SurKit

Surkit is a toolkit for surrogate model development.

Introduction

A surrogate model is a simplified model that can replace a complex or high-fidelity model in engineering problems. Surrogate models can reduce the computational cost and time of evaluating the original model, and facilitate the optimization and approximation processes. SurKit allows users who are unfamiliar with neural networks to get started quickly by selecting the type of surrogate model and providing parameters to automatically generate and train surrogate models.

Installation

System requirements

SurKit requires some prerequisites including NumPy, Matplotlib, and at least one of the following frameworks:

• PyTorch: PyTorch>=1.9.0

• Oneflow: Oneflow>=0.9.0

• JAX: JAX>=0.3.25, Flax>=0.6.4, Optax>=0.1.4

Install with pip package

To Install the stable version of Surkit with pip:

pip3 install surkit

Documentation

Please refer to https://surkit.readthedocs.io/en/latest/index.html.