Paddlesci

PaddleScience 1.1.0 Release Notes

PaddleScience 1.1.0 is now available with multiple feature updates, documentation, code optimizations and bug fixes. This update brings numerous enhancements that improve overall performance and usability. Some of the key updates include:

Examples
[Bracket](https://github.com/PaddlePaddle/PaddleScience/blob/release%2F1.1/examples/bracket/bracket.py), [Aneurysm](https://github.com/PaddlePaddle/PaddleScience/blob/release/1.1/examples/aneurysm/aneurysm.py), [DeepHPMs](https://paddlescience-docs.readthedocs.io/zh/release-1.1/zh/examples/deephpms/), [LabelFree-DNN-Surrogate](https://github.com/PaddlePaddle/PaddleScience/blob/release/1.1/examples/aneurysm/aneurysm_flow.py), [hPINNs](https://paddlescience-docs.readthedocs.io/zh/release-1.1/zh/examples/hpinns/#2), [tempoGAN](https://paddlescience-docs.readthedocs.io/zh/release-1.1/zh/examples/tempoGAN/), and their corresponding network structures are added.

Module
Support for SDF (Signed Distance Function) to all basic geometries; adding [Stan](https://arxiv.org/abs/2204.12589v1) activation functions; and adding [FunctionalLoss](https://paddlescience-docs.readthedocs.io/zh/release-1.1/zh/api/loss/#ppsci.loss.FunctionalLoss) and [FunctionalMetric](https://paddlescience-docs.readthedocs.io/zh/release-1.1/zh/api/metric/#ppsci.metric.FunctionalMetric) to support customized complex loss and metric.

Improvements
Preliminarily support for the [WandB](https://github.com/wandb/wandb) experimental recording module([how-to-use](https://paddlescience-docs.readthedocs.io/zh/latest/zh/user_guide/#14-wandb)); adding unit tests for the Equation module; and adding CUDA 11.6-based docker file.(Thanks to all contributors from community)

Document and bug fix
Adding [usage guides](https://paddlescience-docs.readthedocs.io/zh/release-1.1/zh/user_guide/), [development and reproduction guides](https://paddlescience-docs.readthedocs.io/zh/release-1.1/zh/development/) ; fixing multiple docstring formula description errors in the Equation module.

**We recommend upgrading paddlescience to version 1.1.0 or the develop branch.**

PaddleScience 1.0.0 Release Notes

The PaddleScience 1.0.0 version brings a completely new design of APIs and modules to support different training paradigms in one toolkit: mechanism-driven, data-driven and mechanism-data mixed case. The key features are as follows:

- Newly designed modules
- Built-in high-frequency usage API, such as `PDE`, `BoundaryConstraint`, `InteriorConstraint`.
- Decoupled training/evaluation/visualization.
- Automatic result visualization.
- Support resume training, transfer learning, **automatic mixed precision**, **gradient accumulation** and **distributed parallell training**.
- Enhanced features
- Enhanced geometry module, including interior/boundary sampling and boolean operations, as well as STL-based SDF(signed distance field) weighing.
- Support dirichlet, nuemann, robin and **custom** boundary condition.
- Multiple model architecture such as MLP, Transformer.
- Support high-order differentiation related API, such as jacobian/hessian and L-BFGS optimizer.
- Varied examples
- 10+ classical examples covering fluid, structure, meteorology and more areas.
- Detailed user document and end-to-end tutorial.
- Quickly example experience in AIStudio.

- User friendly
- Structured logger.
- Comprehensive docstrings and type hints.