« first day (3198 days earlier)      last day (2026 days later) » 

00:08
Fun times...
 
1 hour later…
01:22
hey folks
nm, kinda use to seeing it more lively around this time, but its been `a while since i frequented here
im kinda wondering if anybody here has worked alot with earthquakes before
I've got an interesting dataset I'm doing analysis on
besides that fridge is dead, im up to my neck in R hw, and am really annoyed with these deep scrapes i got yesterday when my bike tire skid on something
so its been quite a week so far
howdy folks
01:54
Does anyone mind proof reading a question to ask on the main site? I want it to be clear enough to answer.
Well, its here:
I have been thinking of timescales to the human races doom of using renewables recently, and I haven't been able to find many sources on these times.

For all these questions, assume that increase in the rate of energy consumption stays the same each passing year. That is, our energy consumption increases linearly each year.

If we got all our energy from geothermal, what would be the time scale before the earth’s core got cool enough to cease generation of our magnetic field?
- for this question, you would need to know how much gravitational potential energy was converted to heat in the co
hmm
you want to ask all together?
Yes it does seem like a bit much to ask all at once.
I was thinking I could attach a bounty to the question to reward people for answering all of it.
you have neough points for that/
 
1 hour later…
vzn
vzn
03:12
@tpg2114 sounds like an interesting problem, am curious, have worked on a very hard dataset, financial, may be able to assist, student can contact me here (where there is significant ML experience represented already which could also possibly be leveraged some) or other room to give it a shot chat.stackexchange.com/rooms/9446/theory-salon
03:39
Anyone familiar with string theory?
04:01
for that u want ACM
@ACuriousMind have a question when you're around
04:28
@kylecampbell what do you want to ask?
@JohnRennie I have a couple questions. First, in general, why do string theories require extra spatial dimensions? Also, what's the motivation for compactification?
When you quantise a classical theory the quantum version can have an anomaly
In the case of string theory the quantised theory has an anomaly unless the spacetime dimension is 26 for bosonic strings or 10 for supersymmetric strings.
Basically unless supersymmetric string theory is formulated in a 10d spacetime it becomes inconsistent.
ah I see
The details are way out of my league, but if you Google string theory conformal anomaly or something like that you'll find lots of articles on it.
I'm a bit iffy on what exactly the "anomaly" is, but I take your word for it that it's an issue
04:38
As for compactification, if there are 9 spatial dimensions the obvious question is why we only observe 3 spatial dimensions. Where are the other 6?
The two approaches to this are compactification or the brane world.
In the brane world approach all the standard model particles are anchored to a 3D subset of the 9 spatial dimensions. The other 6 dimensions exist but standard model particles can't move in them.
What exactly is the "brane" in that approach? The 3D subset?
In string theory, D-branes, short for Dirichlet membrane, are a class of extended objects upon which open strings can end with Dirichlet boundary conditions, after which they are named. D-branes were discovered by Dai, Leigh and Polchinski, and independently by Hořava, in 1989. In 1995, Polchinski identified D-branes with black p-brane solutions of supergravity, a discovery that triggered the Second Superstring Revolution and led to both holographic and M-theory dualities. D-branes are typically classified by their spatial dimension, which is indicated by a number written after the D. A D0-brane...
