/sci/ - Science & Math » Thread #12841837

676KiB, 1001x648, single_source_logmap.jpg

View Same Google iqdb SauceNAO

/mcs/ - Mathematical CS General

Anonymous Tue 16 Mar 17:52:25 2021 No.12841837 View Reply Original Report

Quoted By: >>12842085 >>12842895 >>12842903 >>12842908 >>12843053 >>12843357 >>12843948 >>12844565 >>12851902 >>12856672 >>12857507 >>12866899 >>12880924

Since /cspg/ >>12793853 is down for the foreseeable future, I think it's time to bring this thread back.
This is a thread dedicated to the mathematical aspects of computation. It doesn't matter whether or not it's theoretical CS or applied.

Nonetheless, here are some resources for beginners:
https://teachyourselfcs.com/

Free books:
https://z-lib.org/
http://gen.lib.rus.ec/
https://spoon.wiki/Books

There was a wonderful graphics and visualization thread on /g/ on graphics and visualization, so here are some links stolen from there.
>inb4 muh /g/

Intro:
https://gabrielgambetta.com/computer-graphics-from-scratch/

Shader basics:
https://web.engr.oregonstate.edu/~mjb/cs557/

Physics Based Rendering (great AND legally free online):
http://www.pbr-book.org/

CMU Graphics courses:
http://graphics.cs.cmu.edu/?page_id=16

Keenan Crane's discrete differential geometry courses (2020, 2021):
https://brickisland.net/DDGSpring2020/
https://brickisland.net/DDGSpring2021/
Be sure to do the readings (for each respective semester):
https://brickisland.net/DDGSpring2020/category/readings/
https://brickisland.net/DDGSpring2021/category/readings/
Associated text:
http://www.cs.cmu.edu/~kmcrane/Projects/DDG/paper.pdf

Cool research from a graphics lab on developing elastic textures:
http://vcg.isti.cnr.it/Publications/2015/PZMPCZ15/

Anonymous

Anonymous Tue 16 Mar 2021 18:05:48 No.12841902 Report

Quoted By: >>12841935 >>12841962

Why is /cspg/ dead? Also, if /cspg/ is dead I don't see jow this thread is gonna live either because it's even more niche.

Anonymous

Anonymous Tue 16 Mar 2021 18:11:17 No.12841935 Report

Quoted By:

>>12841902
>Why is /cspg/ dead?
it's dead because some real life stuff happened to the guy who organized it. He's gonna be busy for a while, according to >>12811030
>I don't see jow this thread is gonna live either because it's even more niche.
The last time we had an /mcs/, it hit the bump limit.
Either way, I don't see how this is niche on a board about math and science lmao. Math and CS is a ridiculously rich interaction with lots of intersection. The only real qualification I added is that this thread is mostly for CS, not basic coding.

Anonymous

Anonymous Tue 16 Mar 2021 18:16:30 No.12841962 Report

Quoted By:

>>12841902
Original mathematical CS thread for reference:
https://warosu.org/sci/thread/S12303630

Anonymous

Anonymous Tue 16 Mar 2021 18:41:45 No.12842085 Report

Quoted By: >>12869153

>>12841837
Any logicbros around? Any cool uses of model theory in CS? I've seen some old stuff (ie https://www.cs.technion.ac.il/users/wwwb/cgi-bin/tr-get.cgi/1992/CS/CS0711.revised.pdf), but does anyone know anything about the state of the art?

Anonymous

Anonymous Tue 16 Mar 2021 21:11:18 No.12842886 Report

Quoted By: >>12844986

>hurr CS fags can't into math
https://arxiv.org/abs/1810.09250
Even algorithmicists, traditionally thought to do only combinatorics, can do functional analysis mathfags. Cope harder.

Anonymous

Anonymous Tue 16 Mar 2021 21:13:19 No.12842895 Report

Quoted By: >>12842907

>>12841837
>Keenan Crane's discrete differential geometry courses (2020, 2021):
Based as fuck you guys are aware. I love this field and found amazing applications for it in my PhD.

Anonymous

Anonymous Tue 16 Mar 2021 21:14:26 No.12842903 Report

Quoted By: >>12846075 >>12846525

>>12841837
Anyone studying quantum computing / information? Particularly topological quantum computers. I was wondering if anyone has any good references for learning more about anyons and braiding.
Good intro
https://arxiv.org/abs/1705.04103
Notes I’m still going through since they’re pretty dense lol
https://web.math.ucsb.edu/~zhenghwa/data/course/cbms.pdf

Anonymous

Anonymous Tue 16 Mar 2021 21:15:02 No.12842907 Report

Quoted By: >>12842951

>>12842895
Oh shit, what did you do for you PhD thesis? What's your subject of study?

Anonymous

Anonymous Tue 16 Mar 2021 21:15:13 No.12842908 Report

Quoted By: >>12842940 >>12853232

>>12841837
Lads what libraries to you use for visualization?

I have a system of fluids coupled with moving solid state bodies, I would idealy also like to plot vector fields. I don't know if it's even possible to do this in a single library.

Anonymous

Anonymous Tue 16 Mar 2021 21:21:36 No.12842940 Report

Quoted By: >>12842959

>>12842908
don't most people use autodesk CFD for fluids? Dunno how it interacts with solid state stuff though.

Anonymous

Anonymous Tue 16 Mar 2021 21:23:10 No.12842951 Report

Quoted By: >>12843041

>>12842907
I am a chemical engineer, I found some applications for it in a solutions thermodynamics problem that has allowed us to scale simulations from a very low number of mixture components to accurately simulating mixtures properties of hundreds of components. Essentially I just use a slightly modified version of the cotan formula to speed up the regression of an underlying surface.

It's really satisfying because we actually tried a lot of conventional numerical methods first, I had read Keenan Crane's notes back in 2016 and at some point I realized that I could apply it to solve a multilevel surface optimization problem we had. It's been one hell of a ride seeing how well it works on my physical system.

I think you CS friends did amazing work in this field. We have a lot to learn from how you simulate fluids in computer graphics too. I believe it will replace many of the "rigorours" CFD methods that we are currently using at some point in the future. I am happy to see that DDG is becoming more popular in general too.

Anonymous

Anonymous Tue 16 Mar 2021 21:24:16 No.12842959 Report

Quoted By:

>>12842940
>autodesk
I'll take a look at this thanks.

Anonymous

Anonymous Tue 16 Mar 2021 21:48:12 No.12843041 Report

Quoted By: >>12845494

>>12842951
Damn, it's really cool to hear Crane's work being useful in various areas of science and engineering. That, and your research sounds really interesting. If you don't mind me asking, what optimization problem were you studying? What were the other methods you used and their shortcomings?
I also know analytic combinatorics, another theory extensively studied in CS, has been used in chemical synthetic methods, infared techniques, etc..
I knew in the back of my mind that graphics and visualization research has been motivated by these things, but it's ridiculously cool seeing it work out to such a success.
>We have a lot to learn from how you simulate fluids in computer graphics too.
To be fair, there's a lot of work that has to be done here from the CS side as well lol, but I agree that papers in SIGGRAPH is constantly pushing the boundaries of how fundamental the computer has become to modern engineering.
I like to study problems in science and engineering (these days mostly stuff used in condensed matter, and also dip my heels into topological states / anyons since topological quantum computing / information is a big field now) from the CS/math side, and it's easy to feel out of the loop. So hearing about these big developments in CS being used in other fields always puts a smile on my face and reminds me that the work is worth it.

Anonymous

View Same Google iqdb SauceNAO 1611632211157.jpg, 97KiB, 709x783

Anonymous Tue 16 Mar 2021 21:51:22 No.12843053 Report

Quoted By: >>12843704

>>12841837
very cool links, OP. will look into. i've been writing my own graphics site and these will give me ideas and help me find any holes in my coverage of topics.

Anonymous

Anonymous Tue 16 Mar 2021 23:14:33 No.12843357 Report

Quoted By:

>>12841837
Closely related to discrete differential geometry is discrete complex analysis. See chapters 5 and 6 of
https://arxiv.org/abs/math/0504358

Anonymous

Anonymous Wed 17 Mar 2021 00:46:04 No.12843704 Report

Quoted By:

>>12843053
Cool, glad to know it was helpful

Anonymous

Anonymous Wed 17 Mar 2021 01:43:01 No.12843932 Report

Quoted By: >>12844565

Do any of you guys work on NP-hard problem approximation?
Any cool open problem you want to discuss?

Anonymous

Anonymous Wed 17 Mar 2021 01:45:12 No.12843948 Report

Quoted By: >>12844048

>>12841837
i'm guessing the phase encodes the geodesic and the magnitude encodes the distance?

Anonymous

Anonymous Wed 17 Mar 2021 02:13:17 No.12844048 Report

Quoted By:

>>12843948
https://nmwsharp.com/research/flip-geodesics/
"Our FlipOut procedure can be used to efficiently introduce a geodesic edge path from a single source to all other vertices (left). Reading off distance to each vertex and the direction of the edges emanating from the source yields a discretization of the logarithmic map (right). The cyclic color palette encodes the angular component of the logarithmic map, while stripes denote isocontours of its magnitude."

Anonymous

Anonymous Wed 17 Mar 2021 04:51:43 No.12844565 Report

Quoted By: >>12844605

>>12841837
>>12843932
Anyone here do the blue optimization book?
https://web.stanford.edu/~boyd/cvxbook/
I'm thinking of reading this since I want to learn more about semidefinite programs and maybe conic programs.

Anonymous

Anonymous Wed 17 Mar 2021 05:10:00 No.12844605 Report

Quoted By: >>12844717

>>12844565
I read that a few years ago.
It's a great book but it can be slow sometimes.
His lectures are available on youtube I think but it's kind of so-so because he goes on a tangent a lot.

Anonymous

Anonymous Wed 17 Mar 2021 06:10:12 No.12844717 Report

Quoted By:

>>12844605
Thanks. Good to know.

Anonymous

Anonymous Wed 17 Mar 2021 08:26:41 No.12844986 Report

Quoted By: >>12845044

>>12842886
combinatorics is really common in NP hard problems. true chads work on this stuff.

Anonymous

Anonymous Wed 17 Mar 2021 08:51:37 No.12845044 Report

Quoted By:

>>12844986
I like combinatorics a lot, but I think I still have a slight preference for mathematics with more "structure" in its study if that makes sense. Combinatorics is the bread and butter, don't get me wrong, but I think the parts of CS that are steeped more in algebra, logic, and analysis are more interesting than the combinatorial problems.
>NP-hard
the latest progress on P vs. NP comes from algebraic geometry and representation theory
https://www.math.tamu.edu/~jml/simonsclass.pdf
http://pcwww.liv.ac.uk/~iken/teaching_sb/summer17/introtogct/gct.pdf

Anonymous

Anonymous Wed 17 Mar 2021 10:48:49 No.12845287 Report

Quoted By: >>12846505

I have a humanities degree and will be starting a compsi degree next September. How do you recommend I split my free time to prepare myself with some self studying? I work full time and have other obligations, so I don't have a ton of time right now and had thought of self-studying one hour of maths and one hour of programming everyday, but I had negleted the actual applied computer science part.
What do you think would be the most efficient way to spend my time?

Anonymous

Anonymous Wed 17 Mar 2021 12:45:19 No.12845494 Report

Quoted By: >>12846075 >>12846122

>>12843041
>Damn, it's really cool to hear Crane's work being useful in various areas of science and engineering.
Yeah and to be honest I think it's only just starting to pick up. There are countless applications for this work. A lot of complex problems can solved by shifting your perspective to models based on DEC. I'm predicting a 6 figure impact factor explosion for him in the next 5 years.

>. If you don't mind me asking, what optimization problem were you studying?
I don't want to say on 4chan sorry, there are only two other research groups in the world that study this particular problem. Some of my work has made into popular open source libraries that you might be familiar with too.

>I also know analytic combinatorics, another theory extensively studied in CS, has been used in chemical synthetic methods, infared techniques, etc..
Haven't studied this at all, seems interesting.

>...graphics and visualization research has been motivated by these things, but it's ridiculously cool seeing it work out to such a success.
The only reason I think the engineering community hasn't pick it up yet is because it's "computer graphics", but they haven't really looked deeply at what numerical science and underlying algorithms of CFD actually does. I think there is some work left to introduce more rigour than "it looks nice", but it will replace conventional CFD

>topological states / anyons since topological quantum computing
Any good resources/review papers you would recommend for getting into this? Seems like a very exciting field.

>and it's easy to feel out of the loop. So hearing about these big developments in CS being used in other fields always puts a smile on my face and reminds me that the work is worth it.

I think CS has already replaced chemistry as the central science. Nearly all the top h-index scholars are computer scientists now. The work in "mathematical CS"/combinatronics is an extremely useful technology to everyone in science and engineering.

Anonymous

Anonymous Wed 17 Mar 2021 14:23:13 No.12845704 Report

Quoted By:

What is a good next step after Rosen?

Anonymous

Anonymous Wed 17 Mar 2021 16:27:16 No.12846075 Report

Quoted By: >>12846122 >>12846525

>>12845494
Ah fair, thanks for describing your research! It sounds very interesting. As for graphics, I got the impression that EE and CS “know” that SIGGRAPH and “computer graphics” really means graphics and physical visualization, but I feel like mechE and material science people are generally unaware since CAD covers a lot of use cases. Though funnily enough I know that a decent amount of geometers in CS like Jean Gallier are co authors in aerospace and other older fields of engineering.
>Any good resources/review papers you would recommend for getting into this?
There are some good resources here >>12842903
The first link is an easy enough introduction but the second might take some time since it’s definitely more technical. I’m still perusing through and asking my physics friends for resources from the physics side, but I think Kitaev’s original paper on the matter isn’t actually that bad a read.
>I think CS has already replaced chemistry as the central science.
Hmm, I wonder if that’s giving CS a bit too much credit, but I do see your point in the analogy. In traditional engineering, materials (and thus chemistry) is fundamental since you have to build out of something. Today, a lot of engineering is about solving problems using computers, and you build solutions out of algorithmic knowledge.
>Nearly all the top h-index scholars are computer scientists now
Really? I guess being in quantum information so long has skewed my perception since I feel like in the minority of computer scientists working on stuff related to physics problems. Then again, the reason error correction is such a hot topic is because it gives the hardware engineers a (hopefully feasible) end goal, and the guys at Microsoft Q seem to be pushing out lots of interesting research from the math/CS side that directly affects implementations. My bias works against me again lol.

