I look like a caveman
This is supported on $x\in [0,2]$, so if $x<0$ or $x-1>2$ then $[x-1,x]\cap [0,2]=0$ and the integral vanishes.
mathematicians are allowed eccentric haircuts Dodsy
get a big math beard
So the support of the result will now be on $[0,3]$.
@BalarkaSen I wish I could grow a beard
@SteamyRoot Restarted my computer and it fixed itself
16:03
whoa, hold on
Probably we need to consider 3 cases now: $0<x<1$, $1<x<2$, $2<x<3$.
I dont see it, let me turn some gears
okay got it :)
back in a bit
"limit notation" (or first principles) to find the derivative is bs.
how so Dodsy
16:11
I just hate doing the work :P
:P
just get it from proofwiki.org
hi i got something im now sure about . if $H: X \times X \to Y $ is continuous and i define $h : X \to Y$ by $h(x) = H(x,x)$ is it follows that $h$ is also continuous? i think yes but not sure how to show it.
huh , maybe because $A = \{ (x,x) : x\in X\} \subset X \times X$ so $ H$ restricted to $A$ is continuous .
oh it's not actually that bad
but I hate simplifying functions.
16:23
:P
The topology on the diagonal of the product might be different than the topology on $X$
what if they're both metric spaces
well they have to be or else continuity doesnt make sense
nevermind
@GFauxPas they can be topological spaces
but if $A\subset X$ and $f : X \to Y$ is continuous , then $f$ restricted to $A$ is continuous @AlessandroCodenotti
I dont know topology so there goes that
16:26
$X,Y$ topo. spaces, this claim stays correct
When a question asks you to determine the equation of a tangent
and specifies that it's parallel to another line
does that change the method at all?
Yes, but $h(x,x):\Delta\to Y$ will be continuous, with $\Delta=\{(x,x):x\in X\}$, the topology on $\Delta$ might be different from that on $X$
(Where $\Delta$ gets the subspace topology from the product)
@AlessandroCodenotti I thought "cofinal" is like the final section of the ordinal
I mean, I guess you'd then know that the slope is the same
like $\operatorname{cof}(\omega+3) = 3$
16:27
@AlessandroCodenotti i just need to say it is continuous :P
Continuity makes no sense if you don't specify the topology on the domain and that on the codomain
oh fuck
does anyone else have problems with microsoft word and equation editor?
i dont understand what you mean. i need just to justify why this function is continuous.
@Dodsy just say it
I just typed an entire document in TeXStudio :p
As a function from which space to which space?
16:29
I also know microsoft word equation editor
There goes all of my work
because I backspaced the equation..
@Dodsy Ctrl+Z?
No if you have an empty equation
i have $f,g :I \to Y$ that homotopic, so there is $H : I \times I \to Y$ s.t bla bla. now i define $h: I \to Y$ by $h(x) = H(x,x)$ , i just need to say $h$ is continuous
and you backspace
then the program with say not responding
and you have to force quit it.
16:30
oh... microsoft word is famous for that
that's why TeXStudio is much better
More like infamous :P
@Dodsy and then what happened?
And now all of my work is gone :C
@Dodsy why is it gone?
@AlessandroCodenotti so is it fine or there is something i need to check here?
16:32
you've typed all of your work without saving it once?
well it was probably 10 questions..
I just started at 12
you silly kids
if the document is small, then there's no impact as you can just type everything again
well I wrote a lot of math.
if the document is large, then you should have saved it
gosh darnit
So it's either all my fault
or it doesn't matter
16:33
It's fine if you want to say that it is continuous as a function with domain the diagonal, you need to justify it if you want to think about it as a function $X\to Y$
@Dodsy correct /s
@Dodsy and, switch to TeXStudio :D
or pencil and paper :P
-_-
yeah cause I'm in the stone ages.
Mar 4 at 15:52, by Alessandro Codenotti
that's cofinality. For an ordinal $\kappa$ $\text{cof}(\kappa)$ is the order type of the smallest cofinal subset of $\kappa$. Where a subset $Y$ of an ordered set $X$ is called cofinal if for every $x\in X$ there is an $y\in Y$ with $y>x$
I wrongly remembered the definition. You were right all along.
but I'm also right all along... you're just choosing a final segment of the ordinal, for countable ordinals.
Hence my confusion with $\aleph_\omega$, since the set chosen isn't the final segment...
@AlessandroCodenotti
@user685272 btw is that you skull?
16:39
You don't choose a final segment, but a cofinal one, one which is unbounded if you prefer
@AlessandroCodenotti for countable ordinals, that's the same as a final segment, right?
really?
Then who do I play chess with!?
@AlessandroCodenotti I'm wrong again. For $\omega^2$ there isn't a final segment. The set is $\{\omega n | n < \omega\}$
Mar 4 at 15:55, by Alessandro Codenotti
The cofinality of $\omega^2$ is $\omega$ since the subset $\{\omega n:n\in\Bbb N\}$ is cofinal in $\omega^2$ and has order type $\omega$
Sneaky pete.
The cofinality of every countable limit ordinal is $\omega$ I think
(For successor ordinals, is it $1$?)
16:44
@AkivaWeinberger yes
(Because $\{\alpha\}$ is cofinal in $\alpha+1$)
@AkivaWeinberger precisely
OK, thanks
The cofinality of an ordinal can't be of higher cardinality than the ordinal itself, so of course it'd be $\omega$ For countable limit ordinals
Well, cofinalities can only be $\omega_\alpha$ numbers, right?
@AkivaWeinberger why?
I'm all new to cofinality
I don't have a proof, but… you never get something with cofinality $\omega^2$, right?
16:47
If i have 4 real numbers specified through the relation a < b < c < d, and i construct the set U = { x real : (x-a)(x-b)(x-c)(x-d) < 0}
Because if $f:\omega^2\to\kappa$ is your unbounded function, you can do $n\mapsto f(\omega n)$
@AkivaWeinberger what about $\omega+1$?
Not sure how this argument would work in general
Well i do not know about you, but when i always tried to find an example set, it would never work, because the whole product has to be < 0
@LeakyNun Oh. Right… OK, an $\omega_\alpha$ number or $1$ :P
Or $0$, for the $0$ ordinal
@YannikK. What if $a<x<b<c<d$, for example?
16:48
@YannikK. what do you mean by "example set"? I thought the set is determined by the 4 numbers.
$(x-a)$ is positive, while $(x-b)$, $(x-c)$, and $(x-d) are negative,
@GFauxPas back
so $(x-a)(x-b)(x-c)(x-d)$ would have sign $(+)(-)(-)(-)=(-)$ (abusing notation slightly)
welcome back :)
So I think all of that suggests taking 3 cases: $0<x<1$, $1<x<2$, $2<x<3$.
Which I think corresponds to the following geometry:
16:50
You have five cases. $x<a<b<c<d$, or $x$ is between $a$ and $b$ (which I did above), or $x$ is between $b$ and $c$, or between $c$ and $d$, or after $d$.
@YannikK.
If I have my triangle graph of base width 2, and I want to pick some length-1 interval
would someone please help delete this?
-2
Q: Show that: $X^{(\mu \nu )}Y_{\mu \nu }=X^{(\mu \nu )}Y_{(\mu \nu) }$