By "can't move in them", do you mean the SM particles can't interact whatsoever with the other 6 dimensions?
@JohnRennie is this kind of discussion of dimension based on expanding the symmetry groups we see in representation theory and things like the gell-mann matrices
i.e.: string theory is a representation theory with symmetries that encompass gell-mann and other symmetry groups and evolves out of that same perspective in relation to symmetry groups, just trying to describe more than hadrons
@kylecampbell Open strings have boundary conditions on their ends, and one of the conditions is that the ends have to travel within a Dirichlet brane. If the SM particles are associated with the ends of the string then the SM particles can only move within that brane.
But my knowledge of this is so limited that I can't say much more.
@Skyler don't know, sorry.
04:45
ok
Im just wondering if his question could be adequately described by first expanding his understanding of symmetries. Like if you described SU(2) vs SO(3) it might click for him
From the little I know of the subject that approach doesn't seem the right one
The symmetry involved here is conformal symmetry.
A conformal anomaly, scale anomaly, or Weyl anomaly is an anomaly, i.e. a quantum phenomenon that breaks the conformal symmetry of the classical theory. A classically conformal theory is a theory which, when placed on a surface with arbitrary background metric, has an action that is invariant under rescalings of the background metric (Weyl transformations), combined with corresponding transformations of the other fields in the theory. A conformal quantum theory is one whose partition function is unchanged by rescaling the metric. The variation of the action with respect to the background metric...
oh ok, so its a matter of scale invariance and renormalization groups
@kylecampbell did you want to ask about compactification, or is it now obvious why it's used?
@JohnRennie No, I understand the motivation now
I'm just reading...
I have another question but I'll just save it for another day
Appreciate it!
@kylecampbell when ACM does show up he'll no doubt point out that what I've said is at best a caricature, but a proper explanation would probably be incomprehensible :-)
04:58
he's missing the ACM signal though, so will he even answer our calls?
Is ACM not an omnipotent chat AI
man, ive changed computers twice since i made that bat signal photoshop for him, i wonder where that file is
btw, rejecting the null hypothesis meant getting a p-value sufficiently small (below your cutoff) right?
aka accepting your alternative hypothesis of some correlation or etc
05:54
morgen
06:25
maybe I'll try to render the arc reactor...probably too hard tho...but maybe I can try...:D
07:08
after 1 hour of work...I got 2 empty cylinders LOL
close enough
07:51
@Slereah morgen
@RyanUnger yes and no
08:12
Struggling with a question of graphical data representation. I have a MCMC sampling of a posterior PDF for a model. For most of the parameters I'm happy to plot the marginalized median and 68% confidence interval since the distributions are ~Gaussian. The exception is one parameter which is constrained to be >=0. It clearly prefers values near 0, which is physically reasonable, but because of the constraint the distribution is more like half of a Gaussian, the side >=0.
Plotting the median and 68% interval doesn't seem sane, but I'm not sure what I could reasonably replace it with. I can make something reasonable-seeming up, but I wonder if there's something more or less conventional that I should use. Anyone run into this before?
 
