Generate a Dataset

Generating synthetic data with the Inductiva API is a structured process that begins with a base simulation model and scales to thousands of variations, producing a diverse and robust dataset.

The typical workflow includes the following steps:

1. Set Up the Base Case Start by preparing the configuration files for a base case simulation model of the system under study. This step often requires domain expertise and a solid understanding of the simulation software being used.
2. Generalize the Base Case Generalize the configuration files to allow variations of the base case. Our Templating System enables dynamic substitution of variables at runtime, making it easy to modify simulation (hyper)parameters through Python. This supports both exhaustive and randomized exploration of the configuration space.
3. Generate Synthetic Data at Scale Use the API to launch thousands of simulation variants in parallel on cloud infrastructure. Output data from each simulation is automatically collected and made available for post-processing and downstream use.

Whether you're a Machine Learning engineer or a simulation expert, the Inductiva workflow provides a scalable and efficient solution for generating synthetic datasets to support physics-informed ML models.

Ready to dive in?

Explore these hands-on tutorials to jumpstart your journey:

• Generate an OpenFOAM Dataset
• Create Synthetic Data for Physics-Informed ML

Or, if you prefer to read, check out our blog post:

• SimuStruct Dataset (0.1)