Anonymous

Anonymous Wed 17 Mar 2021 16:35:33 No.12846100 Report

Quoted By: >>12846128 >>12846544

Any discrete optimization chads in this thread?

Anonymous

Anonymous Wed 17 Mar 2021 16:43:38 No.12846122 Report

Quoted By: >>12846608

>>12845494
>>12846075
On the topic of graphics in materials science, the last link in the OP is some really cool research on elastic microstructures that came out of a graphics lab in Europe that Crane has worked with in the past
http://vcg.isti.cnr.it/Publications/2015/PZMPCZ15/

I think my shock for CS’s high regard in other communities is that my experience prior to grad school was basically “I’m not like other CS majors” during my bsc since we all know the codemonkey and bad curriculum stereotype. Even though I “knew” that was wrong in the back of my mind, since I know CS PhD’s who work at nvidia for memory virtualization, firmware, and hardware, and I know CS BSc’s who work for JPL at NASA. But I have a hard time “feeling” it since it always felt like CS didn’t have as much prestige until the last 4-5 years after many YouTube channels and big companies have popularized things like AI/ML research. It doesn’t help that the shitposting here sometimes gets to me. It’s legitimately a shock to hear about CS in such a positive light from an engineering researcher on /sci/ of all places; imagine that lmao. Thanks for that anon, I really appreciate it.

Anonymous

Anonymous Wed 17 Mar 2021 16:44:40 No.12846128 Report

Quoted By: >>12846322

>>12846100
I’m not an discrete optimization expert, but I’ve done a few problems using LP duality and the matching polytope. What’s up?

Anonymous

Anonymous Wed 17 Mar 2021 17:37:57 No.12846322 Report

Quoted By: >>12846358 >>12846360

>>12846128

Just checking, I feel like no one gives discrete opt (or really opt in general) any love on /sci/

Anonymous

Anonymous Wed 17 Mar 2021 17:50:32 No.12846358 Report

Quoted By:

>>12846322
Unfortunately barely anybody talks about combinatorics or discrete mathematics on /sci/. Aside from the shitposting, I think the floor is mostly to analysts here because of the reverence that this board has for signal processing and fluids (these fields are great, but haven’t felt the combinatorial influence as much as the analytic one)
In general I think /sci/ is mostly biased toward EE, physics, and some parts of math. Mostly just analysis and ring theory lol

Anonymous

Anonymous Wed 17 Mar 2021 17:51:34 No.12846360 Report

Quoted By:

>>12846322
If it makes you feel any better, discrete optimization is used everywhere lmao. So it is high impact research

Anonymous

View Same Google iqdb SauceNAO a0ee910fddb6e86db0ff69d755392f55.jpg, 48KiB, 564x663

Anonymous Wed 17 Mar 2021 18:05:40 No.12846409 Report

Quoted By: >>12846508 >>12846530 >>12851541

Are there any math books/lectures that teach with programming notation? I'd like to try studying calculus and analysis this way. Or any math.

Anonymous

Anonymous Wed 17 Mar 2021 18:28:51 No.12846505 Report

Quoted By:

>>12845287
Learning CS is a continuous process.
There are a lot of new concepts that will take time for it to "click".
So instead of cramping a ton of material when the deadline of something is close, you should start learning concepts now. It also looks like you will be busy so you should start learning bit by bit before September.
I would recommend spending 10-15 hours a week to learn from online classes (there are many good one) or books.

Anonymous

Anonymous Wed 17 Mar 2021 18:31:06 No.12846508 Report

Quoted By: >>12846676

>>12846409
I don't think there is. Why would you want something like that?

Anonymous

Anonymous Wed 17 Mar 2021 18:36:29 No.12846525 Report

Quoted By: >>12846791 >>12846843

>>12846075
Yeah, you are right, only EE and CS really know SIGGRAPH. Also another thing is though that the solvers used for fluid simulation in SIGGRAPH (especially MPM) are so good (meaning fast and scalable) now that they are being adopted by mech/materials engineers who are trying to make them rigorous. Part of the issues is that CS will do a simulation that "looks good" and call it a day without doing any kind of physical validation on the solver. But those solvers that just "look good" are so good that they are actually more accurate than many rigorous multiphase CFD methods.

For example de Vaucorbeil and co-authors prepared this autistically long review that I enjoyed reading:
https://www.researchgate.net/publication/336796234_Material_point_method_after_25_years_theory_implementation_and_applications

But back to the topic of DDG, Keenan's work is only the most famous use cases in graphics, it's already being used for a lot of other things in math and physics too. There is a huge SFB grant (in Germany that funds basically an entire small institution of around 50 research staff members) that works in DDG, although it is less famous than Krane's group they do interesting work (some of their papers are hard to read though):

https://www.discretization.de/projects/

The group that works on the Fokker-Plank equation is a good example imo of how this is being used to solve problems that conventional methods suck at.

> >>12842903
Thank you for the links. Sorry I'm retardedly blind and missed your post.

>Really?
Yes, but mostly CS working in imagining (MRI etc) which sense because it has so many useful practical applications and of course the eternal ML meme. Guys like Yann LeCun etc. are all over 100 now. I've seen so many researchers with 160+.

The thing is though both imaging and ML are immensely useful technologies everywhere. I think that CS truly deserves its current spotlight, CS advanced by far the most of any field in the previous decade.

Anonymous

Anonymous Wed 17 Mar 2021 18:37:17 No.12846530 Report

Quoted By:

>>12846409
>programming notation
...such as? Function signatures are directly lifted from the math notation, just in a different order. While there is less abuse of it, I don’t see how notation to allocate resources, assign variables, and return would necessarily help in studying calculus. Researchers in CS don’t even study analysis using code. See
https://arxiv.org/abs/1305.0453

At some point you have to sling the lingo. I suggest Abbott’s Understanding Analysis at first and then moving onto baby Rudin after finishing up derivatives in Abbott. For calculus, any MIT OCW course should suffice

Anonymous

Anonymous Wed 17 Mar 2021 18:39:41 No.12846544 Report

Quoted By:

>>12846100
Not discrete, but I've published in global optimization and multilevel optimization before.

Anonymous

Anonymous Wed 17 Mar 2021 18:59:24 No.12846608 Report

Quoted By: >>12846843

>>12846122
Kek. Yeah I think maybe your original problem is browsing /sci/. Past a PhD level someones' background really does not matter. At our (national) lab we have people from every field: math, physics, CS, engineering, even biochemistry etc. and it's not as straight forward as the narrative /sci/ (and tbf a certain class of irl professors) wants to build. Sometimes the physicist is the math phobic experimentalist while the CS and engineers only do modelling. A CS researcher will be teaching combinatronics to a mathematician who used to work in CFD modelling etc.

Though, of course, certain thinking patterns emerge depending on people's backgrounds, but it always depends on the individual. There are people who (at least appear to) know everything and are good at everything, and people who know nothing even about their own discipline outside of their niche field. But, you know, like we've always said that even on /sci/ while some undergrads spout their insecure major politics shit. The only thing that really surprised me about research in fact is that the "best" researchers here (who have won certain prizes so big that I can't even say which prizes) are the ones that are by far the most insistent on dumbing down everything we do. And they are right. You have no idea how many Nature papers that look simplistic have an extremely deep and intricate theory and mathematical background that was suppresed to get the paper published to a wider audience.

But yeah anyway, congrats on you CS bros, it has been very inspiring for all of us in the research community to see the immense progress you made in the field while we were dicking around nanotech and superconductor memes.

Anonymous

Anonymous Wed 17 Mar 2021 19:19:34 No.12846676 Report

Quoted By:

>>12846508
just wanted to see if it was a thing, I've already passed calculus but would be cool if there was a book like that. I mean the first computers were largely used to automate derivative calculation, etc.

Anonymous

Anonymous Wed 17 Mar 2021 20:02:57 No.12846791 Report

Quoted By: >>12846843 >>12847105

>>12846525
Thanks for your links as well! No problem on missing the others, since small posts are easy to miss on general threads.
>Part of the issues is that CS will do a simulation that "looks good" and call it a day without doing any kind of physical validation on the solver. But those solvers that just "look good" are so good that they are actually more accurate than many rigorous multiphase CFD methods.
I'm not super familiar with the state of the art, but as far as I'm aware, most of the methods that come out of CS departments have their own mathematical rigor and correctness that probably works on idealizations of certain cases. By rigor here, do you mean the model's capacity to do real world simulation with nontrivial deviations from the ideal cases? I'm just a bit confused about the difference in rigor between the new solvers and the rigorous multiphase CFD methods.
>https://www.discretization.de/projects/
I'll have to take a look at the German lab's work in DDG, but I'm glad that it's spreading in popularity internationally. We've seen how traditional mathematics
Adding on to both points above, we've seen how traditional mathematics have given us incredibly powerful computational ideas that help us understand complex phenomena. I'm curious to see how the reverse gets more and more popular, ie computation yields research with a largely traditional flavor. I'm already pretty interested in the work that gets published in Experimental Mathematics.
>Yes, but mostly CS working in imagining (MRI etc)
Ah that makes sense, yeah computer vision and imaging profs generally have a lot of citations lmao. Their work is super cool.
>of course the eternal ML meme
I'm actually pretty interested to see how ML turns out, as someone who's dipped into some of the theory side of ML. There's a lot of sharp theory work and a lot of sharp engineering work to be done, and the most interesting stuff generally happens somewhere in the middle.

Anonymous

Anonymous Wed 17 Mar 2021 20:16:01 No.12846843 Report

Quoted By: >>12847105

>>12846525
Continuing from >>12846791
I'm starting to feel like ML research getting sufficiently complex past just being hypercompetitive that the memes about low quality research are becoming less true every day. Nowadays if you don't understand the basics of RKHS or measure theory and functional analysis in general, it's seen as a disadvantage rather than a blocker to running stuff on a better architecture.
>>12846608
Ah yeah, I've heard the national labs do a lot of interdisciplinary research. I mean, I know because my own work is interdisciplinary, but I hear national labs are even more mixed melting pots when it comes to doing work. That being said, one thing that disappointed me when I looked at places like Lincoln Labs is that the CS people they employed always seemed to be working on stuff I felt was "boxed in" or somewhat closed off from the rest of the facility, though I was *only* looking at their lead researchers. What I'm trying to say is that when I saw their quantum computing lab, it was mostly physics and EE, and their CS labs seemed be focused on cybersecurity and cryptography. That might just be LL's reputation for being communications research focused though.
I am curious to see if I should spring into a national lab, since I've always felt like my interests never fit cleanly into a single subject.

If I have to give CS credit though, I think I agree with you in that it's the field that's advanced the most in the last few decades. We went from solving cute balls and bins problems in the 80s to GPT-3, DDG, practical motion planning in robotics, and effectively feeding and clothing like 30% of the logicians arounds the world.
>while we were dicking around nanotech and superconductor memes.
lel who knows, now we might have enough computing technology to tackle these problems with greater force. I mean, I know the condensed matter physicists are interested in quantum computers explicitly because it can solve the Hamlitonians they like lel

Anonymous

Anonymous Wed 17 Mar 2021 20:25:26 No.12846878 Report

Quoted By: >>12846930

Can people even make money working in complexity theory / algorithm analysis nowadays?
I was reading a good book on TCS (by Arora) and I just realized that he jumped to the ML ship years ago because that's where the money is at.
I did a master in ML/Computer Vision but switched to approximation algorithms in my current PhD because I feel like I can't keep up with the fast pace of the ML community (and because more than 50% of the PhD students of my year in my school want to work in ML). I don't know if I made a career mistake.

Anonymous

Anonymous Wed 17 Mar 2021 20:38:09 No.12846930 Report

Quoted By:

>>12846878
>Can people even make money working in complexity theory / algorithm analysis nowadays?
I mean there are good positions that pay decently for TCS
>I was reading a good book on TCS (by Arora) and I just realized that he jumped to the ML ship years ago because that's where the money is at.
except that's the wrong realization. He does ML theory, not just mainstream ML. If you've read any of his work, it's clear he did it because he finds computational problems in that setting interesting.
>I did a master in ML/Computer Vision but switched to approximation algorithms in my current PhD because I feel like I can't keep up with the fast pace of the ML community (and because more than 50% of the PhD students of my year in my school want to work in ML). I don't know if I made a career mistake.
Do what interests you. If you're in CS, you can switch easily from one thing to another. Just be sure to keep publishing in the meantime while you make your jump.

Anonymous

Anonymous Wed 17 Mar 2021 21:15:41 No.12847105 Report

Quoted By:

>>12846791
>By rigor here, do you mean the model's capacity to do real world simulation with nontrivial deviations from the ideal cases? I'm just a bit confused about the difference in rigor between the new solvers and the rigorous multiphase CFD methods.

Yes I mean for the more general cases. The idea with CFD in general is that it should converge to the correct behaviour for any continuous system if you refine your discretisation. In particular you can precisely compute your errors to determine the accuracy of your simulation to a physical system. This is important when desiginging a process/it's control system. But then of course, like graphics, CFD also has methods that only work for certain use cases. When you start working with multiphase systems it becomes a bit rarer to find rigorous methods

>>12846843
National labs are the promised land. High pay and you can focus on research 100% of the time no teaching duties etc. to disctract you. Even though it's more corporate you still maintain some academic culture since most researchers and PIs of course are still coupled to universities. I will stay here as long as I possibly can, though I know I will eventually have to do some kind of teaching to advance my career.