4 hours later…
12:04
Man there's no good rigorous stuff on Hilbert's axioms
Like no formal system stuff
People throw around that we have line segments without defining them first
12:40
create it yourself
12:51
I am
It's not easy
Especially because like
A lot of axiom sets of Hilbert's geometry are written in words
and a bit vague
So I have to make sure to write them out correctly
Otherwise, those axioms could apply to say
The empty set
A universe of discourse of no points
13:06
the insight that you are gaining is vital to a deep understanding
probably not
Synthetic geometry isn't like
hugely used
I mean
it's been used for thousands of years
It just sort of became less popular in modern times
times will always be a changin'
Sure, but only because we will all die in the hellscape of global warming
Math skills will not be too important in the Mad Max of the future
gotta start lifting to be your own Lord Humongous
The Ayatollah of Rock'n'rollah
13:16
The Shareef don't like it
Rockin' the Casbah
Rock the Casbah
The Shareef don't like it…
14:00

  Basic Mathematics

This room is meant for all basic mathematical discussion, incl...
@Slereah come to this^ room please
 
1 hour later…
15:06
a typhoon is coming, so it's raining.
vzn
vzn
cutting edge science in the 21st century o_O
 
3 hours later…
17:46
@user400188 why would taking energy from wind or tides stop the earths rotation?
18:02
@vzn the amount of fake things being shared is what really drives up the wall
18:32
had a nice chat with a neuromorphic computing researcher
very enlightening :D
I have a question concerning GR
in page 189 Schutz, he mentioned,
" It is
important to say ‘there exist coordinates’ rather than ‘for all coordinates’, since we can find coordinates even in Minkowski space in which gαβ is not close to the simple diagonal (−1, +1, +1, +1) form of ηαβ " end.
true
see e.g. spherical coordinates
19:25
I see
19:40
yarp
@EmilioPisanty hmm, I think I've found the closest equivalent to that integral in G&R
namely, 3.954.2 on page 504
(using symmetry to restrict to positive x and getting exp(ip x) -> cos(px) in the process)
general & relativity?
odd abbreviation bro
Gradshteyn and Rhyzik
weird way of spelling general relativity bro
The book of integrals that was written in the woods over vodka shots exclusively apparently
19:53
the source for that particular formula seems to be Erdelyi's Table of Integral Transforms
which seems legit
@Skyler what do you mean string theory evolves out of expanding symmetry groups beyond Gell-Mann matrices?
citation to Erdelyi looks legit
the form G&R gives is cute enough, though I still prefer using erfc
actually, huh, Erdelyi has the erfc form
so the original source for that formula looks better than G&R's version...lol
also, G&R's $\Phi(x)$ is just erf(x). But they also use erf(x) in places and I don't see why
20:09
mmhm...so Hebbian learning just implements PCA...
interesting insight...
@Semiclassical any word back on your phone interview yet?
no. which is likely a message of its own
i mean, i dunno when they make these decisions but I haven't heard a thing
so my guess at this point would be they went with someone else
@enumaris PCA = ? (My off-the-cuff guess would be principal component analysis)
@Semiclassical thanks! I'll look it up when I'm back at the office
@Semiclassical yes, principal component analysis
20:14
perhaps, they will make a last minute decision
it's not impossible. but they haven't said anything either way
where are you applying to?
generally a phone interview warrants a reply
unlike just sending a resume
if i was applying to a company, yes
i applied to one of the private liberal arts colleges in the area, for an adjunct physics instructor opening
o.O
I see..
I dunno what conclusions to draw.
I guess I'll know for sure in a month, lol
20:18
when does their semester start?
I feel like they should respond to you...
probably after labor day? lemme see
@enumaris Should, yes.
but you're applying to a bunch of places right? I dunno about academia but for industry you'd apply to a whole bunch of jobs...
I haven't applied to enough places. This one was a bit of an odd one out, given that a few of my profs forwarded the email posting to me
And when I try to look at Indeed.com listings my brain just shuts down
are you looking for industry stuff?
My experience with Indeed is that the positions there are generally low-level
20:22
i guess? i can't envision a long-term future for me in academia
U gonna transition to data science?
like 5 of my PhD peers all went into data science lol
that does seem like the default option, sigh
I don't have a lot of enthusiasm for that route. maybe I would if I properly understood it
I'm sure there's other applications Physics PhDs can get into
I just talked to a professor in Neuromorphic computing who's training was physics
20:26
i'm just very bad at spotting those things (and very good at finding reasons why posting X wouldn't work)
When I was looking for "something a Physics PhD could do" I mostly found just engineering related stuff...but those jobs generally prefer engineers
I found a posting for a "turbulence physicist" from SpaceX...but they also said no to me lol
I guess astrophysics is still too far from fluid physics
@enumaris I don't think there are many jobs out there where physics PhDs are actually preferred over someone who studied the subject
data science is actually one area..since "data science" degrees are still relatively new lol
20:31
