-
Notifications
You must be signed in to change notification settings - Fork 283
Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
- Loading branch information
Showing
1 changed file
with
70 additions
and
0 deletions.
There are no files selected for viewing
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Original file line number | Diff line number | Diff line change |
---|---|---|
@@ -0,0 +1,70 @@ | ||
# GeoFNO 设计文档 | ||
|
||
> RFC 文档相关记录信息 | ||
| | | | ||
| ------------ | ------------------ | | ||
| 提交作者 | ADream-ki | | ||
| 提交时间 | 2025-03-05 | | ||
| RFC 版本号 | v1.0 | | ||
| 依赖飞桨版本 | release 2.6.2 版本 | | ||
| 文件名 | 20250305_CoNFiLD.md | | ||
|
||
## 1. 概述 | ||
|
||
### 1.1 相关背景 | ||
|
||
[NO.14 CoNFiLD 论文复现](https://github.com/PaddlePaddle/community/blob/master/hackathon/hackathon_8th/%E3%80%90Hackathon_8th%E3%80%91%E4%B8%AA%E4%BA%BA%E6%8C%91%E6%88%98%E8%B5%9B%E2%80%94%E5%A5%97%E4%BB%B6%E5%BC%80%E5%8F%91%E4%BB%BB%E5%8A%A1%E5%90%88%E9%9B%86.md#no14-confild-%E8%AE%BA%E6%96%87%E5%A4%8D%E7%8E%B0) | ||
|
||
传统的大涡模拟(LES)和直接数值模拟(DNS)虽然能够提供详细的流场预测,但其计算需求极高,限制了在实际工程中的应用。基于深度学习的代理模型虽然计算效率较高,但它们通常依赖确定性框架,无法准确捕捉湍流的混沌和随机特性。CoNFiLD模型通过结合条件神经场编码和潜在扩散过程,旨在克服这些挑战,实现高效且鲁棒的时空湍流生成。 | ||
|
||
### 1.2 功能目标 | ||
|
||
1. 复现 CoNFiLD 代码,实现完整的推理流程。 | ||
2. 保持精度与论文精度一致,相对误差在 ±10% 以内。 | ||
3. 产出论文相关文档、图片、视频等。 | ||
|
||
### 1.3 意义 | ||
|
||
复现 CoNFiLD 代码,能够使用 CoNFiLD 模型进行推理。 | ||
|
||
## 2. PaddleScience 现状 | ||
|
||
PaddleScience 套件暂无 CoNFiLD 代码案例。 | ||
|
||
## 3. 目标调研 | ||
|
||
- 论文解决的问题: | ||
CoNFiLD 模型解决在复杂、不规则三维域内快速生成复杂时空湍流的问题 | ||
- 链接: | ||
代码:[https://github.com/jx-wang-s-group/CoNFiLD](https://github.com/jx-wang-s-group/CoNFiLD) | ||
论文:[https://doi.org/10.1038/s41467-024-54712-1](https://doi.org/10.1038/s41467-024-54712-1) | ||
|
||
需将 Pytorch 代码转换为 Paddle | ||
|
||
## 4. 设计思路与实现方案 | ||
|
||
参考 PaddleScience 已有代码实现 CoNFiLD | ||
|
||
1. 数据预处理 | ||
2. 模型构建 | ||
3. 超参数设定 | ||
4. 模型推理的评估指标 | ||
|
||
### 4.1 补充说明[可选] | ||
|
||
无 | ||
|
||
## 5. 测试和验收的考量 | ||
|
||
复现 CoNFiLD/ConditionalNeuralField 和CoNFiLD/ConditionalNeuralField 中的模型,跑通对应的 cass 案例。 | ||
|
||
## 6. 可行性分析和排期规划 | ||
|
||
- 202408:调研 | ||
- 202409:复现代码并作调整 | ||
- 202410:整理项目产出,撰写案例文档 | ||
|
||
## 7. 影响面 | ||
|
||
丰富 PaddleScience 的应用案例,在 ppsci.arch 中新增 CoNFiLD model |