HNHacker News
TopNewBestAskShowJobs

jksk61

57 karma · joined January 12, 2023

submissionscomments
jksk61··on Modular forms, the ‘fifth fundamental operation’ of math
then you took the wrong courses.
jksk61··on Lean, Coq and other proof assistants: Visualising proofs as trees
well, for lean I know there's a discord (with a fair amount of experts) and a fair share of resources nowadays. However, I did try 3 years ago and the jump between solving a simple algebra proof and contributing to Mathlib (or even understand something in it), is quite high.

Coq is well designed/easy to understand but it is missing a core mathlib like Lean. Moreover, most proofs are designed to "compile" quickly and not to be human readable.

jksk61··on Chinchilla’s death
maybe it is just a sierpinski triangle
jksk61··on An Introduction to Graph Theory
binary search is essentially an algorithm to find zeros, which is just algebra/analysis.
jksk61··on LaTeX3: Programming in LaTeX with Ease (2020)
it's like suggesting to use julia for ML 5 years ago instead of using Python, because it is new. Nowadays is somewhat decent but still lags behind PyTorch and none cares, same will be for Typst unless something major happens.

For example, how I'm supposed to use a markdown language that doesn't have a tikz-like library? Sure you can still write commutative diagram but that's not all tikz does.

jksk61··on Geothermal Ahead of Schedule
if they are too far away you'd have a loss of efficiency since I guess they'll transfer heat and not current, no?
jksk61··on How to Build a Big Prime Number
I guess that algorithm used to get the prime number they want to generate, involves sampling from some kind of distribution such that by Law of large numbers it converges in probability to the same thing (but of course much more advanced than that).
jksk61··on A tutorial quantum interpreter in 150 lines of Lisp
maybe not of your interest, but if one is more from a ML background, there's this nice book/overview by Maria Schuld and Francesco Petruccione, called Machine Learning with Quantum Computers, which is mostly regarding variational algorithm and quantum kernel machine. However, the best intro material is the one by Nielsen and Chuang.

Moreover, if you feel more comfortable by running examples there are qiskit, pennylane and other libraries which shares a lot of tutorials and notebook (and some can be run on a true quantum computer if you have the money or the time to wait in queue!)

jksk61··on Telling the Bees
death is scary because we decided to protray it in that way. What's scary in an endless sleep? From a sleep deprived person pov, it seems like a pretty good ending.
jksk61··on Fixed Spelling Errors or Typos
everyone makes mistakes
jksk61··on Methods for increasing efficiency of solar cells with silicon and perovskite
still lab stuff, not scaled and not that impressive since other techniques using silicon (in lab) are roughly the same from an efficiency pov.
jksk61··on Vivaldi on Android bypasses restrictions to let users access Bing Chat
Yes, but there are also better restaurant you can go to, so why wasting money, time and your taste to eat a mcdonald salad?
jksk61··on Vivaldi on Android bypasses restrictions to let users access Bing Chat
what you are missing is: 1. Keeping the hydrogen at high pressure requires a lot of energy. 2. Hydrogen produces H2O in gas form, which in itself it is still a greenhouse gas. 3. True, hydrogen can be produced by "green" technologies, but what is the energy lost throughout the procedure? It is quite a loss (about 20%).

Moreover, it is theoretically viable to produce all electric energy via renewables, albeit impracticle and 100 years away. Thus, I can't see how saying that """EVENTUALLY""" hydrogen might be renewable is good for the environment.

Hydrogen is good for a very niche part of our production system (and spoiler it is not car industry or house heating where better alternatives already exist[1][2]).

[1] https://spectrum.ieee.org/home-heating-with-hydrogen-is-ill-... [2] https://www.nature.com/articles/s41928-021-00706-6

jksk61··on Vivaldi on Android bypasses restrictions to let users access Bing Chat
using hydrogen for transportation is the same as saying you ate healthy because you ordered a salad at Mcdonald.
jksk61··on US urged to reveal UFO evidence after claim that it has intact alien vehicles
what mysterious propulsion patents? I'm out of the loop.
jksk61··on Bayes is guaranteed to overfit
Yes and no. Suppose for example you give MNIST to a SVM and fit the model, then test it only on 0 and 1 digits, which are generally well discriminized, you'll get almost 100% accuracy in test, whereas 97% or less in training. (ye probably need some preprocessing, like using PacMAP or UMAP or whatever but the point is the same)

However, that's just because I decided the right data to test it onto. So, you can't really say much on a model using that definition.

jksk61··on Public Money, Public Code
italy does: both dataset and public websites, but it's italy and the list is more incomplete than not.

see https://github.com/italia

jksk61··on You Have a New Memory
I don't know, but this seems kind of a repeated argument over new technology we are not able to understand. Even cars/motorbikes were (although mainly jokingly) assumed a man's partner.

That's not to say that it is unsettling, but I just want to point out that many of the "weird" behaviours are mainly dependent on the subject (person) and not the object (smartphone/internet/socials).

jksk61··on Choose Your Weapon: Survival Strategies for Depressed AI Academics
i'm not saying LLM are modeling linguistics in any way lol. i only meant that there's some kind of phenomena related to scaling+attention that produces good enough result for most "human language stuff", which is kind of unexpected (i mean everyine knows that if you build a large enough model you can teach it any function, but cmon it is architecture+scaling that made it possibile not scaling alone). Moreover, the architectures used, from attention layer, even LSTM for that matter are not completely understood, are being used because "they works" just as in the old days of electromagnetism the empiral laws "just worked" for their usage.

btw, in other languages i guess it is decent although it depends on which language, at least gpt4.

jksk61··on Choose Your Weapon: Survival Strategies for Depressed AI Academics
lmao even in my university (the serbian uni), we have at least calculus+linear algebra before any nn course. also to "learn" what's a cnn you just need gradients not integrals (unless you use some kind of non-lipschitz function as activation?), plus the idea of what a convolution is... but even Mobius knew it back in the 800s.

anyway i think your statement that industry is light years away from unis is just misleading. i think the two are trying to answer different questions: 1. how can i achieve a "somewhat" decent chatbot that gets me rich albeit not even knowing what it does [industry in case you wondered] 2. try to understand, quantify and measure how well a model works, is it stable? does it converge if we have small datasets? and so on so forth.

just my two cents, to conclude i think a good analogy to the current climate is the 700-800s with electromagnetism: plenty of people discovered "empirical" laws but didn't understand really the phenomenon.

jksk61··on Sh1mmer – An exploit capable of unenrolling enterprise-managed Chromebooks
There are different reason to favour handwriting: 1. Development coordiantion eye-hand; 2. Computers can go to maintenance as you said; 3. Creativity, how many times did you doodle on a white word file? 4. They'll have plenty of times in uni/job to stare blankly at a LCD screen 5. Paper is less tiresome on eyes.
jksk61··on Europe's largest deposit of rare earth metals discovered in Sweden
As one could argue that the earth is flat and nothing is real.
← PreviousPage 2 of 2