12:29 AM
This QuantumMagazine article quantamagazine.org/… said Leonard Susskind proposed that the ever-growing interior volume of a black hole can be indentified as the “complexity” of the black hole — roughly a measure of the number of computations that would be needed to recover the black hole’s initial quantum state, at the moment it formed. But I don't know what the complexity here means despite the above elaboration.
I don't know what the number of computations therein means.

12:41 AM
In computer science, the computational complexity, or simply complexity of an algorithm is the amount of resources required for running it. The computational complexity of a problem is the minimum of the complexities of all possible algorithms for this problem (including the unknown algorithms). As the amount of needed resources varies with the input, the complexity is generally expressed as a function n → f(n), where n is the size of the input, and f(n) is either the worst-case complexity, that is the maximum of the amount of resources that are needed for all inputs of size n, or the average-case...

1 hour later…
2:09 AM
@danielunderwood lol... I actually have written code that's been held up for months because people can't agree on what to name the directories

2:20 AM
@Avantgarde I see. If you could upload the TeX file somewhere (or just copy paste it to Overleaf), I could try to help. :)
Otherwise it's best to create a minimal working example and ask on LaTeX SE.

6 hours later…
8:17 AM
Thanks. I'm done with autocompletion. I'll get to bibliography later.

morning

2 hours later…
10:11 AM
God I hate R so much

10:24 AM
I hate rainy days at hometown.

We had Storm Deirdre blow in from the Atlantic over the weekend. The worst of the strong winds are past now, but it has been raining on and off since Friday. Lovely.

10:52 AM

11:10 AM
I've read 1Q84 and quite enjoyed it, but I'm not a huge fan of Murakami's books. I am not a fan of aphorisms, his or anyone elses.

I don't Murakami at all. But I find these text reflects what I experience.

2 hours later…
1:46 PM
you should be ready for upcoming possible storms.

2:32 PM
I don't know why every set of string notes does not do what Rebbi does, push NG the full way first
Even the normalization of the wave equation is completely obvious this way
No ghost is still driving me nuts

2:51 PM
@JohnRennie hmmmmm.
is that coming our way?
I sure hope not

Kafka on the shore was very disturbing for me in 6th grade

We have our own storm over here. Though it's more of snow and ice much earlier than usual than a large storm

3:10 PM
OK, this
0

Two electrons d apart have potential energy. Release them, they will be repelled according to Coulomb's law. But accelerating charges radiate, producing radiation, resulting in breaking action. How much energy us lost instead of becoming kinetic?

obviously runs afoul the homework policy, but it's gotten it's hooks into me and I want to answer it.
Or at least to talk about how you would approach answering it and then look at someone else's work.
The way I figure it you would find the multi-pole expansion of two identical point charges as a function of $q$ and separation and figure the radiation from the evolution thereof. Maybe.
It has been two decades since graduate E&M for me, after all.

I'd feel better about that question had they done something as simple as note what the result would be if you didn't worry about radiation
(i.e. what's the potential energy of that scenario)
I also question whether it makes sense to do that problem using classical E&M, since we're talking about two electrons specifically

@dmckee this is nowhere near obvious to me
I really can't picture any EM course that would ask this question
it's ridiculously hard
it puts you in all sorts of tricky Abraham-Lorentz situations
In the physics of electromagnetism, the Abraham–Lorentz force (also Lorentz–Abraham force) is the recoil force on an accelerating charged particle caused by the particle emitting electromagnetic radiation. It is also called the radiation reaction force or the self force. The formula predates the theory of special relativity and is not valid at velocities of the order of the speed of light. Its relativistic generalization is called the "Abraham–Lorentz–Dirac force". Both of these are in the domain of classical physics, not quantum physics, and therefore may not be valid at distances of roughly...
including (i) the EOM is third-order in time, not second order, and (ii) if you're not careful, you get some extremely ugly runaway solutions

3:32 PM
a variation on that would be: suppose you have two identical particles which start from $x=\pm \infty$ with speed $v$. If you ignore radiation, they'll move towards each other until their mutual repulsion brings them to stop, then they'll move apart and ultimately exit at $x=\pm \infty$ with speed $v$.
What happens if you include radiation? (My supposition would be that it's an absolutely miserable calculation.)