The department pages are pretty arbitrary and based on which grants PIs were able to bring in rather than reality. Everyone tends to share coffee so people often work together on projects. My lab in particular has a lovely culture where we have a seminar every Friday so we are able to understand everyone's topics.

>now we might have enough computing technology to tackle these problems with greater force.

Kek, yeah unironically that is my hope. After I am finished with my current project we will either have god-tier nanoscale battery tech or at least near perfect understanding and control of self-assembled systems (or, more likely, I will be unemployed).

Anonymous

Anonymous Wed 17 Mar 2021 23:46:11 No.12847633 Report

Quoted By: >>12847638

It's closer to the math side I think, but has anyone gone through Alice and Bob Meet Banach? It seems cool but I don't know enough geometric analysis to read it without struggling. According to this link (http://www.mit.edu/~aram/advice/quantum.html) it's "the definitive treatment of the probabilistic method in quantum information."

Anonymous

Anonymous Wed 17 Mar 2021 23:47:15 No.12847638 Report

Quoted By:

>>12847633
Forgot link to the book. Here it is:
http://math.univ-lyon1.fr/~aubrun/ABMB/index.html
It has the link on the bottom but you have to remove the "REMOVETHIS" part of the link

Anonymous

Anonymous Wed 17 Mar 2021 23:52:55 No.12847669 Report

Quoted By: >>12847693 >>12847725

>found a paper that does a cool thing
>convoluted explanations and badly described algorithms
>no implementation anywhere

why is it so hard to replicate anything from these

Anonymous

Anonymous Wed 17 Mar 2021 23:58:54 No.12847693 Report

Quoted By: >>12847809

>>12847669
Give a source. I highly doubt it was really convoluted and more that your background wasn't sufficiently technical lmao.
>the part time parliament and PAXOS
Leslie Lamport is an exception, not a rule

Anonymous

Anonymous Thu 18 Mar 2021 00:06:34 No.12847725 Report

Quoted By: >>12847809 >>12848472 >>12849230

>>12847669
>Researchers who have good ideas
>Researchers who can code nicely
>Researchers who can write things cleanly
It's a tough combo. A lot of papers are written by PhD students so you probably won't have it all.

Anonymous

Anonymous Thu 18 Mar 2021 00:31:11 No.12847809 Report

Quoted By:

>>12847693
Cant recall anything from the top of my head.
Admittedly im not in CS or maths.

>>12847725
Most likely you are right.

It wouldnt take anything to just post a minimal working example most cases.

Anonymous

Anonymous Thu 18 Mar 2021 04:06:05 No.12848472 Report

Quoted By:

>>12847725
People generally acquire those skills over the PhD and their post doc. Honestly, while not always scalable, I’ve seen many libraries come out of research labs that are great. But some of the horror stories are true too

Anonymous

Anonymous Thu 18 Mar 2021 04:08:04 No.12848478 Report

Quoted By: >>12848509 >>12848787

There’s an odd absence of shitposting on this thread. On one hand, that’s really nice since CS isn’t cut any slack on this board, but on the other, all the fighting keeps the thread bumped at the top for hours.

Anonymous

Anonymous Thu 18 Mar 2021 04:21:09 No.12848509 Report

Quoted By: >>12848658

>>12848478
There are like 12 people in this thread right now so shitposts are kept to a minimum.

Anonymous

Anonymous Thu 18 Mar 2021 05:13:32 No.12848658 Report

Quoted By: >>12849233

>>12848509
lol you're right. It'll probably pick up if this thread persists to the next day

Anonymous

View Same Google iqdb SauceNAO sadAI.png, 789KiB, 720x698

Anonymous Thu 18 Mar 2021 05:53:12 No.12848780 Report

Quoted By: >>12848846 >>12848856 >>12848864 >>12848893 >>12849237

>all the CS chads ITT who are doing fun shit
>meanwhile I slave away sending up commits at a full time job

golden handcuffs are real. god I miss academia and actually doing interesting stuff. the money wagecucking is good but is soul crushing. I don't want to accomplish nothing other than having a lot of money.

Anonymous

Anonymous Thu 18 Mar 2021 05:55:45 No.12848787 Report

Quoted By:

>>12848478
yeah I came in this thread expecting the following:
>CS is not really math
>CS is not really a science
>you're just a codemonkey in training

Anonymous

Anonymous Thu 18 Mar 2021 06:14:47 No.12848846 Report

Quoted By:

>>12848780
gib monee

Anonymous

Anonymous Thu 18 Mar 2021 06:18:19 No.12848856 Report

Quoted By: >>12848867 >>12855350

>>12848780
Anon just do some side projects if you have time.

Anonymous

Anonymous Thu 18 Mar 2021 06:22:37 No.12848864 Report

Quoted By: >>12855349

>>12848780
You could always try and get a masters funded by your place of work. Try and see if you can take some of the interesting courses. There's no way to jump right into research without taking a pay cut, but there's no reason why you shouldn't be able to dip your feet into it, either

Anonymous

Anonymous Thu 18 Mar 2021 06:23:39 No.12848867 Report

Quoted By: >>12849237

>>12848856
I think anon's sad about not doing interesting things near the cutting edge, not just an interesting side project he can bang out in a couple of weekends

Anonymous

Anonymous Thu 18 Mar 2021 06:35:31 No.12848891 Report

Quoted By: >>12849016 >>12849238

Math major, CS minor, looking for coursework advice given that I intend to work either as a dev or math postdoc monkey after graduation.

Courses:
>2x: Real Analysis, Complex Analysis, Abstract Alg, Diff Geo, Topology
>OS, Algos & Complexity, Machine Learning (2x), Rando Algos & Probabilistic Analysis

That's pretty barebones, so I'm considering adding courses in Numerical Analysis (2-3x) and Numerical Optimization (2x) for employability. Otherwise, these slots would be taken up by third courses in Analysis, Algebra, Diff Geo or Topology.

Despite being directly interested int topics, I'm reluctant to add the numerical analysis and optimization courses, because 1) I just want to fucking graduate and 2) I feel a third course is way more valuable than an intro, i.e anyone can self teach and intro, not everyone can study analysis for a full year.

I also have no statistics or probability in my curriculum, and I'm not sure how bad that may sting later. I can afford up to 3 additional terms, but seriously just wanna GTFO asap at this point.

Anonymous

Anonymous Thu 18 Mar 2021 06:37:32 No.12848893 Report

Quoted By: >>12848897

>>12848780
academia is gay, retard

Anonymous

Anonymous Thu 18 Mar 2021 06:38:42 No.12848897 Report

Quoted By:

>>12848893
cope

Anonymous

Anonymous Thu 18 Mar 2021 07:47:50 No.12849016 Report

Quoted By: >>12850407

>>12848891
It literally sounds like you know what you want to do but want validation

Anonymous

Anonymous Thu 18 Mar 2021 09:21:24 No.12849230 Report

Quoted By: >>12850162

>>12847725
Yes, but certain researchers make a point out of keeping the paper as convoluted as possible and don't want to explain themselves. They are also the same ones that complain about being on their 3rd postdoc with their sub 4 h-index so...

Anonymous

Anonymous Thu 18 Mar 2021 09:23:04 No.12849233 Report

Quoted By:

>>12848658
I would be shitposting a lot more if you started this thread on the weekend.

Anonymous

Anonymous Thu 18 Mar 2021 09:25:47 No.12849237 Report

Quoted By: >>12850355

>>12848780
>>12848867

Yeah this, even industry research positions that hire PhDs is soul crushing. You never get to work on cutting edge ambitious ideas. Mostly your "research" is about trying to implement something academics wrote on an industrial scale or improving on old technology.

Enjoy your PhDs lads, it is the last time you will have fun with your work.

Anonymous

Anonymous Thu 18 Mar 2021 09:28:27 No.12849238 Report

Quoted By: >>12850407

>>12848891
>Numerical Analysis (2-3x) and Numerical Optimization (2x) for employability.

To be honest with you I've found that most employers don't even know what optimization really means. If you have an industrial or financial engineering department try to look for a course in Operations Research. It's basically dsicrete optimization, but on supply chain or stock portfolio problems. That will actually get you hired, I regret not doing something like that during my Masters.

Anonymous

View Same Google iqdb SauceNAO buchi automata.jpg, 61KiB, 680x630

Anonymous Thu 18 Mar 2021 10:05:35 No.12849280 Report

Quoted By: >>12850913

anyone know any good textbooks/other resources for linear temporal logic, computation tree logic, buchi automata, and model checking (i.e., the intersection of these things)?
been doing this in my theory classes, finding it particularly enticing

Anonymous

View Same Google iqdb SauceNAO Capture1.png, 10KiB, 689x130

Anonymous Thu 18 Mar 2021 11:25:45 No.12849450 Report

Quoted By:

Brainlet here. I don't understand. Why is this true?
This is a problem proving a sequence is Cauchy.

Anonymous

Anonymous Thu 18 Mar 2021 12:38:04 No.12849620 Report

Quoted By: >>12849998

Can I skip directly to Elements of Automata Theory by Sakarovitch if I have not done Sipser? I want a rigorous intro and I've heard that the book does it perfectly. Would I miss anything?

Anonymous

Anonymous Thu 18 Mar 2021 14:58:40 No.12849998 Report

Quoted By:

>>12849620
Sipser is super clear but very basic I’m that it’s a survey that’s at the level of sophomore math maturity. Sakarovitch is much harder, but he takes you very deep into the abstract algebra in authority. It’s a trade off.
I’d say you should breeze through Sipser in like 2 months and then start Sakorovitch

Anonymous

Anonymous Thu 18 Mar 2021 15:53:38 No.12850162 Report

Quoted By:

Anyone read this (and the actual paper)?
https://www.quantamagazine.org/new-algorithm-breaks-speed-limit-for-solving-linear-equations-20210308
>>12849230
Yeah these people exist as well but there are only a few of them.
If a paper has shit writing or their codes are bad, I would just assume it's just the author is bad at those.

Anonymous

Anonymous Thu 18 Mar 2021 16:53:43 No.12850355 Report

Quoted By: >>12850374 >>12851240

>>12849237
>Yeah this, even industry research positions that hire PhDs is soul crushing.
I've seen some interesting industry research stuff at places like Microsoft and IBM. However, it looks like core CS research at Google focuses a lot on fast theoretical problems under the hood of selling advertisements to people, so I can imagine that being soul crushing.
Msoft Q and IBM Q researchers are academics who regularly publish results for journals and conferences.

Anonymous

Anonymous Thu 18 Mar 2021 17:02:20 No.12850374 Report

Quoted By: >>12850441

>>12850355
Those jobs are super interesting, but my perception of those kind of positions is they are only available to the most elite researchers at the most elite American institutions. The only people that I've seen make it there come out the top CMU or Stanford groups.

Anonymous

Anonymous Thu 18 Mar 2021 17:12:06 No.12850407 Report

Quoted By: >>12850925 >>12850948

>>12849016
well im not sure how much value they really add. and I know I could learn the material outside of class in a month anyhow. I would prefer to do that and take the opportunity to learn upper div math in more depth, which I'm not as confident I could (or would) self learn later, and which can't really be applied to a github project as easily as numerical analysis or optimization. and am I being an idiot by not having probability / stats? (covered a bit in discrete math, and I'll be taking ML, so I hope that's sufficient)

>>12849238
i was worried HR might not really care about these things. I heard they all pair very well with machine learning, but personally I'd rather just learn this shit on the side and apply it in a github project where I can explain what I learned about these topics. do you think that'd be sufficient over taking the classes? I'm interested in this stuff so I know I'll learn it anyway

Anonymous

Anonymous Thu 18 Mar 2021 17:19:15 No.12850441 Report

Quoted By:

>>12850374
>The only people that I've seen make it there come out the top CMU or Stanford groups.
It's not really about the school. It's about presence at conferences and your advisors. There are plenty of excellent advisors at great but not "top" schools.
But yes, they are competitive

Anonymous

Anonymous Thu 18 Mar 2021 18:12:04 No.12850733 Report

Quoted By: >>12850856

>muh manifolds

Anonymous

Anonymous Thu 18 Mar 2021 18:36:14 No.12850856 Report

Quoted By:

>>12850733
What about them?

Anonymous

Anonymous Thu 18 Mar 2021 18:48:33 No.12850913 Report

Quoted By: >>12851813

>>12849280
>model checking
handbook of model checking springer

Anonymous

Anonymous Thu 18 Mar 2021 18:51:14 No.12850925 Report

Quoted By: >>12851209 >>12851536

>>12850407
Only engineers know what optimization even means. HR is incredibly clueless.

Anonymous

Anonymous Thu 18 Mar 2021 18:56:38 No.12850948 Report

Quoted By: >>12851536 >>12851559

>>12850407
>i was worried HR might not really care about these things. I heard they all pair very well with machine learning, but personally I'd rather just learn this shit on the side and apply it in a github project where I can explain what I learned about these topics.

Also to add to my previous post I've contributed to scipy.optimize and other open source libraries, I don't see it getting me a job in a lucrative OR/ML/Data position. Having been to a lot of conferences where FAANG and oil recruits from my impression is that they only care about OR and maybe something like scikit-learn or one of the more major ML libraries would get you hired with them. But I don't know in general, it's not really my industry, just my impression.

The optimization firms themselves prefer hiring their own PhD students that they supervised.

So yeah the tl;dr of it is they think it's "too theoretical" for them, that's why I would recommend you couple it with OR or financial math etc.

Anonymous

Anonymous Thu 18 Mar 2021 19:34:11 No.12851185 Report

Quoted By:

>Master student next to my lab who works in ML was able to published a paper.
>My friend in that lab told me that the master student ran the benchmark script wrong so the experiment result was a botch.
>Advisor ignored that and decided to publish it anyway.
Does this shit always happens to popular research fields, where the pressure to publish something is strong?

