Home etc Blogs Google Releases TensorNetwork, an Open Source Library for Efficient Tensor Calculations

Google Releases TensorNetwork, an Open Source Library for Efficient Tensor Calculations

0
2722
Google

 

  • Google introduces TensorNetwork in a series of papers
  • TensorNetwork uses TensorFlow as a backend
  • Significant computational speed-ups of up to 100x achieved when using a GPU and TensorNetwork library

Google

A new open source library is seen as a valuable tool for solving many of the world’s toughest scientific challenges because it could help scientists understand the complexity of quantum systems better.

To deal with the complexity of quantum systems, scientists use data structures called tensor networks. These structures enable them to focus on the quantum states that are most relevant for real-world problems while ignoring other states that aren’t relevant.

However, widespread use of tensor networks in the machine learning community is hindered by the following reasons:

1) a production-level tensor network library for accelerated hardware has not been available to run tensor network algorithms at scale

2) most of the tensor network literature is geared toward physics applications and creates the false impression that expertise in quantum mechanics is required to understand the algorithms.

To address the above challenges, Google has released TensorNetwork, an open source library for efficient tensor calculations.

Google AI team, who developed the library in collaboration with the Perimeter Institute for Theoretical Physics and X, is hoping that it will become a valuable tool for physicists and machine learning practitioners in future.

Performance in Physics Use-Cases

Google introduces TensorNetwork in a series of papers, the first of which presents the new library and its API and provides an overview of tensor networks for a non-physics audience. Their second paper focuses on a particular use case in physics, demonstrating the speedup that one gets using GPUs.

TensorNetwork uses TensorFlow as a backend and is optimized for GPU processing, which enables speed-ups of up to 100x when compared to work on a CPU.

“We compare the use of CPUs with GPUs and observe significant computational speed-ups, up to a factor of 100, when using a GPU and the TensorNetwork library,” Google writes in its open source blog.

“TensorNetwork is a general-purpose library for tensor network algorithms, and so it should prove useful for physicists as well,” it adds.

In future, Google is planning time series analysis on the ML side and quantum circuit simulation on the physics side. Together with the open source community, it will also continue to add new features to TensorNetwork.

NO COMMENTS

LEAVE A REPLY

Please enter your comment!
Please enter your name here