Blog

Posts about research and programming

Posts by Johan Larsson on statistics, optimization, and scientific computing in R, Python, Julia, and Rust, plus notes on the software he develops.

Three intercept-update strategies take different paths toward the same logistic regression solution.

How to Train Your Intercept

A comparison of intercept update strategies in coordinate descent

How intercept updates affect convergence in regularized generalized linear models, and why one safeguarded Newton step can make a large difference.

26 October 2026
Johan Larsson
Panache: An Editor Companion for Markdown

Panache: An Editor Companion for Markdown

Panache is a language server, formatter, and linter for Quarto and other Markdown flavors. It makes it easy to write technical documents with citations, cross-references, and executable code chunks.

9 October 2026
Johan Larsson
Eunoia: Euler and Venn Diagrams in Five Languages

Eunoia: Euler and Venn Diagrams in Five Languages

Eunoia is a Rust engine for area-proportional Euler and Venn diagrams, now with bindings for R, Python, Julia, and JavaScript.

4 July 2026
Johan Larsson
Basin: Numerical Optimization in Rust

Basin: Numerical Optimization in Rust

Basin is a Rust library for numerical optimization, with pluggable linear-algebra backends, first-class constraints, and WebAssembly support.

10 June 2026
Johan Larsson
Introducing Tomat: A Pomodoro Timer for Status Bars

Introducing Tomat: A Pomodoro Timer for Status Bars

I’ve decided that the world needs yet another pomodoro timer. yapt was unfortunately already taken, so I named it after the Swedish tomato: Tomat!

12 January 2026
Johan Larsson
What’s New in Moloch 2.0.0

What’s New in Moloch 2.0.0

Moloch 2.0.0 is out, with a redesigned color system, new themes, improved documentation, redesigned itemize lists, and various other refinements.

5 December 2025
Johan Larsson
Qualpal

Qualpal

My pet project Qualpal is now a stand-alone C++ library as well as web app and was just published in the Journal of Open Source Software!

22 October 2025
Johan Larsson
LUSEM’s Best Thesis Award 2024

LUSEM’s Best Thesis Award 2024

Lund University School of Economics and Management awarded my PhD thesis on optimization and algorithms for sparse regression its best thesis prize for 2024.

6 May 2025
Johan Larsson
Cramér Prize for Best PhD Thesis

Cramér Prize for Best PhD Thesis

I received the 2025 Cramér Prize for the best PhD thesis in statistics defended in Sweden, for my work on optimization and algorithms for sparse regression! 🥳

5 February 2025
Johan Larsson
New Paper on Feature Normalization and Regularized Regression

New Paper on Feature Normalization and Regularized Regression

We have a new arXiv paper on normalization for regularized regression (lasso, ridge, and elastic net), focused on how to standardize binary features.

21 January 2025
Johan Larsson
Moloch: A Revival of the Metropolis LaTeX Beamer Theme

Moloch: A Revival of the Metropolis LaTeX Beamer Theme

Moloch is a fork of Metropolis, the popular but no longer maintained LaTeX beamer theme. It fixes the theme’s outstanding issues and adds new features.

30 May 2024
Johan Larsson
PhD Thesis

PhD Thesis

My PhD thesis on optimization and algorithms for sparse regression has been published. Here I give a brief summary of the papers included in it.

22 May 2024
Johan Larsson
New Julia Package: ProjectRoot

New Julia Package: ProjectRoot

Introducing ProjectRoot.jl, a small Julia package for robust file referencing relative to your project root, inspired by R’s here and Python’s pyprojroot.

13 May 2024
Johan Larsson
SLOPE 0.2.0

SLOPE 0.2.0

An update to the SLOPE R package for sparse regression with the sorted L1 norm, a generalization of the lasso that handles correlated predictors gracefully.

14 April 2020
Johan Larsson
Polygon Labeling with polylabelr

Polygon Labeling with polylabelr

polylabelr is a new R package that finds the visual center, the pole of inaccessibility, of a polygon, handy for labeling Euler diagrams and maps.

29 October 2018
Johan Larsson
Finding the Farthest Points in a Point Cloud

Finding the Farthest Points in a Point Cloud

An algorithm for finding the most mutually separated points in a cloud, used by the qualpalr R package to pick maximally distinct qualitative colors.

30 October 2016
Johan Larsson
Introducing eulerr

Introducing eulerr

This is an introduction to my new package, eulerr, an R package that computes and plots eulerr diagrams of set relationships.

19 October 2016
Johan Larsson
Introducing qualpalr

Introducing qualpalr

Let me introduce qualpalr: an R package that generates qualitative color palettes with distinct colors using color difference algorithms.

15 October 2016
Johan Larsson
No matching items