The need for zero-dimensional tensors.

[AndPuQing]

需求描述 Feature Description

I am now trying add Paddle in deepxde(https://github.com/lululxvi/deepxde) as backend. For some reason, we need to distinguish between zero-dimensional tensors and 1D tensors, as follows：

import torch

a = torch.tensor(3)
print(a.ndim) # 0
b = torch.tensor([3])
print(b.ndim) # 1

import paddle
a=paddle.to_tensor([3])
print(a.ndim) # 1
b=paddle.to_tensor(3)
print(b.ndim) # 1

替代实现 Alternatives

No response

[paddle-bot-old]

您好，我们已经收到了您的问题，会安排技术人员尽快解答您的问题，请耐心等待。请您再次检查是否提供了清晰的问题描述、复现代码、环境&版本、报错信息等。同时，您也可以通过查看官网API文档、常见问题、历史Issue、AI社区来寻求解答。祝您生活愉快～

Hi! We've received your issue and please be patient to get responded. We will arrange technicians to answer your questions as soon as possible. Please make sure that you have posted enough message to demo your request. You may also check out the API，FAQ，Github Issue and AI community to get the answer.Have a nice day!

[jzhang533]

@iclementine is investigating on this.

[iclementine]

Currently, paddle has limited and inconsistent support for zero-dimensional tensors due to a previous design to fake a zero-dimensional tensor with a 1d tensor with shape [1]. After that, a lot of APIs has been hacked to deal with zero-dimensional tensors, which is incorrect mathematically.

For example, indexing an 1d tensor with an integer returns a 1d tensor with shape [1] instead of a 0d tensor. Reduction on a tensor at all dimensions returns a 1d tensor with shape [1] instead of a 0d tensor.

It's a known issue and we are planning on fixing it. But fixing this requires a throughout scanning of all APIs, which might requires some time.

[AndPuQing]

Right now I'm using a 1D tensor to simulate a zero-dimensional tensor which is indeed problematic, so expect the problem to be solved. Thank you for your reply

[zhwesky2010]

@AndPuQing Hello, is there other API except to_tensor? We are considering supporting 0D, but Many APIs need to be modified. If you encounter any problem with the API of 0D, you can report this API to us, and we will support it firstly.

[zhwesky2010]

@AndPuQing 你好，飞桨将于2.5版本全面支持0维Tensor，上述反馈的0D问题也将同时解决。

欢迎试用并安装develop 版本的paddle whl包，将体验最新的0维特性：https://www.paddlepaddle.org.cn/install/quick?docurl=/documentation/docs/zh/develop/install/pip/linux-pip.html
安装命令：
pip install paddlepaddle-gpu==0.0.0.post112 -f https://www.paddlepaddle.org.cn/whl/linux/gpu/develop.html

---

Hello, Paddle will fully support the 0-dimensional Tensor in version 2.5, and the 0D issue will also be resolved at the same time.
Welcome to try and install the develop version of the paddle whl package, and you will experience the latest 0-dimensional features: https://www.paddlepaddle.org.cn/install/quick?docurl=/documentation/docs/zh/develop/install/pip/linux -pip.html
Install Command:
pip install paddlepaddle-gpu==0.0.0.post112 -f https://www.paddlepaddle.org.cn/whl/linux/gpu/develop.html