SLOPE

SLOPE is a C++ library for Sorted L-One Penalized Estimation (SLOPE). Its main purpose is to serve as a backend for higher-level language packages in R and Python, but it can also be used as a standalone library in the off-chance that you want to fit your models entirely via C++.

SLOPE solves the following optimization problem:

$$\min_{\beta} \frac{1}{n} f(X\beta + \beta_0) + \alpha \sum_{j=1}^p \lambda_j |\beta_{(j)}|,$$

where $\beta_{(j)}$ is the $j$th element of non-increasingly sorted vector of $\beta$, and $f$ is a convex loss function. SLOPE is a generalization of the lasso.

See the following sections for more information:

• Getting Started
• Dependencies