sparse-approx-gsm 1.0.0

Sparse Approximation by the Generalized Soft-Min Penalty

sparse-approx-gsm

Python (+ optional CUDA) implementation of Sparse Approximation by the Generalized Soft-Min (GSM) Penalty — a sparse-recovery solver based on the Trimmed Lasso.

Tal Amir, Ronen Basri, Boaz Nadler — The Trimmed Lasso: Sparse Recovery Guarantees and Practical Optimization by the Generalized Soft-Min Penalty. Weizmann Institute of Science.

Install

With CUDA extension:

NVCC=/path/to/nvcc
PYBIND11_DIR=$(python -c "import pybind11; print(pybind11.get_cmake_dir())")
CMAKE_ARGS="-DCMAKE_CUDA_COMPILER=$NVCC -Dpybind11_DIR=$PYBIND11_DIR -DCMAKE_CXX_COMPILER=/usr/bin/g++" \
  pip install --no-build-isolation . -v

CPU only (omit CUDA flags):

pip install --no-build-isolation .

--no-build-isolation is needed on conda environments to avoid linker conflicts. See CLAUDE.md for full details.

Usage

from gsm_python import sparse_approx_gsm, SparseApproxGSM, gsm

# High-level solver: min ||Ax - y||_2  s.t.  ||x||_0 <= k
x = sparse_approx_gsm(A, y, k, profile='normal')

# GSM function and gradient
mu, theta = gsm(z, k, gamma)

Optional GPU

Install CuPy for the matching CUDA version (e.g. pip install cupy-cuda12x), then pass a CuPy array to gsm() or use SparseApproxGSM(..., use_gpu=True).