Anonymous

Anonymous Thu 18 Mar 2021 19:39:37 No.12851209 Report

Quoted By: >>12851299

>>12850925
I always have a hard time explaining to people what optimization/operations research is. Either i explain what I actually do all day which is just applied math + some coding or I explain it like I would day 1 of an undergrad class and then I just sound like a supply chain major.

Anonymous

Anonymous Thu 18 Mar 2021 19:44:29 No.12851240 Report

Quoted By:

>>12850355
yah sometimes its depressing thinking about how many people have studied variations of the matching problem just to help trick people into CONSOOOOOOming

Anonymous

Anonymous Thu 18 Mar 2021 19:53:21 No.12851299 Report

Quoted By:

>>12851209
Just pick some easy to explain problems that people would understand and add
>It sounds easy but it's hard to solve for the best value
I usually go with 2D rectangle packing because my friend in mechanical engineer was actually faced with this problem and he asked me how to solve it optimally.

Anonymous

Anonymous Thu 18 Mar 2021 20:29:00 No.12851536 Report

Quoted By:

>>12850948
>>12850925
ok, I think I'll just stick with stats and ML then. HR should know stats and ML at least. not much of a point in taking it when I'd rather self teach this anyway (so I can make an interesting project and not be rushed through a typical course).

Anonymous

Anonymous Thu 18 Mar 2021 20:29:37 No.12851541 Report

Quoted By: >>12863574

>>12846409
not sure if thats what you mean but
calculus with julia
coding the matrix
and discrete mathematics and functional programming

Anonymous

Anonymous Thu 18 Mar 2021 20:31:03 No.12851559 Report

Quoted By: >>12872765

>>12850948
> don't see it getting me a job in a lucrative OR/ML/Data position
and to be clear, you don't feel like a couple classes in optimization would help any either?

Anonymous

View Same Google iqdb SauceNAO 1601885923752.gif, 18KiB, 120x140

Anonymous Thu 18 Mar 2021 20:33:21 No.12851578 Report

Quoted By: >>12853224

Any of (You)'s ever go from a math undegrad to CS Masters?

Anonymous

Anonymous Thu 18 Mar 2021 20:55:54 No.12851813 Report

Quoted By:

>>12850913
perfect mate thanks a lot, covers just about everything i'm into atm
<3

Anonymous

Anonymous Thu 18 Mar 2021 21:05:55 No.12851902 Report

Quoted By: >>12852015

>>12841837
anyone knows how to properly self-study ML/AI engineering and Data Science for someone from non-CS background? I graduated in transportation engineering, and would like to study ML

Anonymous

Anonymous Thu 18 Mar 2021 21:22:54 No.12852015 Report

Quoted By: >>12852037 >>12856110 >>12861758

>>12851902
Use this, easier book
https://web.stanford.edu/~hastie/Papers/ESLII.pdf

Data science
https://www.cs.cornell.edu/jeh/book.pdf

Anonymous

Anonymous Thu 18 Mar 2021 21:26:48 No.12852037 Report

Quoted By:

>>12852015
thank you my friend!

Anonymous

Anonymous Fri 19 Mar 2021 00:13:28 No.12852845 Report

Quoted By: >>12852854

Anybody here good with combinatorics/graph theory? How does one derive the formula for counting the number of labelled forests on $n [\math] vertices with <script type="math/tex"> k [\math] connected components? I have a recurrence relation figured out, but I have no idea how to get the result from Cayley's formula without using generating functions. It should be <script type="math/tex"> k n^{n-k-1} [\math]. This problem has been bugging me for weeks.$

Anonymous

Anonymous Fri 19 Mar 2021 00:14:53 No.12852854 Report

Quoted By:

>>12852845
$k n^{n-k-1}$

Anonymous

Anonymous Fri 19 Mar 2021 01:52:45 No.12853214 Report

Quoted By: >>12853215

Should I start with Stanley's combinatorics or an easier book like Bona's A Walk Through Combinatorics? I'm finding it nowadays that I need combinatorics for CS.

Anonymous

Anonymous Fri 19 Mar 2021 01:54:06 No.12853215 Report

Quoted By:

>>12853214
Stanley is really good but fuck it's difficult. Van Lint-Wilson is also good.
I haven't read Bona's book

Anonymous

Anonymous Fri 19 Mar 2021 01:55:30 No.12853224 Report

Quoted By:

>>12851578
There should be no issue if you're going to theoretical CS or somewhere math heavy. You might not have the best time if you're trying to systems like cache design, but even then you could likely find an in through scheduling theory.

Anonymous

Anonymous Fri 19 Mar 2021 01:59:27 No.12853232 Report

Quoted By: >>12873409

>>12842908
OpenFOAM solver and Paraview for post processing, its opensource

if you have some bigtime money: D'assault systems SIMULIA

Anonymous

Anonymous Fri 19 Mar 2021 07:52:11 No.12854120 Report

Quoted By: >>12854182 >>12854318 >>12854322

How do you guys start to work on discrete / combinatorics problems?
I always have troubles with these because it feels like it requires some huge leaps of logic to get to the answer.
I can never get to the answer by doing something like chipping down the problem.
Calculus is fine to me though. Not sure why.

Anonymous

Anonymous Fri 19 Mar 2021 08:26:49 No.12854182 Report

Quoted By: >>12854423 >>12855258

>>12854120
>Calculus is fine to me though. Not sure why.
Because combinatorics is straight up harder lmao. You have to do a lot of work to think outside the box and even more work to convince yourself that you're thinking in the right way

Anonymous

Anonymous Fri 19 Mar 2021 10:02:05 No.12854318 Report

Quoted By:

>>12854120
combinatorics problems don't really have general strategies. that's what makes it so hard

Anonymous

Anonymous Fri 19 Mar 2021 10:04:15 No.12854322 Report

Quoted By:

>>12854120
yeah you have to actually think to solve combis, not just apply calc cookbook type stuff

Anonymous

Anonymous Fri 19 Mar 2021 10:45:50 No.12854423 Report

Quoted By:

>>12854182
True

Anonymous

Anonymous Fri 19 Mar 2021 15:29:26 No.12855258 Report

Quoted By: >>12858452

>>12854182
Yeah my problem is that I never know if I'm going the right way.
It's either right or not right most of the time and my hunch doesn't mean shit.

Anonymous

Anonymous Fri 19 Mar 2021 16:07:38 No.12855349 Report

Quoted By:

>>12848864
>you could always try and get a masters funded by your place of work

Unironically good advice. Thanks anon.

Anonymous

View Same Google iqdb SauceNAO dizzytiredsadupset.gif, 280KiB, 128x128

Anonymous Fri 19 Mar 2021 16:09:03 No.12855350 Report

Quoted By: >>12855941

>>12848856
Easier said than done. I've been self studying anomaly detection and have been trying to implement it on some hobby project, but it is really difficult to do so when I'm blasting for 5-6 hours a day during the weekend.

Anonymous

Anonymous Fri 19 Mar 2021 16:27:17 No.12855394 Report

Quoted By: >>12855841 >>12858968

What do you guys think of Code Parade?
https://www.youtube.com/watch?v=yY9GAyJtuJ0
I feel like he's the 3blue1brown of CS.

Anonymous

Anonymous Fri 19 Mar 2021 18:35:18 No.12855841 Report

Quoted By:

>>12855394
Seems pretty cool

Anonymous

Anonymous Fri 19 Mar 2021 19:14:07 No.12855941 Report

Quoted By: >>12866060

>>12855350
have you tried creating synthetic data?
i need to rig up some anomaly detection for threat detection, and my line of attack will be to generate synthetic 3d data for simulating what my sensor sees, and use that same data to render images with uniform colors representing classes/types/labels to detect.

for instance, suppose you have a simulation that models subsurface radar for landmine detection. you generate random scenes with the relevant material parameters and random landmine locations. then you also render the scene using 1 color to label NORMAL and another to label ANOMALY. use these in conjunction as training data

Anonymous

Anonymous Fri 19 Mar 2021 20:30:02 No.12856110 Report

Quoted By:

>>12852015
nice, these look great.
Do you have any resources on bandits?

Anonymous

Anonymous Fri 19 Mar 2021 22:52:50 No.12856672 Report

Quoted By:

>>12841837
Can most recursive solutions to problems be written as tail recursion?

Anonymous

Anonymous Sat 20 Mar 2021 00:17:55 No.12857001 Report

Quoted By: >>12857247

People shilled the Kozen book on the theory of computation, but the only copies available are used ones for $100. If it's actually good, why doesn't it get reprints?

Anonymous

Anonymous Sat 20 Mar 2021 01:13:06 No.12857247 Report

Quoted By:

>>12857001
lol what, just use libgen / z-lib.
most people use sipser because it's very clear, but Kozen is a good reference too. Both are fairly easy though

Anonymous

Anonymous Sat 20 Mar 2021 02:50:43 No.12857507 Report

Quoted By:

>>12841837
A great resource on Discrete Differential Geometry came out from a CSE / math professor like 6 months ago:
https://cseweb.ucsd.edu/~alchern/teaching/DDG.pdf

Anonymous

Anonymous Sat 20 Mar 2021 06:12:19 No.12858006 Report

Quoted By: >>12858704

Do you guys think we can have some group activity going on?
Solve a problem together?
Book/paper reading?
Code something together?
Might be fun.

Anonymous

Anonymous Sat 20 Mar 2021 09:57:08 No.12858452 Report

Quoted By:

>>12855258
i suggest reading alon-spencer on the probabilistic method. there are a lot of great problems in that book that teach you how to stare at something and come up with some sort of approach

Anonymous

Anonymous Sat 20 Mar 2021 11:09:05 No.12858704 Report

Quoted By: >>12859678

>>12858006
I am not solving any problems
I am not reading any books
and
I am not coding anything
Might be hard

Anonymous

Anonymous Sat 20 Mar 2021 13:32:43 No.12858968 Report

Quoted By:

>>12855394
Yeah, same cringe.

Anonymous

Anonymous Sat 20 Mar 2021 18:00:44 No.12859678 Report

Quoted By:

>>12858704
why are you here then lmao?
we don't have to do anything together but if you're not gonna do anything at all, even alone, what's the point of being here?

Anonymous

View Same Google iqdb SauceNAO 0123D203-8097-4896-9AB4-430C6526 (...).jpg, 241KiB, 1000x1000

Anonymous Sat 20 Mar 2021 23:46:14 No.12860906 Report

Quoted By: >>12862641 >>12862714

Watch 2 minute papers.
https://youtu.be/dVa1xRaHTA0

Anonymous

Anonymous Sun 21 Mar 2021 05:36:15 No.12861758 Report

Quoted By:

Anyone here into ML/AI? Do you know any good computational learning books? I see that statistical learning is already covered in >>12852015

Anonymous

Anonymous Sun 21 Mar 2021 13:11:12 No.12862641 Report

Quoted By: >>12863396

>>12860906
Already do, although I wish he would also do more videos from other areas of computer science.

Anonymous

Anonymous Sun 21 Mar 2021 13:35:57 No.12862714 Report

Quoted By: >>12863390

>>12860906
I can't watch content like this and not get mad that the academic world is still stuck in 1800 forcing us to present our work in a contrived writing style, in a format that's both too short to present full derivations for everything and (usually) too long to be writeable in one sitting, and making 15 minute live presentations at conferences which end up serving for nothing except as marketing for your paper.

I don't think the solution is necessarily to upend the entire process of academic research, but god damn just replacing live presentations with pre-recorded videos with high production value would solve so much.

Anonymous

Anonymous Sun 21 Mar 2021 16:38:30 No.12863390 Report

Quoted By: >>12864347

>>12862714
I mean, that’s usually what reading groups are for right? They invite speakers for a reason

Anonymous

Anonymous Sun 21 Mar 2021 16:39:32 No.12863396 Report

Quoted By:

>>12862641
Agreed. Having a 2 minute papers for pure theory would be really cool

Anonymous

Anonymous Sun 21 Mar 2021 17:08:25 No.12863542 Report

Quoted By: >>12863565

What would a theoretical computer science chart look like? Would it be all that different from teachyourselfcs?

Anonymous

Anonymous Sun 21 Mar 2021 17:13:50 No.12863565 Report

Quoted By: >>12866258 >>12866651

>>12863542
Yes it would be different and focused very heavily on math.
I’ll put out a detailed list soon

Anonymous

Anonymous Sun 21 Mar 2021 17:17:39 No.12863574 Report

Quoted By:

>>12851541
these seem interesting, thanks

Anonymous

Anonymous Sun 21 Mar 2021 18:38:00 No.12863870 Report

Quoted By: >>12864162

How do I get a research internship in NLP?
t. Non American

Anonymous

Anonymous Sun 21 Mar 2021 19:52:56 No.12864162 Report

Quoted By: >>12865911

>>12863870
no clue. try reaching out to some professors?

Anonymous

Anonymous Sun 21 Mar 2021 20:49:05 No.12864347 Report

Quoted By: >>12864441

>>12863390
There are three things on this earth that are unavoidable: death, taxes, and reading groups going to shit after 2 sessions.

Anonymous

Anonymous Sun 21 Mar 2021 21:12:21 No.12864441 Report

Quoted By: >>12864455

>>12864347
Lmao what? My local mathematical physics reading group is great.

Anonymous

Anonymous Sun 21 Mar 2021 21:15:39 No.12864455 Report

Quoted By: >>12864487

>>12864441
I have never been a part of a reading group that didn't either stop meeting, have most of the attendees start skipping sessions, or simply meets so infrequently it's not even worth considering.

Anonymous

Anonymous Sun 21 Mar 2021 21:21:09 No.12864487 Report

Quoted By:

>>12864455
it helps if you have a paper you want to read but don't have the knowledge or expertise to do so

Anonymous

Anonymous Sun 21 Mar 2021 21:33:39 No.12864545 Report

Quoted By: >>12865292 >>12865665

Can I use Haskell for anything?

Anonymous

Anonymous Sun 21 Mar 2021 21:36:31 No.12864560 Report

Quoted By: >>12865321

How do I get into using proof assistants and automated theorem proving?

