Quantum computers are entering the scaling phase, where processors with increased size will enable new computational capability which is not possible with classical computers. A critical enabling step for this new phase is optical links, which will be a cornerstone technology of distributed quantum computing data centers.
What is distributed quantum computation and why is it important?
Distributed quantum computation is a method by which a quantum computer can be constructed out of discrete modules networked together. Today’s most advanced high performance computing gains its strength from networking extremely large numbers of cores together, and this same approach will be required to bring quantum computers to the scale where they can start tackling the most challenging and useful problems. Optical links play a particularly important role here, as the high frequency of the optical photons enable the quantum states to be routed between modules at room temperature, avoiding the need for ever-larger cryogenic systems, and enabling a flexible, reconfigurable and modular approach to constructing the quantum computer.
What does a distributed quantum computer look like?
In practice, distributed quantum computers will contain many interconnected quantum computing modules, somewhat similar to a classical data center. On the system level, however, the computer can be split into multiple layers that handle quantum information at increasing levels of abstraction. The highest level (yellow layer in above Figure) is composed of many quantum processors which are linked together (pink layer). Each link is constructed from a number of entangled states: the base resource for remote gates (blue layer). At the fundamental level, entanglement is generated by hardware components such as the qubits of the quantum processor, microwave-to-optics transducers for connections to optical links and quantum memories (green layer). Together the layers extend computations from a single quantum processor to the whole quantum computer. Each layer passes requests to the layer below and delivers resources to the layer above. In order to simplify the operation of the system, it is then important that each layer can act quasi-autonomously.
What are we covering here?
In our work, we outline a way to connect quantum computers together in a distributed architecture. As an illustrative example, we focus on connecting superconducting quantum processors together using optical links.
In the main body of the paper we work our way up the stack of the quantum computer. We start by looking at the transducers that enable the quantum computer to connect to the optical network. We outline the most important metrics for understanding the quality of a particular transducer, including metrics for how easy it is to operate many channels in parallel. We then briefly examine the state of the art for these metrics.
We then look at transduction protocols which create the critical resource for distributed quantum computation: entanglement between separated computing nodes. We examine how different protocols affect either the rate or quality of entanglement distribution, and how other technologies such as quantum memories can help boost performance. Due to optical loss creating distant entangled states is inherently probabilistic, but with sufficient rates and state storage, the protocols can deliver high fidelity entangled states at pre-defined times, or ‘on-demand’. Operating this way greatly simplifies the relationship with the other layers in the technology stack. With these protocols established, we then look at the rates and quality of entangled states that we expect could be delivered in the near future.
Finally, we reach the top of the technology stack and review some of the leading distributed quantum computing architectures and look at how they can be enabled with the devices and protocols we have introduced.
What are the most important takeaway messages?
- On-demand entanglement between remote quantum processors via optical links can be expected within the next few years.
- Many transducers operating in parallel are necessary for full scale quantum computing: integration and scaling are crucial.
- Combining different quantum technologies together such as microwave-to-optics transducers and quantum processors or quantum memories is a key step towards distributed quantum computing.
What are the main challenges?
Quantum processors operate at high rates and are susceptible to errors. This requires larger quantum processors to provide redundancy for error correction. While enabling links over much greater scales than available through direct microwave-frequency connections, microwave-to-optics transduction is typically less efficient and noisier than local connections. The clearest improvements to the quality of optical links come from core performance improvements at the component level. However, increasing the number of channels operating simultaneously, or merging multiple entangled states through a process called distillation can provide effective routes to improving the rate or quality of the entangled states in the system.
Why write this?
There are many ideas in the community about distributed quantum computing, quantum processor architectures, microwave-to-optics transducers and other supporting hardware technologies, and the range of possible realisations can be a bit overwhelming. By reviewing these works and presenting a few concrete solutions, we direct readers through some of the many possibilities for near-term distributed quantum computing and connect hardware metrics to more abstract system performance. We hope that this helps quantum hardware and software developers, and other quantum enthusiasts to adopt these technologies early: accelerating the expansion of quantum computing technology.
What’s next?
Now it’s time to make this happen. QphoX is in a strong position as a manufacturer of microwave-to-optics transducers to improve their performance and scale up the number of transducers which are available. We will work with partners to combine transducer technology with quantum processors and quantum memories. This is part of a broader effort of the community to push towards distributed quantum computing.