@EmilioPisanty I don't think so. It's moved off to the north now. But the pressure charts for the next week don't look good. There are a couple of deep Atlantic lows headed our way.

I think one can at least anticipate that KE will be lost along the way, so the distance of closest approach will be larger and the exit speeds will be smaller than $v$.
But that's not a calculation :S
I suspect at least some version of this problem is treated in Jackson...which is a pretty good reason for me to not want to touch it

@EmilioPisanty it's an interesting question and I've voted to reopen it. I think David might have been a bit hasty in closing it.

It's an interesting question, but I'd be hard pressed to say that the OP has presented it well

Hi

4:02 PM
@JohnRennie good.
@JohnRennie @dmckee @DavidZ now reopened.

@EmilioPisanty now we just need someone to answer it :-)

'ndeed

A simpler version of this might be to replace one of the moving electrons with a fixed charge
still plenty hard, but at least only one of the charges is moving

Sup guyz
I was wondering, in the Brainbow technique in neuroscience, how did the researchers at MIT managed to genertically change the DNA of the targetted organism so it's neuron structure will be colorful (each neuron will have a color)?
How did they understand exactly what DNA and where to put it?

5:05 PM
@danielunderwood nice, thanks :D
I haven't had time to look deeply into a lot of the stuffy I should be looking into tho

5:30 PM
bleh, scheduling these 2 follow up interviews has been a pain
Anybody here have experience in 3d modeling?

@enumaris I have quite a lot with Blender.

I've heard of that before, it's an open source 3-d Modeling software?

Yep, pretty good for anything. Animations, modelling game characters and props...

Interested in learning more about the field in general
So far the books I've found online are about specific software for 3D rendering
which is a very practical approach...but specialized for one particular software
I was looking for more of a general introduction to the field itself

Here is my advice. Do not read books. Download Blender (or any software you choose) and start experimenting. You'll fail and mess up but that's how you learn the best. You can watch some youtube videos or google things you want to do.

5:40 PM
well my goal isn't really to become a 3-D rendering artist

I've messed with inventor and solidworks, but those are more engineering than art

I wanted to look into AI-augmented 3-D rendering which is a pretty new field. But I have 0 knowledge of 3-D rendering so I was thinking of getting some understanding of it
but I suppose playing around in a 3-D rendering software would give me some experience

I'm interested in neuroscience. Do you know where I can find help and answers to questions? I mean, I come here to learn and ask questions about physics but what about neuroscience?

maybe here

I think this is kind of the inverse of what you're looking for, but you may be interested. My understanding is that they do reinforcement learning in a simulated environment with Unreal aireverie.com

5:43 PM
right, that's using a simulated environment to train AI

In terms of rendering with AI, that's pretty far out there I think. Just getting into rendering on its own is a bit of a journey

I'm more looking for using AI to simulate environments :D
I know the folks over at nvidia are doing it
I feel like there's a big opportunity in that space
esp with all the movie and game studios that need to use 3-D rendering extensively

@enumaris If you just want to be able to model and render in 3D, just play around with the software and the mind will learn naturally, but if you want to know how things work and get into the technical side of the things, better read a book or take an online course. :)

Are they using it just for lighting or more general stuff? Lighting systems are an interesting topic

@danielunderwood I think they are experimenting with a lot of stuff
GANs were mentioned
they also mentioned like using AI to help anti-aliasing
stuff like that
and lighting w/ ray tracing was definitely mentioned
time to found my own start up to do AI-based 3-D rendering boys
lol
make some deals w/ them moovie studios

5:47 PM
Yeah a substitute for ray tracing was my thought. Ray tracing is too expensive to do realtime, so you usually have something like Phong or Torrence-Sparrow approximations in games. Movies to ray tracing, but I'm sure they'd go for something faster if it's realistic

@NovaliumCompany yeah I was thinking of trying to get an understanding of the technical side...but all the books I've found so far are just for how to use a specific software
@danielunderwood I feel like there's gotta be a TON of opportunities to make things faster and more realistic

and I'd be 100% shocked if deep learning (or some replacement for it) didn't play a role in the next great advance

@enumaris What exactly are you intersted in learning?

