What topic of CompSci (especially theoretical CompSci) is ridiculously hard to understand?
>inb4 undergrad topics that crush plebs (pointers, off-by-one errors, etc.)
I'm talking about something notoriously hard. Not beginner problems.
>>7952548
big O notation
>>7952548
>CS
>hard
>>7952563
Also, if you can't contribute, contain your shitposting.
>>7952548
Lots of people have trouble with the Pumping Lemma when they first learn it.
>>7952554
Thank you, but would you please be more specific. Complexity notation in its principles, is actually very easy.
Multivariate gradient descent can be fairly hard, but I think an actual mathematician would only consider it a moderately difficult problem.
Most of the difficulty in CS comes from figuring out what to do, rather than exactly how to do it (assuming your math is good).
>>7952568
Thank you also, but thats also not what i'm asking about. I asked for things at P-NP problem level.
Writing a compiler that makes univalence compute, i.e. implement an intuitionistic type theory with logical = equality that respects type equivalence \sim of whatever structures you look at (e.g. equivalence of group representation Z_2=({0,1},+)=({1,-1},ยท) and the like).
https://en.wikipedia.org/wiki/Homotopy_type_theory#Univalence_axiom
The categorical semantics of equivalence types are very hard to get, you need quite a lot background.
>>7952582
PS it's the open question of the field
Analytic combinatorics can be an ass-kicker for those who are not fairly strong in mathematics.
Integer optimization.
Airplanes companies would pay you a lot of cash if you can think of a better model than the current ones.
>>7952554
>big O notation
Agreed, been working in software development for over 20 years and NEVER ever needed this knowledge. I let the software library developers worry about sorting/searching optimization.
>>7953599
And that is why those developers make double what you do.
Congratulations on being a low achieving retard.
>>7953599
>big O
>hard
Are you serious? You can easily do it if you know how to do proof by induction which you learned in high school.
>>7953611
>learned in high school.
People who study CS do so specifically because they never went to high school.
There is a law that states a minimum of 60 IQ to enter high school, poor CS majors cannot do that so they have to wait until 18 and jump into universities.
>>7953613
Great joke mate :^). Back to /sci/ , oowww wait you are allready in the right retarded place.
> le physics meme
>>7953613
That explains why every horsefucker, furry, and weeb is in CS.
>>7952554
omicron**
>>7953599
Dumb asses... Only a complete idiot would waste development time to write a search algorithm or a sorting algorithm. You call a library function to do this. Sure I COULD spend my time and try to out optimize the libraries but that would be stupid. The times when Big O notation would be used is PRECISELY the times when a library function should be used. Library programmers specialize in this. Respect to them, but how boring to spend a career increasing the optimization of a few searching algorithms by 3 to 4 percent.
>>7952548
P versus NP
Is this answer cheating?
Computational Complexity theory: Space Complexity & Time Complexity
Quantum Computing
Machine Learning
Cryptography
Computational Number Theory
Computational Geometry
Computer Algebra
Artificial Intelligence
To name a few...
>>7953709
it isn't that hard really.
Can all problems that can be verified in O(n^k)(NP) also be solved in O(n^k) (P).
Everybody thinks not but no one has proved it yet.
>>7953679
>I glue interfaces together
>I r so smrats!
>>7953679
You are a living, breathing example of why there is so much shitty software floating around. I'm not really complaining though as I made a very nice living solving problems that idiots like you were too stupid to solve.
>>7953605
oh so you wrote all your sorting algorithms yourself? Did you write ALL of your algorithms yourself? You don't use anyone else's libraries?
That's what I thought bitch
>>7953802
What is it with you and sorting algorithms? Can you comprehend that when you start mixing and matching algorithms to solve larger problems that it might be useful to be well versed in the computational complexity of the library functions that you are stringing together?
I know this is a bait thread to propagate the "CS majurs r dum" meme but I'll bite anyway. Nothing I would classify as ridiculously hard to understand but here are some topics that my peers have had trouble with.
NP complete reductions
Markov decision processes
Neural network backpropagation
Logistic and gaussian models
Multivariable gradient descent in general (optimization algorithms, inverse kinematics, etc)
Radiance, radiosity, and other graphical models dealing with light
Enforcing database integrity
>>7953816
that was my first post in this thread dude I just used it as an example b/c it was the first thing that came to my head
>it might be useful
keywords buddy
It's impossible to prove whether certain algorithms will terminate according to turing.
Prove him wrong.
>>7952548
Nothing in CS is hard to understand, it's hard to do.
That's why everybody on shits on it; they understand it, but they can't do it.
>>7952554
That is considered basic, since it's always included in basic material (CLRS for example).
>>7953613
If you say CS is so easy, none of the problems yet mentioned will be hard for you. Why haven't you solved them?
>>7953831
>NP complete reductions
This. And this again.
Actual research level complexity theory is brutal.
>>7954043
u sure mate? It's pretty ez. Just read Sipser.
>>7953187
Whats wrong with rebooting your airplanes computer every 255 days?
Most CS subfields are hard once you go deep in them, but of course it really depends on your background.
Cryptography is hard.
Machine learning requires calculus, algrebra and statistics, specially if you ever hope to use it for something practical and not for some small project.
Programming languages require logic, type systems require type and category theory, etc
For me, distributed and concurrent programming is hard, not because the concepts are hard, but just because the problems being solved are hard. You have to get around physical properties of the network, understand the implications of latency and time when you have a bunch of concurrent operations on different machines.
>>7952548
The halting problem. There are intuitive proofs that it's just blatantly undecidable but obvious it can be solved in local cases so just saying "undecidable" doesn't really answer the entire question.
Technically this is the only problem in computer science.
>>7954325
I'm pretty sure he's talking about integer programming with respect to flight/baggage/etc scheduling, which is an NP-hard problem, but is useful as many combinatorial optimization problems can be modeled as IP, and those problems themselves are often NP-hard as well, so doing IP can be a reasonable thing to do. Although something like an airline would probably care more for tractable approximation algorithms if they have enough things to schedule, although those can involve things similar to IP. A lot involve linear programming directly or indirectly (relaxations, primal-dual algorithms, etc).
>>7952554
I fucking love this shit, "I don't understand this basic concept because my professors baby me".
>>7952548
I'll throw my hat in the ring.
Language Theory.
>>7952548
Stuff that's not covered in undergrad CS
>>7954333
>Machine learning requires calculus, algrebra and statistics
Are you implying that this makes machine learning hard?
Rofl, CS students never cease to amuse me
>>7955887
I thought the same thing when I read this. However, the poster is correct to a certain extent. I just think he really just doesn't know why.
For example, consider the support vector machine. I find them to be a lovely mathematical construction that essentially reduces to a quadratic programming problem. Computationally solving a QP problem is no joke and I suspect the are many math undergrads out there who have never even seen such a problem. It makes solving a problem such as, say, stochastic gradient descent, look like the first year calculus problem that it really is.
>>7953956
>Nothing in CS is hard to understand, it's hard to do.
That's why everybody on shits on it; they understand it, but they can't do it.
this explains the sci meme about comp sci, people dont realize how hard it is that things work properly
