A major obstacle in the development of powerful quantum computers is the growing number of cables required to control a computer as the number of qubits increases. Researchers at Chalmers University of Technology in Sweden have now demonstrated that several qubits can share the same cable – without significantly increasing computation time. Their study is the most comprehensive of its kind and could become an important piece of the puzzle in developing quantum computers. These computers have the potential to revolutionise such areas as drug development and logistics.
The power of quantum computers lies in what are known as “qubits”. Unlike a conventional computer “bit”, which can have the value 1 or 0, a qubit can have the values 1 and 0 simultaneously – and all states in between, in any combination. This means a quantum computer with 20 qubits can simultaneously represent a combination of more than one million different states, resulting in enormous computational power.
“The global quantum technology race is in full swing, with tech giants currently in the lead with quantum computers based on more than 100 qubits. But to solve real-world societal challenges, quantum computers will need grow much further in size, with thousands or more well-functioning qubits,” says Anton Frisk Kockum, Associate Professor of Applied Quantum Physics at Chalmers University of Technology. At Chalmers, researchers have been developing Sweden’s largest quantum computer within the Wallenberg Centre for Quantum Technology.
Engineering challenges slow scaling
However, scaling up quantum computers comes with practical challenges. For many types of quantum computers to work – including those based on superconducting circuits – they must be cooled to temperatures close to absolute zero, that is -273.15°C. Cooling is achieved using helium in cryostats surrounding the quantum computer. To control quantum computations, signals are sent through cables from electronics outside the system to the cooled qubits inside it. But those cables emit heat that affects the temperature inside the cryostat, something which risks causing the qubits to lose their ability to continue the computation.
“Since each qubit currently requires its own cable, there’s a limit to how many qubits a system can contain before the temperature becomes too high and the quantum computer stops working. There are also physical limitations, since the cables take up space in the cryostat,” says Ingrid Strandberg, a staff scientist in quantum technology at Chalmers.
Smart cable sharing challenges previous concerns
An alternative but relatively unexplored approach is to allow several qubits to share the same cable. Instead of controlling qubits in parallel with one cable each, they are controlled sequentially in rapid succession using fewer cables. The process requires microwave switches to be installed next to the quantum processor to route each control signal to its target qubit – a procedure known as time-domain multiplexing. However, the method involves a presumed trade-off. If qubits must “wait” their turn to receive signals, computations may take longer. To investigate how significant this delay really is, the researchers tested how different types of quantum processors are affected when the number of control cables is reduced. The results were surprisingly positive.
“We can see that for many common quantum algorithms, the number of cables can be drastically reduced without the computations becoming significantly slower or the runtime increasing very much. In some cases, such as the gates that connect two qubits, you can even share cables with no additional time cost at all, limited only by how the qubits are interconnected,” explains Marvin Richter, a PhD student in quantum technology at Chalmers and the study’s lead author.
Important step towards large-scale quantum computers
The Chalmers researchers’ computer simulations and mathematical analyses are the most comprehensive to have been conducted in this area. A particularly important conclusion from the results is that computation time increases logarithmically, not linearly, when individual qubits share cables.
“That's a slower increase than previously feared,” says Simone Gasparinetti, Associate Professor of Quantum Technology at Chalmers and co-author of the study. “Allowing multiple qubits to share cables could be an important step towards large-scale quantum computers. These results give us even stronger motivation to develop the necessary fast, low-dissipation microwave switches to implement this technique.”
More about the study
In the theoretical study, computer simulations were performed on quantum processors of different sizes, up to around 1,000 qubits. The main focus was on a processor with 121 qubits arranged in an 11×11 grid. In the study, the researchers varied the number of qubits per cable from one per cable and up to 121. In simulations of the largest systems, with up to 1,000 qubits, as many as eight qubits per cable were tested.
The article “Overhead in Quantum Circuits with Time-Multiplexed Qubit Control” has been published in the journal PRX Quantum.
The authors are Marvin Richter, Ingrid Strandberg, Simone Gasparinetti, and Anton Frisk Kockum, all active at the Department of Microtechnology and Nanoscience at Chalmers University of Technology. Numerical results were produced using high-performance computing resources at the Chalmers Centre for Computational Science and Engineering (C3SE).
The study was funded through support from the Knut and Alice Wallenberg Foundation via the Wallenberg Centre for Quantum Technology (WACQT), the Swedish Foundation for Strategic Research, the EU Horizon Europe programme, the OpenSuperQPlus100 project, the European Union and the European Research Council (ERC).