Home Page

Case Study

Marginal Likelihood for Linear Regression (C/C++: LAPACK & GSL)

2025
  • C/C++
  • LAPACK
  • GSL
  • Statistical Computing
  • Performance Optimization

Two high-perf C/C++ versions (LAPACKE, GSL) for LM marginal likelihood; GEMM/solve/log-det with careful memory/layout; matched R baseline and spec.

Problem & Motivation:

Compute evidences fast and accurately in native code.

Data & Approach:

  • Linear algebra kernels via BLAS/LAPACK & GSL; attention to row/col major; tests vs R.

Results:

  • Numeric parity with R; strong runtime characteristics on medium-n.

Limitations:

CPU-only; no batching across many models.