Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

【Hackathon 5th No.53】NSFnets (Navier-Stokes Flow nets): Physics-informed neural networks for the incompressible Navier-Stokes equations #695

Merged
merged 31 commits into from
Jan 8, 2024
Merged
Changes from 1 commit
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
Prev Previous commit
Next Next commit
a
DUCH714 committed Dec 20, 2023
commit 52bbeb6638875814d4d9f07e0578c5e33adc79f9
12 changes: 8 additions & 4 deletions docs/zh/examples/nsfnet.md
Original file line number Diff line number Diff line change
@@ -9,7 +9,11 @@
python VP_NSFNet1.py

# VP_NSFNet2
python VP_NSFNet2.py
# linux
wget -nc https://paddle-org.bj.bcebos.com/paddlescience/datasets/NSFNet/cylinder_nektar_wake.mat -P ./datasets/
# windows
# curl https://paddle-org.bj.bcebos.com/paddlescience/datasets/NSFNet/cylinder_nektar_wake.mat --output ./datasets/burgers_sine.mat
python VP_NSFNet1.py DATASET_PATH=https://paddle-org.bj.bcebos.com/paddlescience/datasets/NSFNet/cylinder_nektar_wake.mat

# VP_NSFNet3
python VP_NSFNet3.py
@@ -18,13 +22,13 @@

``` sh
# VP_NSFNet1
python VP_NSFNet1.py mode=eval
python VP_NSFNet1.py mode=eval EVAL.pretrained_model_path=https://paddle-org.bj.bcebos.com/paddlescience/models/nsfnet/nsfnet1.pdparams

# VP_NSFNet2
python VP_NSFNet2.py mode=eval
python VP_NSFNet2.py mode=eval EVAL.pretrained_model_path=https://paddle-org.bj.bcebos.com/paddlescience/models/nsfnet/nsfnet2.pdparams

# VP_NSFNet3
python VP_NSFNet3.py mode=eval
python VP_NSFNet3.py mode=eval EVAL.pretrained_model_path=https://paddle-org.bj.bcebos.com/paddlescience/models/nsfnet/nsfnet3.pdparams
```
## 1. 背景简介
最近几年,深度学习在很多领域取得了非凡的成就,尤其是计算机视觉和自然语言处理方面,而受启发于深度学习的快速发展,基于深度学习强大的函数逼近能力,神经网络在科学计算领域也取得了成功,现阶段的研究主要分为两大类,一类是将物理信息以及物理限制加入损失函数来对神经网络进行训练, 其代表有 PINN 以及 Deep Retz Net,另一类是通过数据驱动的深度神经网络算子,其代表有 FNO 以及 DeepONet。这些方法都在科学实践中获得了广泛应用,比如天气预测,量子化学,生物工程,以及计算流体等领域。而为充分探索PINN对流体方程的求解能力,本次复现[论文](https://arxiv.org/abs/2003.06496)作者先后使用具有解析解或数值解的二维、三维纳韦斯托克方程以及使用DNS方法进行高进度求解的数据集进行训练。论文实验表明PINN对不可压纳韦斯托克方程具有优秀的数值求解能力,本项目主要目标是使用PaddleScience复现论文所实现的高精度求解纳韦斯托克方程的代码。
2 changes: 1 addition & 1 deletion examples/nsfnet/conf/VP_NSFNet2.yaml
Original file line number Diff line number Diff line change
@@ -13,7 +13,7 @@ MODEL:
hidden_size: 100
activation: "tanh"

mode: eval
mode: train

ntrain: 140000

2 changes: 1 addition & 1 deletion examples/nsfnet/conf/VP_NSFNet3.yaml
Original file line number Diff line number Diff line change
@@ -11,7 +11,7 @@ MODEL:
hidden_size: 100
activation: "tanh"

mode: eval
mode: train

ntrain: 70000