libgcca

libgcca is an open-source C library that provides implementations of generalized canonical correlation analysis (GCCA) and related multi-view learning techniques. It is designed to help researchers and developers integrate data from multiple sources by learning a shared low-dimensional representation that reflects the common structure across views. The library focuses on a practical, numerically stable interface suitable for inclusion in C and C++ projects.

Its core functionality covers common GCCA formulations, including SUMCOR and MAXVAR variants, as well as options for regularization and normalization to improve performance on real-world data. libgcca operates on a set of data matrices (views) that share the same number of observations but may have different feature counts. It computes projection matrices for each view, enabling projection of new samples and generation of a consensus representation.

Usage involves supplying the view matrices to the library, configuring model options (such as dimensionality of the latent space and regularization strength), and invoking the fit routine. After fitting, transformed representations can be obtained for downstream tasks such as clustering or visualization. The library exposes a C API intended to be accessible from C and C++ code, and can be extended through bindings to other languages.

libgcca is distributed under an open-source license and is typically built with standard C development tools. It relies on common linear algebra routines for eigen and singular value decompositions and is designed to be portable across platforms. The project aims to provide a compact, dependency-light solution for multi-view data analysis in statistical learning and data integration workflows.