MaRDI Open Interfaces (Q6534208): Difference between revisions

## MaRDI Open Interfaces: Software project for improving Interoperability in Scientific Computing\N \N''MaRDI Open Interfaces'' is a project that aims to improve interoperability in scientific computing by removing two hurdles that computational scientists commonly face in their daily work.\N \NThese hurdles are as follows. First, numerical solvers are often implemented in different programming languages. Second, these solvers frequently expose significantly different interfaces---they have different function names, arguments order, and call sequences. As a result, switching from one solver to another can take non-trivial effort for code modifications and correctness testing.\N \NThe ''MaRDI Open Interfaces'' project aims to alleviate these problems by providing automatic data marshalling between languages and a set of interfaces for typical numerical problems, such as integration of differential equations and optimization.\N \N## How can it be used?\N \N* The package is a software library and does not impose constraints on how it can be used in computational scripts\N \N* Supported languages are C, Julia, and Python (with more languages to follow)\N \N* A common use case is benchmarking different implementations of the same numerical algorithm or, more generally, different solvers for the same numerical problem\N \N## Installation and documentation\N \NPlease refer to the documentation at [https://mardi4nfdi.github.io/open-interfaces/](https://mardi4nfdi.github.io/open-interfaces/) for deeper view on the goals and implementation details as well as installation instructions, tutorials, and API reference.\N \N'''The ''MaRDI Open Interfaces'' package is currently in beta and is still evolving; however, it can already be used in computational experiments requiring time integration.'''\N \N## Publications\N \NDmitry I. Kabanov, Stephan Rave, and Mario Ohlberger. ''Improving Interoperability in Scientific Computing via MaRDI Open Interfaces''. [arXiv:2504.03628](https://arxiv.org/abs/2504.03628) (Accepted for publication in the ''Journal of Open Research Software''), 2025.