Notes about random things

Setting up NextCloud on Raspberry Pi 4 using k3s

Mon, 20 Nov 2023 14:30:00 +0200

Introduction

I was setting up a NextCloud instance on my Raspberry Pi 4, using k3s, and found out that there are quite some step-by-step guides on how to do that, none of them fully addressed all the issues I had, so I decided to write yet another guide on how to that. Mostly for myself, but maybe it will be useful for someone else. In particular, I faced the following issues:

OneDrive on Linux

Thu, 11 Feb 2021 19:37:53 +0200

OneDrive sync on Linux

There is no official client for OneDrive for Linux, but there are some open source alternatives, e.g.:

OneDrive Client for Linux

TL;DR does what you expect, available only for OneDrive.

Syncs a local folder with remote OneDrive: monitors changes both locally and remotely and synchronizes them.

In short, does pretty much what you expect from a file hosting and syncing service, similar to native clients.

Listing files and folders sorted by size

Sun, 15 Nov 2020 17:10:00 +0200

Sort files and directories by size on disk

The following command sorts files and directories in descending order by their disk usage:

du -hs * | sort -rh

Explanation

`du`

du summarizes disk usage of files, and for directories it summarizes them recursively. The -s option tells du to display “only a total for each argument”. Without it, du also displays recursively the sizes of each nested directory. So for the following file structure

pathos.multiprocessing

Sun, 09 Aug 2020 20:19:18 +0200

Multiprocessing in Python

Although Python is not very well suited for parallel programming, sometimes it could be useful.

If it’s a computation, then we are probably better off using something like Dask, Numba, etc. But if it’s not computations, then there is a built-in solution in Python: multiprocessing.

We’ll stick with computations for examples though, since their are simpler.

Comparison of parallel and not parallel

A quick illustration of why parallelization is great when the problem is embarrassingly parallel.

Acceptance-rejection method for generating random variables

Fri, 10 Feb 2017 20:40:00 +0200

Acceptance-rejection method is a method for generating samples from a distribution, for which the probability density function is known, but inverse cumulative probability function is not known, and thus, using the inverse CDF method is not possible.

Although there is quite a lot of information on the topic available, I will try to explain the method the way that I (a.k.a 5-year-old) understand.

Idea

Majorizing distribution

Let us say that we want to draw numbers from some distribution¹ $ f(x) $—target distribution—but we have only distribution $ g(x) $, such that by multiplying it with some constant $ c $ it is always larger than $ f(x) $,

LaTeX fonts in R Markdown plots

Fri, 03 Feb 2017 19:30:00 +0100

Let us be honest, one of the reasons we use R Markdown to compile documents into PDF is the aesthetic pleasure provided by LaTeX. However, all the efforts can be ruined by wrong fonts in plots that are not the same as in the rest of the document. For example

Well, these are some ugly fonts (not by itself, but in combination with the rest of the document). Would not it be much better to have something like this?

Table and figure captions in R Markdown

Thu, 10 Nov 2016 15:30:00 +0200

R Markdown is an extremely useful tool for producing reports using R. The problem is that decent quality reports require captions for figures and tables, and it is not straightforward to do. The good news is that it is still quite easy.

Pandoc’s Markdown: numbered captions

The key to adding captions is that knitr actually converts your .Rmd file to .md file first, and then uses pandoc to conert it to html, pdf or another format. Therefore, everything that works in Pandoc also works in R Markdown. It is worth noting, though, that Pandoc uses its own extended version of Markdown called Pandoc’s Markdown. Among other things it allows captioning your figures and tables. This is done in the following way for the figures