European Consortium to Develop Quantum Annealer under AVaQus

A European consortium has been awarded €3.0 million from the European Commission’s Horizon 2020 FET Open for a three-year program under the

AVaQus (Annealing-based VAriational Quantum processors) program to design and fabricate a fivequbit quantum annealer with high connectivity tunable interactions, and long coherence time. This project aims to set the stage for next-generation quantum computing systems capable of performing computations and simulations with far greater speeds and efficiencies than classical computers by

utilizing recent developments in superconducting quantum circuits.

Researcher Pol Forn-Díaz, the head of the Quantum Computing Technologies group at the Institute for High Energy Physics (IFAE) in Barcelona is responsible for initiating and coordinating the project. “As coherent quantum annealing technology has continued its progress, an initial attempt at the FET Quantum Flagship was re-elaborated into a FET-Open proposal, since AVaQus targets truly innovative technologies with significant market potential,” explained Form-Diaz.

Quantum Annealers Appropriate for Optimization + Quantum Simulation

Quantum annealers are believed to be more suitable for solving certain types of problems than classical processors, with a shorter time to market than universal quantum computers. “Quantum annealers are not universal quantum processors

in the sense that they cannot just solve any type of quantum algorithm,” commented Forn-Diaz. “But the problems they can solve, optimization +

quantum simulation, are very much ubiquitous in our society, hence very relevant.

“The technological demands are less than gate-based quantum processors as the system remains at its lowest energy state all the time. This means the quantum annealer size, in number of qubits, can grow very rapidly.”

If successful, the system developed by AVaQus would be the first superconducting annealer to perform quantum computation and simulation tasks with hardware designed for coherence. “[This] means that the qubits, and their control and readout circuitry will be designed with the purpose of maintaining a high level of quantum coherence of the qubits,” clarified Forn-Díaz.

“By high coherence we mean that the qubit state will not decohere during the duration of the computation. In other words, we will be using the same kind of superconducting qubit technology that is already working well in gate-based digital quantum processors.”

Forn Diaz noted that quantum annealing may have a positive effect on coherence: “The hardest problems are ones where the solution, encoded in the lowest energy state of the system, is just marginally below the higher excited levels, and the tiniest amount of noise will take you away from your optimal point.

Another area where coherence seems to help is the speed at which the annealing algorithm can run, as there is a connection between quantum tunneling, which affects the speed at which you can anneal, and low dissipation in the system, something you get when qubits are highly coherent.”

Annealers Use Adiabatic Quantum Computing Techniques

The majority of research on quantum systems is centered on universal gate-based quantum computers, an approach that requires sizable arrays of qubits to correct for noise-induced error for operation. Leading quantum computer companies, including IBM, Google and Intel, are currently focused on Noisy Intermediate Scale Quantum (NISQ) devices that are able to function without error correction.

“Quantum annealers are different from gate-based quantum processors in that they use adiabatic quantum computing techniques, which do not require quantum gates, to solve problems,” explained Forn-Díaz. “The computation is performed by evolving the system with all interactions among qubits switched off into a state in which interactions dominate and generate strong correlations, all while keeping the system in its ground state. Pictorially, initially all qubits are equal and don't know anything about each other, but at the end of the computation certain qubits influence their neighbors, resulting in a very complicated arrangement of positive and negative interactions, leading to a highly entangled ground state."

Noisy Processes Less Damaging to Annealer

According to Forn-Diaz, the analog quantum computing of an annealer may be less prone to errors than universal quantum computing systems: “We do not claim that our qubits will be more resistant to quantum gate errors, because there

are no gates used in annealing. What we mean is that the type of errors that plague gate-based quantum processors will not affect us all that much because the way in which information is processed is completely different. In particular, the annealer never leaves the ground state of the system, and therefore the noisy processes that induce decoherence in gate-based processors appear to be less damaging to the operation of a coherent quantum annealer.”

To date, Burnaby, British Columbia-based D-Wave Systems has been the primary commercial developer of quantum annealing devices. In spite of their significant technological accomplishments, the extent to which D-Wave’s computing technology qualifies as quantum technology has generated debate by some academics. In particular, some researchers question whether the company’s system has shown quantum entanglement, an indicator that it relies on quantum effects.

“Their device is not a coherent annealer, in the sense that the coherence time of qubits is much shorter than the total computational time of their system,” Forn-Diaz pointed out. “Some refer to D-Wave devices as incoherent quantum annealers. 

“There is an IARPA-funded project in the US involving groups from around the world named QEO (quantum enhanced optimization) led by MIT that is also aiming at building a coherent quantum annealer. The project is about midway (out of five years) and has so far reported work using 1-2 coherent qubit prototypes.”

Innovations to Include Qubit-qubit Interactions

Forn Diaz explained how the AVaQus team plans to improve upon existing designs: “We want to implement novel types of qubit-qubit couplings which are mathematically harder to deal with for classical computers. In this way, the problems a coherent annealer can solve will be less likely to be solved in a classical computer. The challenge is in identifying such problems.

“For that, AVaQus will have three theory/software teams addressing this important front. In annealers you need very large qubit-qubit interactions as compared to gate-based approaches. Also, the interactions must be switchable and adjustable, something that Google recently introduced in their device whereby they claimed quantum supremacy.

“In the AVaQus approach, the coupling circuits we aim at developing introduce additional physical effects. If we think of qubits as spin-1/2 particles, it would be as if we would want to have spins interact with more than one component of their spin.

“Normally, in annealers and in gate-based processors, the interaction operates along one component, say XX. We want to implement couplings like XX + ZZ, which turn out to be very tough to deal with by classical processors and may lead to the long-sought quantum advantage if one is able to wire up together a sufficiently large amount of qubits.

“In no way can we think of outperforming classical processors with five qubits. However, this five-qubit demonstrator will set the path to scaling up to a large-size device that should be the goal of the following generation of devices in a follow-up project.”

Annealer Development and Rollout

The system will incorporate applications of smallscale quantum annealing algorithms that can be employed for simulations and optimization in logistics, navigation, traffic, finance, quantum chemistry, and machine learning. The qubits will employ aluminum circuits and either silicon or sapphire as substrates. The annealer will need to be cooled via dilution refrigeration to a temperature of 10 mK.

The researchers intend to work in two parallel lines of development, each one using a different qubit and coupler type, namely low impedance and high impedance circuits. At the end of the project, one of the goals is to assess which of the two types of qubits is more suitable to scale-up the device in a later project.

AVaQus is a consortium of eight European partners consisting of five research centers: IFAE, the Karlsruhe Institute of Technology (KIT), the French National Center for (CSIC), and three quantum startup companies: Delft Circuits in the Netherlands, Qilmanajaro Quantum Tech, S.L. (QILI) in Spain, and Heisenberg Quantum Solutions (HQS) in Germany.

They are slated to launch the AVaQus project on October 1, 2020.

OpenSuperQ from the FET Flagship on Quantum Technologies (FET-QT) and the QuantERA-funded project SiUCs (Superinductor-based Quantum Technologies with Ultrastrong Couplings), also coordinated by IFAE, will be collaborating with the AVaQus initiative. Delft, CRNS, and Glasgow will be responsible for designing and fabricating the enabling quantum and classical hardware, while IFAE and KIT will design and validate different types of qubit circuits operating as coherent quantum annealers. HQS, QILI, and CSIC will work on developing quantum software and applications that will eventually run on the coherent quantum annealer, as well as on studying highly connected topologies in larger-scale future devices.