Quantum computers separate order from disorder – only then can the quantum physical processes computed by these systems take place. But as analysis using three different methods has now shown, some of the current quantum computers from IBM, Google and Co are dangerously close to the threshold of a chaotic collapse. This could mean that not all of these superconducting transmonic quantum bits systems are easily scalable, the research team explains. To prevent these systems from falling into chaos, this must be considered when designing qubit processors on a superconducting platform.
Quantum computers are considered the computers of the future. Because thanks to quantum physical processes such as superposition and entanglement, they can test many possible solutions simultaneously and are therefore faster than conventional computers. Companies like IBM, Google and others have already developed the first commercially useful quantum computers, even though they are still too small for most applications and contain too few quantum bits. Most of these systems are based on what are known as transmon qubits. These are virtual particles in the form of charge islands in special superconducting metal coils. For these qubits to perform arithmetic operations, they must remain superimposed as seamlessly as possible while computing. This requires the best possible shielding against external disturbances. On the other hand, they must be connected to each other to form circuits.
Between feedback and disorder
Thus, current quantum computers operate in a fragile equilibrium between order in the form of a coupling between qubits and disorder, which gives the individual qubits sufficient leeway for independent fluctuations. “Chip Transmon not only tolerates, but actually requires random qubit-qubit imperfections,” explains first author Christoph Berke of the University of Cologne. Because coupled quantum bits resemble a system of coupled pendulums, the fluctuations of which can easily lead to uncontrolled large oscillations with catastrophic consequences. The principle of operation is similar to the resonance effect of bridges: when large groups cross them, they must avoid a step march, otherwise resonant vibrations arise.
Also in quantum computers, a deliberately introduced disturbance is to avoid the appearance of such chaotic resonance fluctuations. Deliberately introduced local “tuning” prevents too much coupling of qubits and maintains the delicate balance between order and clutter in multi-bit processors. Current quantum computing systems use various methods to maintain this balance. IBM uses a structure in which qubits of the same frequency are placed alternately in the network to prevent unwanted coupling to adjacent qubits. However, as Berke and his team explain, there may still be feedbacks to the penultimate qubit. On the other hand, quantum computers at TU Delft and Google use active point interference to block unwanted resonances.
An IBM system potentially more prone to chaos
How well these different methods work, Berke and his colleagues have now explored using three different techniques. “In our study, we look at how plausible the principle of ‘stability by chance’ is in practice,” says Berke. On the one hand, it has been shown that even before plunging into chaos, there is a surprisingly vast gray zone in which unwanted resonances are already influencing quantum states, but the threshold of “hard quantum chaos” has not yet been exceeded. On the other hand, tests have shown that at least some of the system architectures used in industry are dangerously close to instability. “When we compared Google systems with IBM systems, we found that in the latter case, the qubit states can be coupled to such an extent that controlled computational operations can be disrupted,” reports Berke colleague Simon Trebst. The system of this computer architecture is more prone to chaotic fluctuations.
This could have ramifications for the planned expansion of quantum computers with more quantum bits: ‘We would go as far as to speculate that this system cannot handle the generalization to larger and two-dimensionally connected matrix geometries that are necessary for more complex applications,’ the researchers write. Senior author David DiVincenzo of RWTH Aachen University adds: “Our study demonstrates how important it is for hardware designers to combine device modeling with state-of-the-art quantum random methods and to make ‘chaos diagnostics’ a routine part of the qubit design for integration with a superconducting platform.”
Source: Christoph Berke (University of Cologne) et al., Nature Communications, doi: 10.1038 / s41467-022-29940-y