GPRat: Gaussian Process Regression using Asynchronous Tasks

GPRat is an open-source library for Gaussian Process Regression. Leveraging the asynchronous many-task runtime HPX, we aim to combine the performance of asynchronous parallelism in C++ with the ease of use of commonly available Python libraries. Thus, GPRat can be conveniently integrated into Python projects without binding overheads or used directly with pure C++ code.

Dependencies

GPRat depends on HPX for asynchronous task-based parallelization. Furthermore, for CPU-only BLAS computation GPRat requires OpenBLAS or MKL. A CUDA installation is required for GPU-only BLAS computations.

Install dependencies

All dependencies can be installed using Spack. A script to install and setup spack for GPRat is provided in spack-repo. Spack environment configurations and setup scripts for CPU and GPU use are provided in spack-repo/environments.

How To Compile

GPRat makes use of CMake presets to simplify the process of configuring the project.

For example, building and testing this project on a Linux machine is as easy as running the following commands:

cmake --preset=dev-linux
cmake --build --preset=dev-linux
ctest --preset=dev-linux

As a developer, you may create a CMakeUserPresets.json file at the root of the project that contains additional presets local to your machine. In addition to the build configuration dev-linux, there are release-linux, dev-linux-gpu, and release-linux-gpu. The configurations suffixed with -gpu build the library with CUDA.

GPRat can be build with or without Python bindings. The following options can be set to include / exclude parts of the project:

Option name	Description	Default value
GPRAT_BUILD_CORE	Enable/Disable building of the core library	ON
GPRAT_BUILD_BINDINGS	Enable/Disable building of the Python bindings	ON
GPRAT_ENABLE_FORMAT_TARGETS	Enable/Disable code formatting helper targets	ON if top-level
GPRAT_ENABLE_EXAMPLES	Enable/Disable example projects	ON if top-level
GPRAT_USE_MKL	Enable/Disable usage of MKL library	OFF
GPRAT_WITH_CUDA	Enable/disable compilation with CUDA support	OFF

Respective scripts can be found in this directory.

We also provide a spack package for GPRat in spack-repo/packages for portable and convenient compilation. When the repository is added to spack, GPRat can be installed with spack install gprat~cuda~bindings~examples blas={mkl,openblas}

How To Run

GPRat contains several examples. One to run the C++ code, one to run the Python code as well as two reference implementations based on TensorFlow (GPflow) and PyTorch (GPyTorch).

To run the GPRat C++ code

• Go to examples/gprat_cpp
• Set parameters in execute.cpp
• The example is built as part of the main project.
  • Go to build/ and execute ./gprat_cpp [--use_gpu] to run the example.
  • If you want to use an installed GPRat version: Run ./run_gprat_cpp.sh [cpu/gpu] [x86/arm/riscv] to build and run the example.

To run GPRat with Python

• Go to examples/gprat_python
• Set parameters in config.json
• Run ./run_gprat_python.sh [cpu/gpu] to run the example

To run GPflow reference

• Go to examples/gpflow_reference
• Set parameters in config.json
• Run ./run_gpflow.sh [cpu/gpu/arm] to run example

To run GPflow reference

• Go to examples/gpytorch_reference
• Set parameters in config.json
• Run ./run_gpytorch.sh [cpu/gpu/arm] to run example

The Team

The GPRat library is developed by the Scientific Computing department at IPVS at the University of Stuttgart. The project is a joined effort of multiple undergraduate, graduate, and PhD students under the supervision of Prof. Dr. Dirk Pflüger. We specifically thank the follow contributors:

• Alexander Strack: Maintainer and initial framework.

• Maksim Helmann: Optimization, Python bindings and reference implementations.

• Henrik Möllmann: Accelerator Support.

How To Cite

@InProceedings{GPRat2025,
  author={Helmann, Maksim and Strack, Alexander and Pfl{\"u}ger, Dirk},
  title={{GPRat: Gaussian Process Regression with Asynchronous Tasks}},
  booktitle={Asynchronous Many-Task Systems and Applications},
  year={2025},
  publisher={Springer Nature}
}