@NovaliumCompany hard to be exact since I have 0 experience in the field lol
At this point it's more just general knowledge of the field

5:51 PM
@enumaris Of which field? 3D AI modelling or?

3D modeling

I don't want to pigeonhole myself into the existing AI-assisted 3-D modeling technologies since that's such a new field

OGRE may be an interesting place to look. It's an open source rendering engine, so you'd have access to the lighting code. Unreal is open source (kind of), but it's a bit of a zoo. I'd say grab a rendering engine and start poking at the lighting code

I want a stronger base in 3D modelling to start off :D
presumably I could also grab Blender and start poking at the code...
but that seems very difficult lol

5:53 PM
Yeah looking into blender and modeling may be a good direction too. I started writing a game engine a few years ago and got stuck not knowing the concepts of materials

like a very difficult way to learn the subject though
surely there are courses in 3D modeling...

Blender also has python plugins and I think you may be able to tweak lighting there

@enumaris You best learn about 3D modelling by downloading a software and learning, that's why all the book are about a specific software. But in most 3D modelling programs, the theoretical side is the same. So it doesn't matter which software you use or which book on a software you read, in the end, you end up with the same general 3D modelling rules and techniques.

Also note that "3D modeling" is probably going to be more art while "3D rendering" is going to be more math/software/physics

^
But for starting off, I'd recommend learning Blender first. Also, this guys is amazing: youtube.com/user/AndrewPPrice

6:02 PM
I remember this book being quite good, but I think there's only one chapter on the rendering side of things. It does explain some quite low-level details about processing as well. There are pdfs floating around if you wanted to see what it has

Morning folks

Also one of the suggested books in that amazon listing is about physics-based rendering and looks pretty neat. It also seems that is has a free online version pbrt.org
"Computer graphics" may also be googlable for technical details

6:32 PM
PBRT I think is the standard book on the subject matter. I have one of the older versions. I don't like how the code is structured at all, and the book is written in Knuth style kind of like a well documented codebase
But the contents ought to be solid.
And if you want to look at specific state of the art things, I guess see what gets published at SIGGRAPH

6:48 PM
@NovaliumCompany thanks for the resource :D
nice resources all around :D
hmmm interesting, this company is dropping future interviews cus they couldn't find someone who's available at the times I am available lol
alrighty then

7:05 PM
"The ideas and software in this book show the reader how to design and employ a full-featured rendering system capable of creating stunning imagery." seems like exactly what I'm looking for :D

I'm disappointed that I didn't find that book back when I was messing around with graphics stuff
I do have to say that that has been the most fun I've ever had with software

let's start a start up for AI 3D rendering lol, fun times all around
We just need to find funding
I can contribute 3 dollars

I can cover the bosonic string section of the startup :\

I too can contribute 3 dollars!

Lets give string theory real world applications!

7:11 PM
lol

"we're just simulating the woldsheets of the photons"
Or is photon bad terminology in string theory?

The world is our oyster, we can call any massless spin 1 thing a photon now...
I really should get those Feynman gravitation lectures

seriously tho, there's gotta be a ton of money to be made in this space...
just gotta figure out how to extract it

@bolbteppa I bought the book a while back and didn't understand a thing when I opened it. I guess I should take a look and see if I'm smarter now
There's also a Feynman lectures on computation that's sitting on my bookshelf

7:17 PM
@enumaris by beating nvidia at their own game. Should be easy

It's pretty nuts, looked in it before, shocking tbh, but at the same time he waves his hands in deriving the spin of things

@danielunderwood they don't have a monopoly on this

Yeah that one. Looks like I got it for around half of the current price too. Clearly I should start investing in used books instead of stocks

Shocking how expensive it is

I feel like since they are chip makers, they have research interest in doing rendering, but like they aren't the ones that are being hired by movie studios to produce photorealistic CGI for the movies.
The competitors would be firms which do that...

7:21 PM
I feel like they probably also want a piece of the ai pie

yeah
but there ain't no place with 0 competition bruh
at least, it's pretty freaking hard to find one lol
It's all in how you execute the idea

I think Epic went in a pretty interesting direction of marketing Unreal towards designers and such on top of game devs. I have no idea how much other competition is in that field though

WE CAN BEAT THEM
HAVE FAITH IN THE HEART OF THE CARDS

Just go a step further and not have any bad cards!