Fernan13The left side is: $X^{(\mu \nu )}Y_{\mu \nu }=\frac{1}{2}\left (X^{\mu \nu }+X^{ \nu \mu} \right )Y_{\mu \nu }$ But the right side does not give me the same. Can you help me please?

the user is still refusing to respond and edit their question as I've demanded.
the question needs to go
Hey duck
sorry I ducked out last night
:)
@AkivaWeinberger where $\alpha$ is a successor ordinal
16:52
@LeakyNun You sure?
Oh, you're right.
I think. Hold on
wait semiclassical
@TheGreatDuck For reference, here's the policy on question deletion: math.stackexchange.com/help/deleted-questions
im going back to $\int_x^{x+1} \chi(t) \, \mathrm dx$
My impression is that a question like that which has been closed in the manner you've indicated will ultimately get deleted.
Yeah, you're probably right… @LeakyNun
16:53
@Semiclassical that's not relevant. The question is off topic. I told the user to either edit the question or their question will be deleted.
I get how that's $t \cdot \chi[0,1]$
> The cofinality of any ordinal α is a regular ordinal, i.e. the cofinality of the cofinality of α is the same as the cofinality of α. So the cofinality operation is idempotent.
having a hard time with the $2-t$ part
they probably won't bother coming back anyways
@AkivaWeinberger from Wikipedia
16:54
they're a new user
...well, that's not really your call. You can indicate that it's probably going to be deleted, but you're not in a position to dictate it
@AkivaWeinberger First why are you always putting a $ in front of a variable ? Well if x is for example smaller than a its also smaller than all others variables and the whole product is positive
@YannikK. Oh. Do you have MathJax installed?
There's a process in place for how such questions get deleted, and I imagine it will play itself out.
It's in the room description. It makes the math stuff format nicely.
16:54
@Semiclassical i was stating it as an indication. I just don't want to look like an idiot.
:p
For example, if I type $a^b$, it turns it into an actual superscript.
plus... I was posting it here to see if anyone could cast some delete votes. It's a pretty normal thing to do around here.
...so, because you made a declaration based on authority you don't have, you want to have the mods step in to make sure you don't look bad?
wat...
mods?
@LeakyNun OK, but why must $\omega_\alpha$ never be regular for limit ordinals $\alpha$?
16:56
mods is an exaggeration, come to think of it.
I'm referring to the casting of delete votes...
and dude
@AkivaWeinberger I'm having a conjecture that the cofinality of any omega subscript limit ordinal is, you know, omega
I was just sharing a post needing some delete votes
2much ordinals
nm I got it
16:57
@AkivaWeinberger OK i got it now, but back to my question. if x is for example smaller than a its smaller than all the other vars and hence the whole product is negative
@LeakyNun What about $\omega_{\large\omega_1}$?
@YannikK. Is it? It's a negative times a negative times a negative times a negative…
$(-)(-)(-)(-)$
oh yeah i mean positive xd
Yeah OK :D
so we want $(x \cdot \chi_{[0,1]} + (2-x) \cdot \chi_{[1,2]}) * \chi_{[0,1]}$
fair enough. i'm not sure what I can do as far as issuing deletion votes in any case.
16:58
oh hey the ordinal master is back @AlessandroCodenotti
@GFauxPas Yeah.
and luckily convolution is linear
I'm just going to call that first bit $g(x)$ to stay sane, though.
@AkivaWeinberger alright, that's $\omega_1$.
lol okay
16:58
Cofinality of $\omega_{\large\omega_1}$ is $\omega_1$ I think
I am slow at writing
@AkivaWeinberger I agree
@LeakyNun even for $\omega$ the subset of even numbers is cofinal but not final
Where I was going earlier is that $g(x)$ is supported only on $[0,2]$
agreed
@AlessandroCodenotti I see, thanks.
16:59
@AkivaWeinberger sounds right
Geometrically, there's a few possible cases for how an interval $[x-1,x]$ can intersect with that.
@AlessandroCodenotti Ah, that's a good example
Explains what "cofinal" is pretty easily

« first day (2507 days earlier)      last day (2810 days later) »