Anonymous

Anonymous Mon 22 Mar 2021 00:32:52 No.12865292 Report

Quoted By:

>>12864545
There are a lot of things you can do with Haskell. Literally look up any body of resources on Haskell

Anonymous

View Same Google iqdb SauceNAO C85DBB6D-4E6A-49B3-89DC-A26E5ED7 (...).jpg, 537KiB, 1565x880

Anonymous Mon 22 Mar 2021 00:39:06 No.12865321 Report

Quoted By:

>>12864560
Pierce’s types book and then HoTT.
Also there’s a lot of interesting stuff near the end of the old thread:
http://www.cs.cmu.edu/~rwh/talks/MURI20-talk.pdf

https://www.cs.uoregon.edu/research/summerschool/summer18/lectures/harper1-1.mp4

http://symbolaris.com/course/constlog-schedule.html

https://www.ias.edu/ideas/2014/voevodsky-origins

https://www.cs.cmu.edu/~cangiuli/papers/abcfhl.pdf

https://scs.hosted.panopto.com/Panopto/Pages/Sessions/List.aspx#folderID=%2207756bb0-b872-4a4a-95b1-b77ad206dab3%22

http://www.cs.cmu.edu/~rwh/pfpl/index.html

Anonymous

Anonymous Mon 22 Mar 2021 01:55:37 No.12865652 Report

Quoted By: >>12865935

What's a good capstone project for CS that is heavy on maths/theory?

Anonymous

Anonymous Mon 22 Mar 2021 02:00:06 No.12865665 Report

Quoted By:

>>12864545
It's Turing-complete, so theoretically, yes

Anonymous

Anonymous Mon 22 Mar 2021 03:03:34 No.12865911 Report

Quoted By:

>>12864162
I have been sending out emails but haven't got any replies yet. Do you know of any that might offer places this late?

Anonymous

Anonymous Mon 22 Mar 2021 03:07:39 No.12865935 Report

Quoted By: >>12866120

>>12865652
Lel look at the resources in this thread. Plenty has been said about discrete differential geometry as both very rich mathematical theory and as something that can be applied everywhere

Anonymous

Anonymous Mon 22 Mar 2021 03:41:07 No.12866060 Report

Quoted By:

>>12855941
I was just going to pull some datasets online to play with

Anonymous

Anonymous Mon 22 Mar 2021 04:05:54 No.12866120 Report

Quoted By: >>12866155 >>12866250 >>12868582 >>12868608

>>12865935
Is this doing for an undergrad? My faculty is really shitty since I'm from a third-world country. All they can suggest is some meme software bs, and I'm too dumb to think up of any real-world applications.

Anonymous

Anonymous Mon 22 Mar 2021 04:18:09 No.12866155 Report

Quoted By: >>12866250 >>12868582

>>12866120
doable*
Fucking autocorrect.

Anonymous

Anonymous Mon 22 Mar 2021 05:02:28 No.12866250 Report

Quoted By: >>12866368

>>12866120
>>12866155
Idk what’s your math background?

Anonymous

Anonymous Mon 22 Mar 2021 05:05:47 No.12866258 Report

Quoted By: >>12866781

>>12863565
Nice. thanks for taking the time. Looking forward to it.

Anonymous

Anonymous Mon 22 Mar 2021 05:40:19 No.12866368 Report

Quoted By: >>12881432

>>12866250
I've done Rudin Ch. 1-8 and some of Munkres' Topology.

Anonymous

Anonymous Mon 22 Mar 2021 07:24:03 No.12866651 Report

Quoted By: >>12866781

>>12863565
Thanks! Looking forward to it too

Anonymous

Anonymous Mon 22 Mar 2021 08:38:56 No.12866781 Report

Quoted By: >>12866783 >>12866792

>>12866258
>>12866651
It’s impossible to list out everything in theoretical CS without tedious motivating many subfields of math. So I’m just going to list out the “obvious” foundations. If you have any questions about whether your favorite subfield of math is in CS, the answer is almost surely either yes or that such math has no application anywhere yet. I’m assuming you have calculus up to at least vector calculus. This is an intro list.

Intro to proofs:
Literally any good math webpage on this from MIT, Berkeley, etc.
Concrete math by Knuth, chapters 1-3

(Basic) probability theory:
A First Course in Probability by Ross is fine

Linear algebra:
Linear algebra done wrong (free here https://www.math.brown.edu/streil/papers/LADW/LADW.html)

Graph theory:
Bollobas or Diestel. Either are great

Combinatorics:
Easy but good: http://www.math.ucsd.edu/~ebender/CombText/
Pain but very good: Enumerative Combinatorics by Stanley, Vol I

Algebra:
Artin or Dummit/Foote. Either are fine. Read up to Galois theory

Analysis (optional but heavily recommended):Bare minimum: Understanding analysis by Abbott, up to chapter 4 probably

Logic:
Enderton

Automata:
Easy but very clear: Sipser
Pain but massive mathematical insight: Elements of Automata theory by Sakarovith

Algorithms:
Jeff Erickson (free here, very good https://jeffe.cs.illinois.edu/teaching/algorithms/book/Algorithms-JeffE.pdf)

Computability/Recursion theory:
Turing Computability by Soare

Type theory:
Pierce
Category theory for programmers (free here, https://github.com/hmemcpy/milewski-ctfp-pdf)
Homotopy Theory Theory (free online)

Complexity theory:
Arora-Barak
The complexity theory companion (https://www.cs.rochester.edu/u/lane/=companion/). Very good, and organizes problem by technique rather than topic.

(cont)

Anonymous

Anonymous Mon 22 Mar 2021 08:40:17 No.12866783 Report

Quoted By: >>12866792

>>12866781
(cont)
Probability theory:
Concentration of Measure for Analysis of Algorithms (free here, very good read http://wwwusers.di.uniroma1.it/~ale/Papers/master.pdf)
Properly rigorous in many ways, but skips probability in full generality. The aim of the book is to collect analytic techniques in probability that show up a lot in CS. You can worry more about measure later if you’re in ML theory or other areas where it’s more necessary.

Analysis of Boolean functions:
O’Donnell (free here, very good read https://www.cs.tau.ac.il/~amnon/Classes/2016-PRG/Analysis-Of-Boolean-Functions.pdf)

These barely scratch the surface and don’t even get into geometry processing (which the resources in OP and across the thread should cover), ML theory, cryptography, quantum information, error correction, etc.. It doesn't even have enough resources a proper treatment of algorithms. But I believe this list to be a good start for anybody interested in theoretical CS.

Anonymous

View Same Google iqdb SauceNAO The Arch-Wizard.png, 1MiB, 1140x4777

Anonymous Mon 22 Mar 2021 08:43:40 No.12866792 Report

Quoted By: >>12867704

>>12866781 >>12866783

Is there an
extreme hardcore level
meme guide for
Mathematical CS
equivalent to pic related?

Anonymous

Anonymous Mon 22 Mar 2021 09:34:06 No.12866899 Report

Quoted By: >>12866912

>>12841837
I'm in the first year of a cs master's program, and I'm pretty set on taking my master's in machine learning and then go on to take a Phd in an ai-related field. What should I learn on my own during the next 4 years in order to increase my chances of being admitted to a Phd program related to ai? What are some good projects I could work on, or open source projects I could contribute to, that would make me stick out among the other applicants? I know 4 years is a long time but starting early will only be beneficial

Anonymous

Anonymous Mon 22 Mar 2021 09:36:40 No.12866912 Report

Quoted By: >>12866920

>>12866899
Is your grad school 4 years long?

Anonymous

Anonymous Mon 22 Mar 2021 09:39:48 No.12866920 Report

Quoted By: >>12867017

>>12866912
Nah I'm in the first year of a program consisting of a three year bachelor's degree and then a two year master's degree, and my goal is to get a Phd after that, which will be in 4 years.

Anonymous

Anonymous Mon 22 Mar 2021 10:20:23 No.12867017 Report

Quoted By:

>>12866920
Gl anon, don't be like me and party away your GPA

Anonymous

View Same Google iqdb SauceNAO constanze-braunschbank-albrechts (...).jpg, 19KiB, 210x240

Anonymous Mon 22 Mar 2021 12:29:42 No.12867295 Report

Quoted By:

At work we use an asynchronous job that deletes tombstoned records' child resources off a series of disconnected databases. I was wondering if there was some rigorous definition of the tradeoffs involved. For example, if you desynchronize the deletion of a parent and child resource, what do you have to ensure the atomicity of the parent and child? You need to check if the children have been deleted yet to report its deletion status, so that "question" would require 2 trips (or some rigorous definition of trips given the depth of the parent-child dependency tree)

how would i even begin to look something up like this? I know someone must have come up with this idea but I don't know how to use google scholar to look for what I'm looking for

Anonymous

Anonymous Mon 22 Mar 2021 14:31:25 No.12867704 Report

Quoted By:

>>12866792
I mean probably lol. Basically if I took everything related to geometric complexity theory, ie algebraic geometry and related pure math topics, and maybe some other topics, I could probably hack together some stupid meme list.

But you forget that what makes the one you posted special: it was made by a Russian prof who unironically believes it should be standard

Anonymous

View Same Google iqdb SauceNAO 46B13D47-D4AD-4855-B46D-927BA91E (...).jpg, 51KiB, 633x474

Anonymous Mon 22 Mar 2021 18:39:09 No.12868582 Report

Quoted By: >>12868608 >>12871050

>>12866120
>>12866155
Honestly in your position, I think a project on understanding a novel cryptographic protocol and implementing it properly would be fine. It’s both heavy on math and on nontrivial software implementation. Maybe focus on curves, LWE, or homomorphic encryption. You might need to read up on some ring theory though. Just know that whatever you make will be a toy model. Aside from the Handbook, below are some basic resources - you might want to go further if you want your project to be more novel. I’ve included them in a pastebin because 4chan thinks it’s spam

Handbook of applied cryptography
https://cacr.uwaterloo.ca/hac/
Other stuff
https://pastebin.com/f7JXWKBN

Anonymous

View Same Google iqdb SauceNAO ringscheme.jpg, 23KiB, 640x428

Anonymous Mon 22 Mar 2021 18:47:10 No.12868608 Report

Quoted By: >>12868956

>>12868582
>>12866120
Check this out

https://eprint.iacr.org/2015/1098.pdf
or
https://en.wikipedia.org/wiki/Ring_signature

Anonymous

Anonymous Mon 22 Mar 2021 20:01:33 No.12868956 Report

Quoted By:

>>12868608
kek that pic

Anonymous

Anonymous Mon 22 Mar 2021 20:47:09 No.12869153 Report

Quoted By: >>12869211 >>12869297

>>12842085
>does anyone know anything about the state of the art?
State of the art from ~2010 was using model theory for graph limits. But everyone and their wife's son was doing graph limits then, not just model theorists, so it wasn't anything special.
Room mate went on to try and do some infinite stuff with model theory, but i have no idea about that because he became a politics-obsessed nutsack.

Anonymous

Anonymous Mon 22 Mar 2021 20:59:10 No.12869211 Report

Quoted By: >>12869246

>>12869153
oh cool, do you have any links / readings / resources?

Anonymous

Anonymous Mon 22 Mar 2021 21:09:20 No.12869246 Report

Quoted By: >>12869297

>>12869211
It was ages ago (to me, anyway). Since CNRS was involved too, you'll probably find it on Inria's HAL.

Anonymous

Anonymous Mon 22 Mar 2021 21:19:29 No.12869297 Report

Quoted By: >>12869365

>>12869153
>>12869246
Oh, you’re talking about graphons right? Didn’t lovasz have a good survey on them? I think they were used for property testing, but a lot of the interesting things related to them had to deal with the analytic questions

Anonymous

Anonymous Mon 22 Mar 2021 21:34:39 No.12869365 Report

Quoted By: >>12869397

>>12869297
Again, it was ages ago so i don't even know who worked on it, but the idea was to get a unified tool to work with limits of dense and sparse graph sequences.

Anonymous

Anonymous Mon 22 Mar 2021 21:42:03 No.12869397 Report

Quoted By: >>12869555

>>12869365
fair
what do you work in now?

Anonymous

Anonymous Mon 22 Mar 2021 22:09:38 No.12869505 Report

Quoted By: >>12869554 >>12869564

serious question. should i even bother grinding through CLRS even though i have no intent on grad school? will i ever be qualified for a job where i would actually need to know this stuff?

besides interview questions obviously.

Anonymous

Anonymous Mon 22 Mar 2021 22:17:29 No.12869554 Report

Quoted By:

>>12869505
Do algorithms by Jeff Erickson instead. Not as big but it’s much more we’ll written and not nearly as dry. Lots of great directed exercises too. It’s also completely free legally.
https://jeffe.cs.illinois.edu/teaching/algorithms/book/Algorithms-JeffE.pdf

Anonymous

Anonymous Mon 22 Mar 2021 22:17:45 No.12869555 Report

Quoted By: >>12870941

>>12869397
Digital banking security. Which means post-quantum crypto.
Before that i used to work on automated verification of lock-free algorithms.

Anonymous

Anonymous Mon 22 Mar 2021 22:18:45 No.12869564 Report

Quoted By: >>12869783

>>12869505
Kind of. Knowing how certain algorithms, and more specifically data structures work really equips you for "serious" software engineering. But yeah, interviews is the most important bit.

Anonymous

Anonymous Mon 22 Mar 2021 23:00:32 No.12869783 Report

Quoted By: >>12870243

>>12869564
I suspect that as software engineering matures, there will be fewer trivial problems dealing with elementary systems. I think there will be more mathematical CS problems that show up naturally in SWE, and thus it'll demand better engineers who know proper CS and math.

