Why Julia?

Some of you may ask, why Julia? Was Matlab-Unfold not good enough?

There are two main answers, one specific to Unfold, one more general.

Initially and specifically to Unfold.jl was the impossibility to adapt many LMMs to Matlab (we tried). Julia's MixedModels.jl results in up to 100x speedup (compared to R's LME4).

More generally, we came up with a list of things that make Julia exceptionally well suited for science:

• 🚀 Easy to write fast code. Solves the 2-language problem: you have simple python-like code, but at the speed of mex/numba/cython. This allows "normal" scientists to produce efficient code without needing the C++ skills.

• 🤝 Easy interoperability with Python, R, and Matlab. If a library does not exist in Julia, simply call it from another language.

• 📦 Built-in reproducible package management. Think of Python's conda-env, but actually working & fast.

• 🈳 Built-in support for unicode via latex \beta=𝛽.

• 🔤 Syntax is easy, a mix of R/MatLab/Python.

Compared to MatLab

All of the above, but in addition:

• 💸 It is free.

• ✅ Automated unit testing and documentation are much easier.

Compared to Python

All of the above, but in addition:

• 🧮 LinearAlgebra integrated. No more np.diff(np.mult(A,B)) etc.