7:50 PM
Evidently a company is giving me a programming portion of a technical demo in a language I've never used...should be interesting

8:39 PM
XD

My interview with them last week used something I didn't know for the technical portion as well. It's actually kind of refreshing that they do that. It's almost like they know people can learn new things

"There are three main audiences that this book is intended for. The first is students in graduate or upper-level undergraduate computer graphics classes. This book assumes existing knowledge of computer graphics at the level of an introductory college-level course, although certain key concepts such as basic vector geometry and transformations will be reviewed here."
I have no background in computer graphics though... D:
@danielunderwood that's nice :D

@enumaris a large portion of that is probably rotation matrices, vectors, and transformations
and maybe a bit of quaternions

yuck
but ok

The guy that likes GR says yuck to those things?

8:46 PM
I like coordinate free GR
I'm gonna study 3D rendering and simulating...then I'm gonna simulate our world...and then train reinforcement learning algorithms to take it over...and then use that for world domination! Muahahahaha!

Feel free to do that in a coordinate free way :D

sounds legit

Speaking of which, I've always been a bit bothered by the use of "tensor" in ML

it be a matrix
but a higher dimensional matrix
apparently in CS that's what "tensor means"
matrix possibly with D>=3
at least that's what my software engineer partner that ghosted me told me...
now that she's ghosted me, I'm not sure how much I trust her word lol

I figured CS people never even heard the word tensor

8:53 PM
The basic foundations for physically based rendering are the laws of physics and their mathematical expression.
cool, seems like I got a leg up on this :D

I only heard it in my relativity course aside from mentions of intertia tensor and field strength tensor without any explanation more than "pretty much a matrix"
Now just approximate all the computationally expensive parts!

needs more AI
On a serious note though, is this field your actual field of work? Or are you just interested in it?
(or you're not interested in it, in which case I misinterpreted something you said before lol)

I could understand tensor as a synonym for an array of arbitrary dimensionality (eg 10-by-10-by-10)
Whereas calling a 1000-by-1000 array a tensor seems strange

1000x1000 is 2D which is part of "arbitrary dimensionality" tho

True enough
I mean, the basic rule is more like “a tensor is an array which isn’t a matrix or a vector”. But even that gets slippery

9:00 PM
Well at one point I decided I wanted to write a game and started doing things with ogre. Then I decided I wanted to do it more from scratch for whatever reason and did things with plain opengl. That ended up being a much larger thing than I had anticipated, but I enjoyed being able to look into the technical details. I've wanted to return to it, but just writing a game engine would be one of those projects that not much comes out of unless I did something really exceptional

I see
What's OpenGL?

It's one of the libraries you use to actually render things. You give it your vertices, textures, shaders, and whatever else for your GPU to compute what is displayed on the screen. There's also stuff to actually get that in a window. I think that's what GLFW and similar are responsible for. Vulkan and DirectX are alternatives to OpenGL with Vulkan being new and DirectX being windows-only

I see
makes sense :D

Shaders are their own language that's executed on the GPU to calculate things. Generally you have a vertex shader that determines where vertices are and a fragment shader that colors triangles based on textures/materials (that's also where lighting is handled to an extent), but you can also have other types of shaders. That's for OpenGL at least. I'm not sure if the others are the same. You may find ShaderToy interesting
The language for shaders is pretty c-ish with vector/matrix math built-in

hmmm
now to AI it up and bypass those parts?
lol

9:10 PM
My hunch would be that you generate a net that can generate lighting and pass parameters for that into the fragment shader or generate a fragment shader from it, but I'm not entirely sure

well certainly nobody is gonna be "sure" at this moment
it is a newly emerging field :D
but I would need to know more about each piece before I could even tell if it makes sense to even try to apply AI/Deep Learning to it

Now the training part would be interesting. My hunch for that would be that you want to generate a bunch of weird geometry and do ray tracing to calculate either a 3d image or 3d movie that you pass to your GAN

so the pbrt book intro mentioned "rendering is turning a 3D model into a 2D image" or something along those lines
but how do you build the 3D model in the first place? That is a question I'm having right now o.o
isn't building a 3-D model a tough part of the problem lol

Well if you were making a game, you'd use something like Blender or Autodesk 3DS. But since you ideally want a bunch of training data, you're going to either have to grab a bunch of high-quality models or generate them

so Blender creates the 3D model and renders it in a way you can visualize right
that's what Blender is for?

