FRIDOM: Framework for Idealized Ocean Models#
About FRIDOM#
Purpose & Flexibility: FRIDOM is a powerful and modular framework originally developed for running simulations of idealized ocean models. Thanks to its modular design, it can be used to simulate any model represented by a set of partial differential equations, such as \(\partial_t \boldsymbol{z} = \boldsymbol{f}(\boldsymbol{z}, t)\).
Minimizing Boilerplate Code: To streamline the development process, FRIDOM provides base classes for common components like grids, differential and interpolation operators, time-stepping schemes, netCDF output, animations, etc.
Easy Model Modifications: Every component of a model in FRIDOM is fully exchangeable without changing the model’s source code. This feature makes FRIDOM an excellent sandbox for testing new ideas and a useful tool for educational purposes.
Balancing Flexibility & Usability: While modular frameworks often compromise user-friendliness for flexibility, FRIDOM strives to be both flexible and easy to use. It offers a high-level API, comprehensive tutorials, and numerous examples.
Performance through Python & JAX: Written in Python for ease of use, FRIDOM overcomes Python’s performance limitations by leveraging the Just-In-Time (JIT) compiler from JAX. This approach allows FRIDOM to achieve speeds comparable to compiled languages like Fortran or C, and it can further accelerate simulations by running on GPUs.
Note
FRIDOM is in an early development stage, and as such, it may undergo significant changes.
Alternatives#
FRIDOM draws inspiration from several existing modeling frameworks and tools, which have influenced its design and capabilities. Some notable inspirations include:
A very powerful ocean model written in Julia with CPU and GPU support. Oceananigans is suitable for both idealized and realistic ocean setups.
An ocean model written in Fortran with many available parameterizations and closures.
A Python implementation of pyOM2 that runs on CPUs and GPUs using JAX.
A pseudo-spectral non-hydrostatic incompressible flow solver written in Fortran.
A Python framework for solving systems of partial differential equations using the spectral Galerkin method.