Google LLC today open-sourced TensorFlow Quantum, an extension to TensorFlow that enables developers to build artificial intelligence models for quantum computers.
Quantum computing is still in an early stage, but the technology is maturing rapidly. IBM Corp. is doubling the processing power of its quantum chips every year, while Honeywell International Inc. recently unveiled a system that it expects to speed up by a factor of 100,000 in the next five years.
To take advantage of tomorrow’s ultrafast quantum machines, researchers will have to write specialized algorithms that can run on qubits, which unlike binary bits can be represented as 1s, 0s or both states at the same time, and they’ll need equally specialized development tools to help with the task.
That’s where TensorFlow Quantum comes into the picture. It provides a set of operators, low-level programming building blocks, for creating AI models that work with qubits, quantum logic gates and quantum circuits. These operators abstract away some of the underlying complexity to reduce the amount of code researchers need to write.
“TFQ allows researchers to construct quantum datasets, quantum models, and classical control parameters as tensors in a single computational graph,” Google researchers Alan Ho and Masoud Mohseni wrote in a blog post.
One potential application for TensorFlow Quantum is quantum data interpretation. Because a qubit can be set to both 1 and 0 at the same time, finding out the results of a computation carried out by a quantum processor is by itself a major challenge. According to Google’s Ho and Mohseni, TensorFlow Quantum could enable engineers to develop AI models that disentangle quantum data automatically.
“The TFQ library provides primitives for the development of models that disentangle and generalize correlations in quantum data, opening up opportunities to improve existing quantum algorithms or discover new quantum algorithms,” the researchers wrote.