They're not new anymore. There's not a university that doesn't have a degree or two catering to exactly that.
yeah but the quality is ...not certain...
And a physics PhD somehow is a guarantee of quality?
nothing is a guarantee of quality
But to me a Physics PhD would at least show you know math
and experimental methodology
some understanding of statistics and correlation vs causation type effects
It's a different branch of math. And the math is never the hard part anyway.
depends on the physics lol...sometimes the math is the hard part...
20:36
amusingly, the side research I've been dealing with lately has been Bell's inequality stuff and how that (almost directly) corresponds to linear correlation in stats
But maybe "prefer" is not the right term for prefer physics PhD to data science degree. Maybe more, there aren't really enough data science degree recipients out there yet to fully saturate the market
I've heard it was all Black-Scholes
In academic physics the math can indeed be the hard part (though even there I'd say coming up with an interesting problem to solve is the most difficult thing), but not really in industry.
The math isn't the "hard part" in industry sure, but you still need good foundations (at least in linear alg and statistics) to understand data science. :)
industry and universities are known to work together
20:39
depends on which part of the university
usually in R&D
the engineering departments, definitely
@enumaris Ah ok, fine, though I think not only are universities educating a lot of data scientists, there's bootcamps and all to that effect, so I would argue the opposite and in my immediate surroundings see that data science job postings want people to hit the ground running, i.e. machine learning experience (either from a previous job or some academic credentials to that effect)
the physics departments...not so much
@enumaris Ah, but I don't think physicists are that well versed in the subtleties, e.g. matrix decompositions etc as are some other people.
@bolbteppa What, data science?
20:41
@alarge in that and in stock math
@alarge yeah...the "hit the ground running" mentality is one that I don't fully agree with lol. But certainly if you augment your PhD in physics with some data science work it's not a "bad" resume for DS positions.
String and quantum field theory are the only real math-based physics things, the rest is calculus (and probability in QM, but not even up to Bayes' theorem) runs
financial math is a different beast than data science
@alarge I think you'd be surprised the lack of math fundamentals a lot of people have lol. The math you learn for physics is actually quite a strong math foundation as compared with a majority of the population.
20:43
Can you imagine invoking Bayes' theorem or even conditional probability properly in QM
eh, I know at least one context where the latter shows up
You can get very far in life with a bit of basic arithmetic
for instance, entangle two electrons in a singlet state and separate them, then measure their spins along directions a,b
What's the probability of getting up for the second particle (along direction b) given that you obtained up for the first particle (along direction a)?
(It works out to be something like sin(theta/2)^2 where theta is the angle between a & b)
if you're happy and you know it clap your hands
clap clap
notes distinct lack of other claps
20:47
:(
so correlation functions show up in QFT and they're related to probability correlations but it's super confusing to make sense of
and even formulating the question of repeated measurements abstractly, probably should be formulated in terms of correlation functions, the only qm explanation I've seen used integral equations :\
When is a matrix non-invertible?
Over a field?
20:50
yeah a standard matrix of real numbers
When you can use it to annihilate some vectors
finite dimensions?
haha
$Det A = 0$
lol yes finite dimensional matrix of real numbers
voila
yeah. which also means Av=0 for nonzero v
a lot of people don't know this - and this fact is actually important in data science
20:51
the bit of linear algebra which I don't have as much experience with is non-symmetric stuff
jordan blocks etc
yuck
I should know why google uses eigenvectors but I don't
the stuff i've been working on has made me a lot better about things like PSD matrices and Gram matrices
And matrix norms, which I knew of but didn't really have much sense of
(this is probably also why I could guess "PCA = principal component analysis")
PCA is important in data science so there u go :D
yeah
and the connection to optimization is neat
I have a sneaking suspicion I'd find the -math- behind data science / machine learning a lot more interesting than the actual topics
maybe you can look for more on the frontiers of data science
like graph neural networks or something
20:59
figure out how to connect machine learning to QM :P
except people have probably figured that out already
Penrose talked about that
I mean I suppose you can use QCs to do ML
but QCs are really undeveloped...
I keep harping on this paper that shows the formal equivalence of Reinforcement learning and light transport
but that's one area that has the exact same math (integral equations) lol
ML does use differential equations...though it's not often phrased as such...
generally first order partial diff eqs
which is basically formally equivalent to the Schroedinger equation (in time) right
@enumaris link?
 
2 hours later…
22:39
@JohnRennie thoughts on Fear Inoculum?

« first day (3198 days earlier)      last day (2026 days later) »