9:17 PM
You create the 3D model in Blender

I don't think generating geometry would be too bad unless you really want the geometry itself to be representative of game environments

but like if I want to generate a room that I can take a VR walk through tour of...
That would combine the problem of generating a 3-D model and also rendering that 3-D model onto a VR headset right

And the rigging etc to animate it. And then add some physics simulations it for fluids and whatnot, And the renderer shoots light and imagines different surfaces as different kind of textures etc so light will bounce off of it differently

that makes sense

You could put AI into use to try and figure out some of the physics of the animations as well
Like how the fluid moves etc

9:19 PM
I guess my question is does pbrt only go from 3D Model -> 2D rendering (presuming you have a 3-D model built already using some other software), or does it take you through the whole process building 3-D model and then 3D model -> 2D rendering

Or how crowds should avoid each other when you simulate a lot of people etc

@alarge yeah, all very interesting to me :D

It goes from 3D scene to a picture, so yes, 3D to 2D, but it needs to read the standard files, so you'll go through the formats and how graphics are represented as vertices etc

I see

I think I saw a paper a year or two ago applying machine learning to try to compute some atomic orbitals

9:21 PM
is the format with which you represent a 3D scene pretty standard across the industry?

Actually one of the real things I had an interest of was physical manipulations of environment (things like blowing up a wall). I think the direction you're looking could be pretty interesting for that as well

cgchannel.com/2018/07/… there's a bit of data that Disney released earlier this year

There are a variety of formats and I think some engines want their own. OBJ is pretty easy to handle and fairly common though

I see...
@danielunderwood yeah those things are definitely interesting to me as well :D

Disney also publishes a lot of papers on their simulation methods etc in case you're interested (you'll often find them in SIGGRAPH)

9:24 PM
simulating (astro)physical environments was my PhD so I have a little bit of exposure to that kind of stuff I guess...though in my case I was mostly solving Schroedinger's equations for various initial conditions.
@alarge thanks! It's a lot of information to go through so I'll be on this for a while probably XD

Schrodinger's equation in astrophysics?

yep
cus I simulated the neutrinos

Also I think this could be interesting in terms of generation. Say you wanted to generate models based on a few sample models or terrain based on real terrain maps

not the supernova itself
@danielunderwood yes...these are all directions which I think there's a ton of opportunity

Are neutrinos not generally relativistic?

9:26 PM
hence my assertion that there's a "ton of opportunity" in this space :D

uhhh typically relativistic

they are, but you can still model them with an approximate Schroedinger's equation
You basically use a Taylor expansion of $H=\sqrt{p^2c^2+m^2c^4}$
for the ultra-relativistic limit

huh neat
It still hurts me to see an operator under a square root

hence the Taylor expansion
it's what Dirac tried at first I think... except in the ultra-relativistic limit rather than the non-relativistic limit

In that case it is, but I've heard other people talk about the square root of an operator like its own operator

9:35 PM
uhhh

9:48 PM
weez

10:14 PM
@danielunderwood there are some sensible versions of that, to be sure. for instance, it's perfectly legit to ask whether a positive operator $B$ exists such that $A=B^2$

muscles sore...
need to hit the gym more often...

10:40 PM
@Semiclassical ahh that does make sense I suppose
And I used to never get sore at the gym, but my doctor recently told me to kind of randomize my workouts and I've been sore pretty much every day since

lol yeah, if you get used to the movements your body gets efficient at doing them and then you don't get sore...
like, my core is not sore even though I cranked up the weights on the abdominal machine
but my chest muscles are sore from chess presses

That's some heavy pieces you're playing with

heh
chest presses :D

11:28 PM
I don't think I've ever managed to make my chest muscles sore. Either that or I confused the feeling with heartburn

well the part that gets sore is more like towards your arms