Conversation
Notices
-
@crunklord420 I am in the process of implementing some stuff from statistics in c. Have you ever dealt with the topic? For example, i want the fastest as most efficient way to get random samples which fit a specific normal/gaussian distribution. I tried Latin Hypercube with Box Muller algorithm, but the results are not better than when I use standard Monte Carlo with Box l-Muller.
-
@Saxophone3784 sorry I don't really know math
-
@crunklord420 No probs. just solved the issue. Chat GPT actually managed to hallucinate a function with a myriad of variables that somehow solved the problem. Literally thought it made up these numbers but they were all correct. I promised to tip, maybe because of that it worked, lmao.
-
@crunklord420 Sometimes I sort of feel bad. All those topics are so interesting, but I do not have any time to get into them. Like I would love a deep understanding of statistical algorithms and everything, but it feels to me I am too far gone to learn all that.
-
@Saxophone3784 people usually use numpy but yeah python is the slowest language that people actually use. I believe in loops in particular are very slow and part of "performant python" is avoiding them, hilariously.
-
@crunklord420 I also tried to do the same stuff in Python with all the fancy libraries. It is still insanely slow compared to what my C code produces. Sometimes I feel like the modern programming has the "resource curse". Since the hardware development is currently on steroids, no improvement needs to be done in the programming area and people just shit out insanely slow pointless programs.
-
@crunklord420 Yup, I tried with numpy/scipy. It takes a different approach. I've worked in Python quite a bit, the basic rule is "if you are using loops, you are doing something wrong". I think at this point Python is just a wrapper for various libraries. No function you use is itself written in Python. You just call them. You stack black boxes on top of each other hoping you get the answer you are looking for. You don't write anything in Python, you just script. It's nice for data analysis, but the moment you actually need something to work fast it all falls apart. I think programming needs a very major shake up, most people don't really understand what they are doing and as a result we have basic applications that can't even display short text without lagging.
-
@Saxophone3784 yeah python is terrible and the worst part of about it is it's guaranteed not to be functional in 2 months as the fragile ecosystem of libraries and interpreter versions is constantly breaking.
-
@Zergling_man @Saxophone3784 I just don't write python. It's not simple or easy to use. The libraries aren't very good. It's just not a very good programming language unless you're comparing it to writing bash scripts.
Python isn't really the right tool for any job.
-
@crunklord420 @Saxophone3784 >python is terrible
Skill issue. Python is fine, just don't use it when you need speed. Use the right tool for the job.
>and the worst part of about it is it's guaranteed not to be functional in 2 months as the fragile ecosystem of libraries and interpreter versions is constantly breaking
Skill issue, stop installing random shit from pypi lol, 90% of it already exists in standard libs, and the rest is like 5 lines to implement yourself.
-
@Saxophone3784 Python package management is so beyond saving that people just distribute their software via Docker files.
No on else does this. It's just Python people.
-
@crunklord420 And it has several package managers which makes 0 sense for me. Why do you need conda? Why do you need different ways of making virtual environment? I always feel like most of it was done by people who just wanted to add it to their CVs. They did not even stop for a moment to think if it actually makes sense.