Anonymous

Anonymous Tue 23 Mar 2021 00:48:41 No.12870243 Report

Quoted By: >>12870318

>>12869783
SWE is at least 30 years old. the main motivator for adopting a system is either for performance reasons, or ease of use. as it matured, it's likely that the amount of people working on trivial problems grew much faster than the people working on the more intricate stuff
i've heard a lot of ML engineers claim "people are starting to realize it's not actually that difficult to do ML grunt work, we're probably going to see the standards for hiring lower"

i think those technical challenges will exist, but will be overshadowed by the sheer amount of grunt work

Anonymous

Anonymous Tue 23 Mar 2021 01:06:17 No.12870318 Report

Quoted By: >>12870468

>>12870243
I mean sure, but the problem is that the more sophisticated the tools become, the more attention people will have to pay to engineer using them. While grunt work might explode, I think SWE senior positions will be in dire need of people with both academic and industrial prowess.
What I’m trying to say is that I think the stories of “I’m a senior engineer, and I never had to use that [basic math in CS harder than arithmetic] shit,” are going to get sparser

Anonymous

View Same Google iqdb SauceNAO file.png, 284KiB, 350x457

Anonymous Tue 23 Mar 2021 01:38:58 No.12870468 Report

Quoted By: >>12870510

>>12870318
Most senior software engineers spend their time doing system design and grunt work (quickly without mistakes) instead of developing any mathematical rigor around the systems the build. It's pretty much - does it work given the constraints? Yes? Ship it.

Not to say that engineering systems that handle specific problems with difficult constraints is not impressive (see zanzibar, a high throughput globally distributed set computation system)

Anonymous

Anonymous Tue 23 Mar 2021 01:48:39 No.12870510 Report

Quoted By: >>12870603

>>12870468
Lmao it depends on the subfield of software engineering

Anonymous

Anonymous Tue 23 Mar 2021 02:11:31 No.12870603 Report

Quoted By:

>>12870510
and company. People only get hired as senior because they demand a certain pay. I was able to turn a job offer into a senior by simply asking for more $.

senior doesn't mean shit, especially in startup hell

Anonymous

Anonymous Tue 23 Mar 2021 02:38:34 No.12870682 Report

Quoted By:

Pleasantly surprised with how positive the CS thread has been so far. Usually you see way more vitriol and shitposting

Anonymous

Anonymous Tue 23 Mar 2021 04:25:44 No.12870941 Report

Quoted By: >>12871317

>>12869555
What kind of tech/theory is involved in the post-quantum crypto space?

Anonymous

Anonymous Tue 23 Mar 2021 04:48:46 No.12871008 Report

Quoted By: >>12871131

How exactly do you become good at Theory of computation? I think I understand the concepts of finite automata, context free languages, pushdown automata but when it comes to exam questions, a lot of it comes down to making these FA or NDPAs etc etc which seems like it requires ingenuity or some trick..

Is there something I am missing? How do I become good at this shit? Is it just a matter of practice and doing many questions so that I will have them all memorized? I've been re-reading my notes on structural induction and graph theory trying to see if there is something I missed..

Anonymous

Anonymous Tue 23 Mar 2021 05:10:32 No.12871050 Report

Quoted By: >>12871117

>>12868582
The problem is the faculty here is expecting us to improve upon an algorithm. Isn't it quite hard to do for undergrads? They kept throwing off Corpo-buzzwords everywhere so I guess my uni is a meme kek.

Anonymous

Anonymous Tue 23 Mar 2021 05:38:08 No.12871117 Report

Quoted By: >>12871263

>>12871050
> Isn't it quite hard to do for undergrads?
...lmao yeah. Making any real improvement to an algorithm in a nontrivial way is tough and usually warrants a paper to be published at a major conference. Imagine telling an undergrad to find better lower bounds for Boolean functions when this is a big open problem in CS and requires lots of technical tricks involving Fourier analysis, etc..
Unless by "algorithm" they mean "implementation of an algorithm?"

Anonymous

Anonymous Tue 23 Mar 2021 05:40:53 No.12871131 Report

Quoted By:

>>12871008
> which seems like it requires ingenuity or some trick..
keep solving problems and draw out everything you do. It's literally just a matter of practice and experience.
For example, for CFGs and regular languages, using the pumping lemma to show languages aren't either is literally a matter of experience and writing it out.
For computability, lots of problems end up being understanding the full extent of reducing to the halting problem.

Anonymous

Anonymous Tue 23 Mar 2021 06:24:52 No.12871263 Report

Quoted By:

>>12871117
> implementation of an algorithm
They're not clear on this, perhaps it's due to their lack of expertise in the field.
I doubt they'll even be able to understand the maths since the profs here lack that background too.

Anonymous

Anonymous Tue 23 Mar 2021 06:48:23 No.12871317 Report

Quoted By: >>12877338 >>12877414 >>12878728

>>12870941
>tech
I wouldn't know. Computers don't have enough qubits yet.
>theory
There's too much to give a complete list. I personally use some algeo, some algebraic number theory, a lot of harmonic anal and repre theory, a lot of lattice theory.

Anonymous

Anonymous Tue 23 Mar 2021 07:20:32 No.12871395 Report

Quoted By: >>12871523

How do you explain interactive proof system to an undergrad?
I'm assuming they don't know about NP yet.

Anonymous

Anonymous Tue 23 Mar 2021 08:16:50 No.12871523 Report

Quoted By:

>>12871395
It’s not hard if you can properly explain Arthur Merlin games using the all powerful but untrustworthy wizard analogy

Anonymous

Anonymous Tue 23 Mar 2021 15:43:39 No.12872677 Report

Quoted By: >>12872733 >>12872827 >>12874700

how do i focus to study i cant see seem to do it

Anonymous

Anonymous Tue 23 Mar 2021 15:58:16 No.12872733 Report

Quoted By:

>>12872677
Coffee and closing out the distractions

Anonymous

Anonymous Tue 23 Mar 2021 16:02:41 No.12872765 Report

Quoted By: >>12872817 >>12872954

>>12851559
No. Classes in general are worthless for your CV btw. No one really cares about them.

Anonymous

Anonymous Tue 23 Mar 2021 16:18:34 No.12872817 Report

Quoted By:

>>12872765
nah it depends. they matter if they launch you into research or a large project. My independent study in algebraic topology launched me into stuff relating to the simplex algorithm and a novel way to use it, which I got to write about and implement. It was a very good thing to have on my resume looking for work since it was hard evidence I could synthesize my theoretical knowledge into practical work, and it was interesting to talk about during the interviews. For the CV, it was another tick of research in my favor.

Anonymous

Anonymous Tue 23 Mar 2021 16:22:12 No.12872827 Report

Quoted By:

>>12872677
>turn off your router
>turn off your phone
>take pen and notebook
>open book
>read
>do the exercises
>revise exercises couple times a week

Anonymous

Anonymous Tue 23 Mar 2021 17:04:16 No.12872954 Report

Quoted By: >>12873308

>>12872765
what do they care about then? just the major? I'm a math major so I'm trying to do some damage control

Anonymous

Anonymous Tue 23 Mar 2021 18:29:47 No.12873308 Report

Quoted By: >>12875579

>>12872954
they care about your experience. classes are nice but not enough

Anonymous

Anonymous Tue 23 Mar 2021 18:48:55 No.12873409 Report

Quoted By:

>>12853232
>OpenFOAM solver and Paraview for post processing, its opensource
Yeah I've used this, but I couldn't find a way to view rigid bodies (it crashes when I try to open the files, but maybe that is just because there are too many triangles).

Anonymous

Anonymous Wed 24 Mar 2021 00:09:01 No.12874700 Report

Quoted By:

>>12872677
Library boi

Anonymous

View Same Google iqdb SauceNAO ExGLb4HWUAI2f1D.png, 21KiB, 819x150

Anonymous Wed 24 Mar 2021 00:15:52 No.12874721 Report

Quoted By: >>12874739 >>12874900

Anonymous

Anonymous Wed 24 Mar 2021 00:20:29 No.12874739 Report

Quoted By: >>12876446

>>12874721
Wrong thread. Mochizuki has no power here

Anonymous

Anonymous Wed 24 Mar 2021 01:22:03 No.12874900 Report

Quoted By:

>>12874721
still better than Shelah

Anonymous

Anonymous Wed 24 Mar 2021 06:05:47 No.12875579 Report

Quoted By: >>12875704

>>12873308
well that's good news, do projects count? or just internships?

Anonymous

Anonymous Wed 24 Mar 2021 07:01:28 No.12875704 Report

Quoted By: >>12877666

>>12875579
Projects count but not as much as internships

Anonymous

Anonymous Wed 24 Mar 2021 09:32:48 No.12876093 Report

Quoted By:

Interesting talk about automata on infinite words
Here's an interesting video on automata on infinite words
https://www.youtube.com/watch?v=pmMkLHIaUaE&ab_channel=SimonsInstitute

Here's the full playlist:
https://simons.berkeley.edu/talks/automata

Anonymous

Anonymous Wed 24 Mar 2021 11:57:22 No.12876446 Report

Quoted By:

>>12874739
Kek

Anonymous

Anonymous Wed 24 Mar 2021 14:11:11 No.12876805 Report

Quoted By: >>12877111 >>12877342 >>12877816

What open problems in CS are possible for an undergrad to do? Also any links on a collection of open problems?

Anonymous

Anonymous Wed 24 Mar 2021 15:52:56 No.12877111 Report

Quoted By: >>12877690

>>12876805
Why would undergrads be able to do open problems

Anonymous

View Same Google iqdb SauceNAO dsfgh.png, 783KiB, 1402x726

Anonymous Wed 24 Mar 2021 16:56:55 No.12877307 Report

Quoted By: >>12879040

Andrew Ng machine learning talk and Q&A.

Starting now.

The link is here.

https://www.youtube.com/watch?v=06-AZXmwHjo

Feel free to join, and now is a good opportunity to ask Andrew any questions you may have.

Anonymous

Anonymous Wed 24 Mar 2021 17:06:04 No.12877338 Report

Quoted By: >>12877426

>>12871317
>algeo
algebraic geometry? not that anon, but i’m doubling in math + cs and have similar research interests. sounds like I should be aiming for a math masters?

Anonymous

Anonymous Wed 24 Mar 2021 17:07:29 No.12877342 Report

Quoted By: >>12877690

>>12876805
retard undergrad personifying dunning kruger: the post

Anonymous

Anonymous Wed 24 Mar 2021 17:25:10 No.12877414 Report

Quoted By: >>12878968

>>12871317
>a lot of harmonic anal and repre theory, a lot of lattice theory.
I know some actual books for these topics, but could you link some papers / books in crypto that use these?

Anonymous

Anonymous Wed 24 Mar 2021 17:26:59 No.12877426 Report

Quoted By: >>12877451

>>12877338
no in the US, you don't get a masters. you just go to a PhD for research instead.
Usually to get into these areas in a way that you're doing research and not engineering is to get the PhD

Anonymous

Anonymous Wed 24 Mar 2021 17:31:46 No.12877451 Report

Quoted By: >>12877495

>>12877426
im pretty naive on this topic, but wouldn’t the masters level coursework make you a stronger candidate for a PhD? of course research is the most important, but all else equal.

Anonymous

Anonymous Wed 24 Mar 2021 17:45:04 No.12877495 Report

Quoted By:

>>12877451
the coursework? absolutely, but you can take multiple graduate level courses in undergrad anyway. I did that for both math and CS - this saves you money against the masters, shows people you're a go getter, etc., and it usually leads to good research.

Anonymous

Anonymous Wed 24 Mar 2021 18:38:37 No.12877666 Report

Quoted By: >>12877862

>>12875704
I have a split between wanting to do internships and wanting to do research opportunities, do you have any heuristics to offer for deciding between the two? Part of me thinks I should go to grad school after undergrad, the other part of me just wants to make cash and gtfo of school. I guess I could do internships anywhere in fall-spring and research in the summer?

And when am I ready for internships or research anyway? I haven't even taken intro analysis yet, so I don't think anyone would want me as a research assistant this summer, right? And I'm not super mature as a programmer, and would probably want to tackle a couple larger projects by myself first before I embarrass myself and have an internship go south.

I have tons of work to do this summer, unrelated to uni (trying to build and run small biz), but I'm trying to get my game plan down for after that.

Anonymous

Anonymous Wed 24 Mar 2021 18:46:08 No.12877690 Report

Quoted By: >>12877816 >>12877852

>>12877111
Curiosity? I heard graph theory and combinatorics is the most accessible field for open problems, and they're heavily related to CS.
>>12877342
I was just asking...

Anonymous

Anonymous Wed 24 Mar 2021 19:14:28 No.12877816 Report

Quoted By: >>12877840

>>12876805
>>12877690
Open combinatorics problems are usually in the "easy to state the problem but difficult to solve" group. They are mostly recreational math (the ones Paul Erdos usually likes to spend his time on). They are not really interesting, theory-wise, if you want to pursuit a career in academia. Most of them are not practical problems so if you want to work in the industry it's kind of time-wasting.
Most people don't even care about things like these:
https://en.wikipedia.org/wiki/Happy_ending_problem
But if you still want to work on open problems to expand your mind, here they are:
http://www.openproblemgarden.org/category/graph_theory
Each problem has a "recommended for undergrad" flag. Most of the flags are a big no.

Anonymous

Anonymous Wed 24 Mar 2021 19:18:50 No.12877840 Report

Quoted By:

>>12877816
>They are mostly recreational math (the ones Paul Erdos usually likes to spend his time on). They are not really interesting, theory-wise, if you want to pursuit a career in academia.
This isn't true though. Most people have this view of combinatorics because it attracts the most recreationally minded people out there, but if you follow anything like Stanley or Van-Lint Wilson, it's clear that they're very interesting theory-wise and have incredible practicality and far reaching implications.

Anonymous

Anonymous Wed 24 Mar 2021 19:21:31 No.12877852 Report

Quoted By:

>>12877690
>I heard graph theory and combinatorics is the most accessible field for open problems, and they're heavily related to CS.
Good luck solving graph theory and combinatorics problems that haven't been solved. These problems come in two flavors: either they're completely trivial, or they're balls to the walls, tear your hair out as you spend weeks trying anything that might work difficult.

