« first day (2883 days earlier)      last day (2052 days later) » 

12:04 AM
Nooo not automated ML!
 
drag and drop bro
you drag your dataset over to the platform and it handles everything else
soon data science jobs will be obsolete
better get that moolah before that happens
time's ticking!
tick tock tick tock
 
Stupid data scientists creating automated data science to make it harder to get a data science job
"Sorry we're cutting your job because the automated system you made is better"
Either that or a genius plot to make data science not be the big thing so they can get paid more
What I'd really be scared of is getting a job and it being my job to do the dragging and dropping
 
12:40 AM
Lol
 
12:52 AM
The new version of "Excel is great for analysis!"
 
1:24 AM
Hi all. Is anyone familiar with conformal gravity in three dimensions (via the gauging of the conformal algebra)?
 
rob
2:07 AM
@DikshitGautam Let's talk here in the chat room instead of in the comments under your question. I'll tell you a little about how the site works and then we can perhaps address your issue.
@DikshitGautam I think that I should have the chat permissions set up correctly for you now. I'd be happy to talk about what's happening with your question, either from the site mechanics side or from the physics side.
 
 
2 hours later…
4:32 AM
\o @rob
 
rob
@user2646 Hey there.
 
how are you?
 
rob
I'm alright. Yourself?
 
fine, thanks
how's life?
 
rob
@user2646 More complicated than I might like, but I'll manage.
 
4:37 AM
welcome @JohnRennie
 
@user2646 morning :-)
 
how are you doing this morning?
 
If a surface is "topologically a cylinder," does that just mean that you can find some differentiable parameterization of that surface in $S^1 \times \mathbb{R}$?
Actually it looks like it may not need to be differentiable. Just continuous and invertible
 
 
2 hours later…
6:34 AM
@enumaris You can cast a flag for that. I think the system prompt doesn't appear if more than two people have commented, or something like that
But I'll take care of this case now.
 
7:06 AM
\o @Loong
 
7:42 AM
@rob Thanks for giving my problem so much attention
 
 
7 hours later…
2:42 PM
@enumaris and @ACuriousMind - Thanks for your help yesterday. It got sorted.
@vzn Well, thanks. But over there in that conversation, I wasn't grappling with anything novel, just an interpretation of the good old version of the same itself. I was just trying to make sense of that.
 
vzn
3:07 PM
@TheDarkSide right. researchers are trying to grapple with it also and formulating novel povs in response (forced to by the facts so to speak).
aha, found it. Khrennikov has some ideas about this. Born’s rule from measurements of classical signals by threshold detectors which are properly calibrated vzn1.wordpress.com/2015/01/26/…
 
3:47 PM
@TheDarkSide No problem :)
 
hmmm
 
 
2 hours later…
6:49 PM
I think these recent proofs are obviously nonsense. Let us hope that we all live long enough to come up with similar nonsense, and that we could remain even half as jolly presenting it to the mathematical community
anyway, hello room. Is anyone here at the moment?
 
@BenNiehoff Hi there! Long time, no see. How've you been?
 
busy, I guess :P
 
Heh, it happens
 
but today I'm letting Mathematica do my work for me
it's been trying to Simplify something for about 8 hours
but it's not going crazy with RAM usage, which is somewhat miraculous I think
 
Well, I have no idea how Simplify works internally, but how do you know it's not just stuck in some sort of loop? :P
 
6:58 PM
because then the memory usage would be going crazy
 
Not if it is an intelligently designed loop!
 
it has something very big to simplify, but from watching its behavior in Task Manager I think it's doing ok
in the meantime I've been reading about how to save definitions to a file so I don't have to re-do this one again
 
I mean, you only get exploding memory usage from stuff like non-optimized recursion
Infinite loops that don't consume increasing memory are easy to write :P
 
and from exploring a very large problem space to search for an "optimal" arrangement of the symbols in an expression
while(1);
so what's new and exciting here, besides deriving the fine structure constant by renormalizing pi?
 
