Pasqal and EDF partner to study smart-charging challenges with Quantum Computing

Quantum Computing startup Pasqal collaborates with the R&D department of electric utility EDF to bring fast solutions to hard optimization problems.

Quantum computers have the potential to solve hard computational problems more efficiently than their classical counterparts. Applications notably encompass computational drug design, materials science, machine learning, and optimization problems. With the rapid developments of quantum hardware, practical quantum advantage is within reach.

With many cities turning to e-mobility to tackle environmental challenges, electric utilities have to account for a growing and more complex load to manage for their production facilities and the grid. One example is the need to schedule resource allocation for shared electric vehicles while taking into considerations their expected and real time availability as well as charging constraints. This class of problem is computationally hard to solve even with large supercomputers and it is expected that a quantum algorithm called Quantum Approximate Optimization Algorithm (QAOA) could improve its resolution.

EDF made smart charging and the development of its infrastructures one of the strong point of its Electric Mobility Plan, launched in October 2018. EDF views smart charging as a true asset for electric vehicle’s users and for the electrical system. Through its subsidiaries, IZIVIA and DREEV, the EDF Group already provides V2G solutions.

Through its Pulse Explorer Program, EDF R&D routinely reaches out to start-ups to explore new ideas in a collaborative way. EDF and Pasqal have formalized a partnership to explore how this algorithm could be implemented on the neutral atoms’ quantum processor developed at Pasqal and take benefit from its unique properties.

The core of the partnership is to finely tune the algorithms according to the hardware’s possibilities and to mitigate the impact of the errors. The level of performance will be gauged on a classical emulator, prior to a real hardware implementation.

Loïc Henriet, head of software development at Pasqal explained: “we have developed our full software stack with specific tools for generic optimization problems, but it is very important that we engage directly with partners working on applications. We need to focus on practical use cases to show that quantum processors can provide a real advantage.”

Marc Porcheron, head of EDF R&D’s Quantum Computing project, said: “utilities such as EDF have to be at the forefront of innovation in high performance computing. It is great to collaborate with Pasqal to explore the new possibilities opened by Quantum Computing for hard optimization problems like the ones we face in the decisive field of smart-charging. I am impressed with the results that have already been achieved with Pasqal, and look forward to implement on their upcoming hardware the quantum algorithms we investigate together.”

To know more

More information on Pasqal’s Quantum Computing stack is available in Pasqal’s technical whitepaper. An abstract is provided here.

Minimizing the total charging time of N vehicles on k distinct charging stations amounts, under some constraints, to solving a combinatorial optimization problem called Max-k-Cut. The objective of Max-k-Cut is to partition the N vertices of a graph into k ensembles, so that the cut is maximal (the cut is the sum of all the weights of edges between nodes that are not in the same group). An example is illustrated on the following figure:

(a) Mapping for seven vehicles on three charging stations. The scheduling and association of each vehicle to a given station correspond to solving a Max-3-cut problem on a 7-nodes graph. The affiliation of a node to a charging station is illustrated by a color-code. The edges contributing to the cost function are illustrated by full black lines, while the ones that are discarded are dashed. The computational complexity of this problem grows exponentially fast with the system size. (b) Optimization landscape for a Max-k-cut problem using QAOA on an emulated Pasqal processor as a function of two variational parameters. One shows the minimum search using both a gradient-based optimizer and a genetic algorithm.

An efficient resolution of Max-k-Cut with a variational procedure involving N ln k qubits has recently been devised by Pasqal’s research team, thus unlocking the potential of Quantum Processors for real-world operational problems. One illustrates in the figure (b) the optimization landscape for one instance of a Max-k-cut problem. As can be seen in the plot, the optimization landscape is very irregular, requiring the design of efficient software methods in the classical optimization loop of the procedure as described in Pasqal’s whitepaper, a genetic algorithm in this instance.