Anonymous

Anonymous Wed 24 Mar 2021 19:22:33 No.12877862 Report

Quoted By: >>12881965

>>12877666
it literally just depends on what you wnat to do. Just know that research experience is invaluable for grad school but you can do it during the year. You might be able to continue it during an internship too

Anonymous

Anonymous Wed 24 Mar 2021 23:20:55 No.12878728 Report

Quoted By: >>12878968

>>12871317
seconding the others on wanting more books

Anonymous

View Same Google iqdb SauceNAO formal_methods.jpg, 24KiB, 330x499

Anonymous Wed 24 Mar 2021 23:32:52 No.12878764 Report

Quoted By: >>12881508

Are there any good job opportunities for people who specialize on formal methods and software verification in general?

Anonymous

Anonymous Thu 25 Mar 2021 00:28:42 No.12878968 Report

Quoted By: >>12879162 >>12888390

>>12877414
>>12878728
>papers
Sure, when i get back home (work trip atm, don't have bookmarks on phone and don't feel like searching for it) i'll make a repo on gitlab with a compilation of papers to read.
>books
I'm not aware of any good books and i'm skeptical there can be such a thing, it's just too broad a field and requires too much depth.
Crypto in general is mostly about reading papers. Post-quantum is very interdisciplinary so you'll have to watch quant-ph, math-ph, math and CoRR. Most of the "interesting" stuff will be in quant-ph, at least while quantum computers are just toys.

Anonymous

Anonymous Thu 25 Mar 2021 00:44:49 No.12879040 Report

Quoted By: >>12879125

>>12877307
I went to this

TLDR for everyone

Basically, he's saying 99% of machine learning research goes torwards the models rather than torwards data collection/cleaning. Whereas studies have shown that by improving the data end of things, the outcome is waayyy more likely to improve than by improving the model.

SO, he's advocating for more data-centric machine learning, rather than model-centric.

Anonymous

Anonymous Thu 25 Mar 2021 01:09:18 No.12879125 Report

Quoted By:

>>12879040
Yeah I kind of understand why he would say that.
Prepping nice data basically makes less work for your models.

Anonymous

Anonymous Thu 25 Mar 2021 01:20:42 No.12879162 Report

Quoted By: >>12879288

>>12878968
>I'm not aware of any good books and i'm skeptical there can be such a thing, it's just too broad a field and requires too much depth
yeah I thought as much. It doesnt look like a field that could have a single or even collection of "tomes" dedicated to just the topic

Anonymous

Anonymous Thu 25 Mar 2021 02:00:32 No.12879288 Report

Quoted By: >>12879439

>>12879162
>doesnt look like a field that could have a single or even collection of "tomes" dedicated to just the topic
Well, there is Goldreich, but it's... Not good. Not in the way i expect a textbook to be - Munkres for example.
It's not a bad book, it does a decent job of conveying the convoluted mess that is crypto, but it's still not enough for me to recommend it.

Anonymous

Anonymous Thu 25 Mar 2021 02:53:28 No.12879439 Report

Quoted By: >>12880105

>>12879288
That’s fair. I anticipate reading the papers soon lol

Anonymous

Anonymous Thu 25 Mar 2021 06:00:25 No.12879877 Report

Quoted By:

http://www.chrishecker.com/Physics_References Is this fine for a beginner to follow or should I use more up-to-date resources? Last edited 2007.

https://perso.liris.cnrs.fr/nicolas.pronost/UUCourses/GamePhysics/literature.html
I will use something from here to learn C/C++:
http://iso-9899.info/wiki/Main_Page
https://hero.handmade.network/forums/code-discussion/t/982-books

Anonymous

Anonymous Thu 25 Mar 2021 07:34:47 No.12880105 Report

Quoted By:

>>12879439
Any progress?

Anonymous

Anonymous Thu 25 Mar 2021 12:41:08 No.12880924 Report

Quoted By: >>12881109 >>12882318

>>12841837
anyone have the serge lang chart?

Anonymous

Anonymous Thu 25 Mar 2021 13:56:13 No.12881109 Report

Quoted By: >>12881558

>>12880924
What do you mean?

Anonymous

Anonymous Thu 25 Mar 2021 15:34:41 No.12881432 Report

Quoted By:

>>12866368
Do more algebra

Anonymous

Anonymous Thu 25 Mar 2021 15:55:08 No.12881508 Report

Quoted By: >>12884705

>>12878764
Lots of industry applications of formal verification lmao.

Anonymous

View Same Google iqdb SauceNAO 1560944356389.png, 1MiB, 865x1645

Anonymous Thu 25 Mar 2021 16:08:24 No.12881558 Report

Quoted By: >>12881711 >>12882318 >>12884173

>>12881109
it's one of those book chart pics like picrel for example, but it's just full of books written by serge lang. I forgot to save it last time I saw it.

Anonymous

Anonymous Thu 25 Mar 2021 16:42:28 No.12881711 Report

Quoted By: >>12882564

>>12881558
lmao. This chart has no pedagogical value, and it has some meme books (I like GEB but it's not relevant, same goes for the philo books like Hegel), but a lot of these books like generatingfucntionology, Vazirani's Approx. Algos, Arora-Barak, Alon-Spencer, etc., are in fact very very good.

TAOCP is a bit overkill given the rest of the list (great book but you need more math background to get the fullest out of it), and field theory is entirely relevant to CS unless you want to do computational physics or study the interface between TCS and physics theory...and hell, it's actually more useful to know QFT and operads than classical theories for that.

Anonymous

Anonymous Thu 25 Mar 2021 17:36:22 No.12881965 Report

Quoted By: >>12882019

>>12877862
OK, and at what point should I consider myself ready for research?

Anonymous

Anonymous Thu 25 Mar 2021 17:45:16 No.12882019 Report

Quoted By: >>12882236

>>12881965
what do you mean? It's a rolling process. You're always "ready" but experience is both valuable to get a taste of what it feels like solving problems with no clear solutions and valuable as a way to show other people you can do it with established researchers.

Anonymous

Anonymous Thu 25 Mar 2021 18:29:26 No.12882236 Report

Quoted By: >>12882261

>>12882019
i guess I just don't get how an undergrad brain can be useful other than for getting coffees. Like how would I be useful at all to a professor who's done PhD level coursework?

For CS, I can see how I could help in some programming project, because that's just building stuff. Maybe I'll try to get some CS research instead, idk

Anonymous

Anonymous Thu 25 Mar 2021 18:33:20 No.12882261 Report

Quoted By: >>12883060

>>12882236
>Like how would I be useful at all to a professor who's done PhD level coursework?
...by reading up, running simulations, making key insights, asking for help when a technical part comes up, doing parts of the books used in PhD courses, etc etc..
>For CS, I can see how I could help in some programming project, because that's just building stuff.
It's the exact same in math as it is in theoretical computer science.
>Maybe I'll try to get some CS research instead, idk
any research is fine.

Research and solving problem is a function of many things. Yes, you need to be able to impart deep insight and technical prowess to solve problems, but really the heavy lifting comes down to being very familiar with the common literature and work in the field. All research, and even mathematics, is about synthesizing old knowledge and then bringing in your own insights. You will learn how to do the latter from your classes and with your PI. Your part in the research, however, is learning how to do the former.

Anonymous

Anonymous Thu 25 Mar 2021 18:37:38 No.12882287 Report

Quoted By:

Guys. Remember to get enough sleep.
I was doing some research problems and it took me days without any result.
I said fuck it and went to bed early. The answer instantly came to me the day later.

Anonymous

View Same Google iqdb SauceNAO WhatsApp Image 2021-03-25 at 21. (...).jpg, 152KiB, 1366x905

Anonymous Thu 25 Mar 2021 18:43:32 No.12882318 Report

Quoted By: >>12882326 >>12882564 >>12882840 >>12883821

>>12880924
>>12881558
Here you go, don't use it.

Anonymous

Anonymous Thu 25 Mar 2021 18:45:17 No.12882326 Report

Quoted By:

>>12882318
>whatapp image

Anonymous

Anonymous Thu 25 Mar 2021 19:31:24 No.12882564 Report

Quoted By: >>12882834 >>12883151

>>12882318
wowie thanks anon. why shouldn't I use it? is it a meme chart. I'm doing basic mathematics rn and I like it so far.

>>12881711
Yeah I had a feeling this was a meme one overall but it's the quickest example I could get. Thanks for letting me know the good ones!

Anonymous

Anonymous Thu 25 Mar 2021 20:37:44 No.12882834 Report

Quoted By:

>>12882564
They’re mostly good. Just avoid these ones:
All the philosophy books (they’re fine but not good for learning CS)
Field theory (good read but not immediately relevant, also read classical mechanics first anyway)
The AI book (I think most famous teacher webpages do basic AI way better than that book)
GEB
Real and complex analysis (do baby Rudin first lol, papa is better but not suitable for a beginner)

I have nitpicks for the rest, but all in all the remaining books make for a pretty good reading list, though not comprehensive.

Anonymous

Anonymous Thu 25 Mar 2021 20:38:46 No.12882840 Report

Quoted By:

>>12882318
Lol'ed

Anonymous

Anonymous Thu 25 Mar 2021 21:28:45 No.12883060 Report

Quoted By:

>>12882261
thanks anon

Anonymous

Anonymous Thu 25 Mar 2021 21:54:32 No.12883151 Report

Quoted By:

>>12882564
>why shouldn't I use it
Because, as with all charts or book guides, it has no value. Just focus on the book you have, Basic Mathematics is a very nice book, you will learn quite a lot just by going through it and doing exercises. What you do after you are satisfied with Basic Math is something you can, and should, worry about later.

Anonymous

Anonymous Fri 26 Mar 2021 00:20:50 No.12883821 Report

Quoted By:

>>12882318
lel

Anonymous

View Same Google iqdb SauceNAO Sci's CS reading list .jpg, 3MiB, 2500x4048

Anonymous Fri 26 Mar 2021 02:18:43 No.12884173 Report

Quoted By: >>12884367 >>12884459

>>12881558
better list

Anonymous

Anonymous Fri 26 Mar 2021 03:44:23 No.12884367 Report

Quoted By: >>12886419

>>12884173
eh not really. Same problem in that there are good books here but no good pedagogy. Also a good amount of extraneous, redundant things. You don't need a book to learn C++. You don't need a book to "program in the UNIX environment." There are better graphics textbooks. Hoffman and Kunze or linear algebra done wrong is better for LA. Linop shouldn't be a full book, it's too easy. Design patterns in software eng, not CS. The AI book should be replaced with a good course. There are way better quantum computing books out there, use either Kitaev or Mike'n'Ike for that. .Use Stanley or Van Lint - Wilson for combinatorics, diestel or bollobas for graph theory.
"matrix analysis and applied linear algebra", lin op, and "numerical linear algebra' don't need to be here if you just chose a good proof based LA book.

There are more problems, but those are the big ones. There are some good books in there but having good books isn't enough to have a good curriculum. Second, as much as I love theory since I study it, there's a disturbing lack of implementation and systems work. You can't learn systems from reading a book alone.

Anonymous

Anonymous Fri 26 Mar 2021 04:42:43 No.12884459 Report

Quoted By: >>12884461 >>12884541

>>12884173
i really dislike the textbook flowchart stuff here, it's pure procrastination masquerading as some temple of knowledge you're supposedly going to conquer, despite not having even started yet..

Anonymous

Anonymous Fri 26 Mar 2021 04:44:09 No.12884461 Report

Quoted By: >>12884558

>>12884459
they're not bad in and of themselves, but I'd appreciate if they actually motivated them or put them together into a cohesive curriculum. Otherwise, it's just masturbating to authors

Anonymous

Anonymous Fri 26 Mar 2021 05:19:28 No.12884541 Report

Quoted By: >>12884558

>>12884459
Totally agreed. I think it is called "action faking", things that feel productive and easy but are totally useless in the bigger picture. Better to decide what book to read after you are done with what you have.

Anonymous

Anonymous Fri 26 Mar 2021 05:28:21 No.12884558 Report

Quoted By: >>12884561 >>12884638

>>12884461
they are bad in the sense that any time spent worrying about such is ridiculous. and the amount of posts here dedicated to these charts demonstrates that far too many anons. all their doing is creating a grandiose obstacle, it serves no purpose to entrench yourself or your education path so heavily.

i know what it feels like to be in that mindset and I've been there, and that mindset for me was associated with big time daydreaming about how badass I'm going to be someday.

>>12884541
right, even a "bad" book will still teach you want to learn, it's not like half the calculus are total bogus or something lmao. a good learner learns from available sources.

Anonymous

Anonymous Fri 26 Mar 2021 05:29:39 No.12884561 Report

Quoted By:

>>12884558
im a bit buzzed so forgive the typos, you meant what I knew anyway

Anonymous

Anonymous Fri 26 Mar 2021 06:23:43 No.12884638 Report

Quoted By: >>12884652

>>12884558
I mean yeah, but I see a lot of the better ones that aren't on /sci/ (think the sparse teachyourselfcs list for programmers who want more CS knowledge) are generally good and seem more like a repository of resources. I get fantasizing about reading those books, but I feel such lists are really great for people who have 3 semesters of formal education before attempting them.

Anonymous

Anonymous Fri 26 Mar 2021 06:30:36 No.12884652 Report

Quoted By: >>12884665

>>12884638
The way to use teachyourselfcs isn't to plot out some big spread sheet. it's to think, 'hey, I wanna learn about networking, what did that trusted source recommend?" and then check it out.

my problem with those book list is that you'll know way better AFTER you read the first book what you should, so you get severely diminishing returns after the first book.