Anonymous
7:38 PM
@BenNiehoff Been rather boring lately
 
Anonymous
Apart from what you mentioned :P
 
I think Sean Carroll's response is more or less the first thing I thought of
 
Anonymous
Sean Carroll, what? Umm, have I missed something?
 
i.e., QED is not even fundamental, so it doesn't make sense to try to derive the constants that appear in the QED Lagrangian
Carroll has a response on his blog
 
Anonymous
7:41 PM
Link?
 
Anonymous
I never saw his blog
 
Anonymous
iirc
 
Anonymous
Ah, thanks
 
7:54 PM
Have you guys listened to his podcast?
 
who
 
Sean Carroll
 
I think the most charitable way to interpret a 'derivation' of the fine-structure constant would be to say: Suppose we ignore the role played by QCD and the weak force at high energy, and just focus on QED as a theory unto itself. is it possible that there's a version of QED (say, some UV-completion of it) such that what you'd get for $\alpha$ (in the limit of zero momentum) can be expressed analytically in some fashion?
 
I have not
 
However, that's a very charitable reading and I don't actually think it applies. The problem is that you'd still need to be recognizably talking about QFT
 
7:57 PM
@enumaris do you listen to/know of any good ML podcasts?
 
nope, the only podcast I've really listened to snippets of is Joe Rogan's lol
 
I'm working through Partially Derivative, which I've liked so far. I think they ended it in 2016 though
 
And, uh, there's no reason to expect that from what we've seen of Atiyah
 
I don't even listen to whole podcasts tho
 
The UV completion of QED is electroweak theory. Or, I should say, a UV completion, since there are infinitely many
 
7:58 PM
Ahh I listen to them when I'm driving or sometimes when working out
 
but QED by itself cannot be taken to high energies
 
@Semiclassical surely the fine structure of hydrogen should not be purely mathematical in nature and things like the fundamental electric charge should play a role?
 
well, in natural units you've got $\alpha=e^2$
so I don't think "knowing the fine structure constant" is different than "knowing the fundamental electric charge"
 
which seems like it shouldn't be derivable from purely mathematical considerations
 
it does seem pretty implausible
What I sketched above is about the most plausible way I could think for that to be meaningful
(And even that seems like a stretch)
 
8:02 PM
The only thing I can think of would be something geometry related......like normal modes or something...but then you'd have other boundary conditions instead
 
my knowledge of how UV-completions of QED work is pretty much nil, soooo
 
same here :D
 
so is mine :D
all you have to know is that pure QED has a Landau pole
and that many different theories in the UV can give you the same theory in the IR
 
too much beyond me
 
 
1 hour later…
9:23 PM
bam, posted another question to physics
second one in like a month
I'm on a roll
Qmechanic is so fast in his edits
hmmm...I edited the title of my question and the edited title appears in "questions" but not in "home"
somehow the two tabs are not going to the same place???
how does that make sense lol
oh, the "home" tab has refreshed it
 
vzn
10:11 PM
guys: (dont want to rain on the parade + stones/ glass houses etc, but...) has Atiyah published anything on physics? not to google drive? o_O
 
Ooh the Singularity
So black holes are AI right?
 
10:34 PM
@danielunderwood maybe...
 
They'd be the best model. Just toss all your training data in there and boom...no more problem to solve
 
only if it's your only copy of the training data
some manager would prolly ask you to go retrieve the data tho
 
10:57 PM
"I can't"
"Find a way"
"..."
or "we've partnered with this company that is selling us their black hole retrieval tool. They won't warranty it for being lost during use, but I think we'll be alright"
 
11:12 PM
"do some voodoo magic on it"
 
Seems about right. Maybe we're the real AI. Give us some input and out comes something like what they wanted
Is there a term for clustering algorithms that let you have variable clusters? Like train it once and be able to go through the clusters at different levels. I don't think that's a great way of explaining it. I think level set trees are an example
My understanding is that they give you a tree of clusters and you can navigate that tree to "zoom in/out". I was looking to see what similar algorithms are out there, but I don't even know what to call them
 
11:49 PM
hmmm...not sure I understand your description well
 
11:59 PM
Yeah I figured that would be the case. I'll ask again once I know more about clustering lol
 

« first day (2883 days earlier)      last day (2052 days later) »