I get that they provide a comfortable sense of guidance, but so do university cirriculums, right? and they read, on average, like 1/10 the books I see in these cirriculums here. my freshman physics year was one single book. calculus I-IV was one book. diff eq, 1 book. lin alg 1 book. that's 4 books for half an undergrad cirriculum, and all I'm taking from here is analysis, topology and abstract algebra so add 3 more books for a total of 7. 7 books for 4 years of studying. I mean just think about that when you see the next chart with a dozen plus books.

just do the thing, you'll know better when you're done

Anonymous

Anonymous Fri 26 Mar 2021 06:35:41 No.12884665 Report

Quoted By: >>12884707 >>12884722

>>12884652
>I get that they provide a comfortable sense of guidance, but so do university cirriculums, right?
I think the idea for better lists is to provide reference and to be better than university curricula. CS undergrad in universities ends up being fairly sparse if you lie anywhere on the more academic spectrum.

That being said, most of these lists are posturing and offer no value whatsoever.

Anonymous

Anonymous Fri 26 Mar 2021 06:53:11 No.12884705 Report

Quoted By:

>>12881508
It doesn't seem to be that many job offers about it.

Anonymous

Anonymous Fri 26 Mar 2021 06:54:19 No.12884707 Report

Quoted By: >>12884722

>>12884665
i understand the intent may be noble, but the list tend toward extreme verbosity, probably so the author feels badass, and the students choose the bigger obstacle so they can envision themselves being more badass upon completion (while not actually getting anything done, and instead shopping for new flowcharts as we see the poor anon above doing).

it's all just super toxic, and we see these recurring threads of procrastinators pooling together and mentally masturbating and ejaculating all over one another.I guess I wouldn't mind the list so much if the standard was like 4 reasonable books that a brainlet could do. I don't look upon my book-chart days fondly and I'm glad I escaped them, my personality type is super susceptible to that shit

Anonymous

View Same Google iqdb SauceNAO ac1413f.gif, 2MiB, 400x284

Anonymous Fri 26 Mar 2021 07:04:36 No.12884722 Report

Quoted By: >>12884734 >>12884796

>>12884707
>>12884665
Thank you guys for all the input! I'm so new to this I wasn't sure what to do.

I will just finish basic mathematics and worry about other stuff later.

Anonymous

Anonymous Fri 26 Mar 2021 07:09:55 No.12884734 Report

Quoted By:

>>12884722
what's your goal?

Anonymous

Anonymous Fri 26 Mar 2021 07:38:28 No.12884796 Report

Quoted By:

>>12884722
just do'nt get lost in studying prereqs, I'd never read Lang's Basic Math's personally, except for fun on a vacation or something. the algebra you need for calc/linear algebra/diff eq is covered in the first few chapters of any high school / non-top-10 uni calculus book like stewarts Calculus books. I'd try opening those up and checking out the prelimiaries chapters.

i assume you want to learn calculus and lower-div undergrad math at least. I'd also recommend the first two chapters of spivak's caluclus for some hard mode upper-level algebra problems

Anonymous

Anonymous Fri 26 Mar 2021 15:19:03 No.12885901 Report

Quoted By: >>12885913

Is spending 30 hours per chapter for a book normal? (Included some exercises)
Usually when I'm taking a course I will just look at the lecture notes and look at relevant paragraphs in the main book a few time.
But I have never actually finished a textbook in my life so I don't know if 30 hours is slow or fast.
I feel like I don't know which section to grind and which to gloss over so it takes me longer than normal courses.

Anonymous

Anonymous Fri 26 Mar 2021 15:21:57 No.12885913 Report

Quoted By: >>12885947

>>12885901
It really just depends on you. if you can solve, if not all, exercises satisfactorily and feel like you can grasp the material in other places you see it, you've probably synthesized it well enough

Anonymous

Anonymous Fri 26 Mar 2021 15:26:39 No.12885929 Report

Quoted By:

>Have a meeting in 45 minutes
>Already prepared for it well
>Be anxious and can't do anything else, not even reading paper
What's wrong with me?

Anonymous

Anonymous Fri 26 Mar 2021 15:30:34 No.12885947 Report

Quoted By:

>>12885913
Thanks I will try some exercise until I feel like I have internalized the main concepts.

Anonymous

View Same Google iqdb SauceNAO 946a725e07427bfa471790eed33e5e2b (...).jpg, 30KiB, 720x834

Anonymous Fri 26 Mar 2021 15:37:45 No.12885985 Report

Quoted By:

Metaheuristics > Deep Learning

Anonymous

Anonymous Fri 26 Mar 2021 17:23:17 No.12886419 Report

Quoted By: >>12886962 >>12887507

>>12884367
>You don't need a book to learn C++.
dumb
>You don't need a book to "program in the UNIX environment."
dumb
>Hoffman and Kunze or linear algebra done wrong is better for LA.
dumb
>Linop shouldn't be a full book, it's too easy.
dumb
>Design patterns in software eng, not CS.
dumb
>"matrix analysis and applied linear algebra", lin op, and "numerical linear algebra' don't need to be here if you just chose a good proof based LA book.
holy shit you're dumb
>there's a disturbing lack of implementation and systems work
There's architecture x2, networks, os, compilers. What do you think is missing?

Anonymous

Anonymous Fri 26 Mar 2021 19:08:35 No.12886962 Report

Quoted By:

>>12886419
>dumb
>I need books to do teach me systems, knowledge and intuition that is the farthest from something a book can “teach”
Unless youre willing elucidate, it’s pretty apparent that you’re the dumb one here. Systems books are much better as references but not as learning tomes

Anonymous

Anonymous Fri 26 Mar 2021 20:50:39 No.12887507 Report

Quoted By:

>>12886419
>holy shit you're dumb
unless "matrix analysis and applied linear algebra" is some brainlet name for functional analysis, this list recommends like 3 undergrad linear algebra books when you only need a single one to advance to more advanced mathematics and even more advanced applications. And even then, there would be way better functional analysis books. Your list is stupid.

Anonymous

View Same Google iqdb SauceNAO 2021-03-26_17-43-02.jpg, 373KiB, 1540x745

Anonymous Fri 26 Mar 2021 22:44:44 No.12887983 Report

Quoted By: >>12887992

I don't get this proof at all.
Isn't L already in EXP by definition? Why do they need to prove the highlighted part?

Anonymous

Anonymous Fri 26 Mar 2021 22:46:30 No.12887992 Report

Quoted By:

>>12887983
Nevermind I was just drunk on coffee.

Anonymous

Anonymous Sat 27 Mar 2021 00:32:34 No.12888390 Report

Quoted By: >>12888405

>>12878968
Any word on those papers?

Anonymous

Anonymous Sat 27 Mar 2021 00:38:28 No.12888405 Report

Quoted By: >>12888455

>>12888390
i will be returning home on thursday, i'm pretty sure this thread will still be up by then - it's a slow board
have patience

Anonymous

Anonymous Sat 27 Mar 2021 00:52:53 No.12888455 Report

Quoted By: >>12889469

>>12888405
this thread is pretty close to the bump limit, so in the case that this one gets archived, I'll make a new one

Anonymous

Anonymous Sat 27 Mar 2021 07:19:53 No.12889469 Report

Quoted By:

>>12888455
Thanks. Link here please. I love this general.

Anonymous

Anonymous Sat 27 Mar 2021 12:34:03 No.12890195 Report

Quoted By: >>12890293 >>12890651 >>12890706

Is it possible to learn normal levels of math, physics, and cs theory focused in computer graphics in 3 months?

I have a very basic knowledge of these 3 fields, but I need to make a game engine from scratch and I would like to understand about 90% of what I'm doing not just copy/paste.

So my question is more in the sense that... would just taking quick math/physics courses online be enough or would I have to read the books? Also... why is that for example a linear algebra/calculus course or series of classes online contain about 4-5days worth of content but the books have like... 500 pages? Like... are those 500 pages of new content not explored during the classes or are they like derivations of the content/other ways to apply the content that can be ignored? I don't wanna get in the level of an awakened mathematician capable of seeing the world in 10 dimensions, just 4 is enough for the quaternions for the rotations I think, anyway I just would like to know if it's doable.

Anonymous

Anonymous Sat 27 Mar 2021 13:11:37 No.12890293 Report

Quoted By:

>>12890195
>why is that for example a linear algebra/calculus course or series of classes online contain about 4-5days worth of content but the books have like... 500 pages?
because the classes online are just lectures, which amounts to ~ 1/4 of workload for the class
and yes, virtually every class skips topics from its reference textbooks

Anonymous

Anonymous Sat 27 Mar 2021 15:32:41 No.12890651 Report

Quoted By:

>>12890195
Just learn as you go. You’ll find that rigid body physics will be the most immediately useful thing you learn.

Anonymous

Anonymous Sat 27 Mar 2021 15:53:44 No.12890706 Report

Quoted By:

>>12890195
3 months is enough for you to understand it but there will likely be some problems when you actually start working on the engine. But at least you know how to look up your problems by then.

Anonymous

View Same Google iqdb SauceNAO 1395047745083.jpg, 515KiB, 1920x1200

Anonymous Sat 27 Mar 2021 16:22:06 No.12890793 Report

Quoted By:

>Shortly after Euclideon released their "unlimited point cloud data" videos, a girl on youtube made some videos about how she though it worked, as did a guy from another channel. Later the videos were removed.
>Euclideon aren't hyping themselves as much as they used to after pivoting to some shitty table top 3D system.

Did Euclideon pay off or threaten these channels or is there a greater conspiracy by graphics cards manufacturers to suppress forbidden rendering knowledge?

Why do I have to draw billions of pixels per frame (as triangles), when there's only a few million pixels to render?

Anonymous

Anonymous Sat 27 Mar 2021 16:23:22 No.12890799 Report

Quoted By: >>12891076

There was a link with many books for all the branches for mathematics including arithmetic years ago but I fail to find it and it's not the stickied one.

Anonymous

Anonymous Sat 27 Mar 2021 17:50:49 No.12891076 Report

Quoted By: >>12891122

>>12890799
sorry, can't say I have a link

Anonymous

Anonymous Sat 27 Mar 2021 18:07:14 No.12891122 Report

Quoted By: >>12891154

>>12891076
I remember it being really easy to find, maybe a wiki or something but I just can't find it anymore.

Anonymous

Anonymous Sat 27 Mar 2021 18:19:27 No.12891154 Report

Quoted By: >>12891165

>>12891122
do you mean this?
https://4chan-science.fandom.com/wiki/Mathematics

Anonymous

Anonymous Sat 27 Mar 2021 18:22:21 No.12891165 Report

Quoted By: >>12891243 >>12891329

>>12891154
No, I remember that it started with early math and had many books even for just arithmetic alone.

Anonymous

Anonymous Sat 27 Mar 2021 18:38:44 No.12891243 Report

Quoted By: >>12891251

>>12891165
...many books on arithmetic alone? how many books do you need for arithmetic?

Anonymous

Anonymous Sat 27 Mar 2021 18:41:20 No.12891251 Report

Quoted By:

>>12891243
I'm just interested to see what they were about really. I remember people posting some really complex books about arithmetic but they were just about proofs I think.

Anonymous

Anonymous Sat 27 Mar 2021 19:07:45 No.12891329 Report

Quoted By: >>12893284

>>12891165
https://4chan-science.fandom.com/wiki/Math_Textbook_Recommendations#Preschool_.28Arithmetic.29

Anonymous

Anonymous Sat 27 Mar 2021 22:40:24 No.12892000 Report

Quoted By: >>12892683

Alright guys, since this is hitting the bump limit soon, I'll be taking votes (informally) on which topic to base the next thread on. I think theming each thread will give us more to talk about (and potentially attract new people). The latter will probably occur as the topics skew towards CS's relationship with the rest of academia.

Here's a short list of some possible topics:
Computer Vision
Randomized and Probabilistic Algorithms
Robotics
Quantum Information and Computing
Programming Language Theory and Compilers

Anonymous

View Same Google iqdb SauceNAO 244173ab86a1ad04301e70e44c1bf71e (...).jpg, 24KiB, 341x500

Anonymous Sun 28 Mar 2021 02:50:40 No.12892683 Report

Quoted By: >>12892780

I want to read pic related but I heard /mg/ said that Combinatorial Geometry is dead now so there's no point?
>>12892000
Computer Vision or Randomized and Probabilistic Algorithms

Anonymous

Anonymous Sun 28 Mar 2021 03:34:07 No.12892780 Report

Quoted By: >>12892814

>>12892683
>I heard /mg/ said that Combinatorial Geometry is dead now so there's no point?
lmao I remember seeing that guy, but I don't know if he was right. Read whatever you think is interesting
>Computer Vision or Randomized and Probabilistic Algorithms
Cool cool, I'll decide between these two if nobody else votes.

Anonymous

Anonymous Sun 28 Mar 2021 03:53:59 No.12892814 Report

Quoted By:

>>12892780
>lmao I remember seeing that guy, but I don't know if he was right. Read whatever you think is interesting
Yeah I will ask around some more.

Anonymous

Anonymous Sun 28 Mar 2021 08:42:46 No.12893284 Report

Quoted By: >>12895707

>>12891329
Thanks but it's not this either, it had more information.

Anonymous

Anonymous Sun 28 Mar 2021 11:15:27 No.12893526 Report

Quoted By:

New thread up here: >>12893521
I'll be archiving the links discussed within this thread. I'll hopefully have something workable up soon so everyone can access them from a common page.

Anonymous

Anonymous Sun 28 Mar 2021 22:43:38 No.12895707 Report

Quoted By:

>>12893284
>Thanks but it's not this either, it had more information.
Probably bait and had Serre's A Course in Arithmetic, Jacobson's Basic Algebra, etc on it

Anonymous

View Same Google iqdb SauceNAO sci meme book chart.png, 3MiB, 3288x2872

Anonymous Sun 28 Mar 2021 22:47:23 No.12895725 Report

Quoted By:

Capcode	All Only User Posts Only Moderator Posts Only Admin Posts Only Developer Posts
Show Posts	All Only With Images Only Without Images
Deleted Posts	All Only Deleted Posts Only Non-Deleted Posts
Ghost Posts	All Only Ghost Posts Only Non-Ghost Posts
Post Type	All Only Sticky Threads Only Opening Posts Only Reply Posts
Results	All Grouped By Threads
Order	Latest Posts First Oldest Posts First

Your latest searches