Rigetti says quantum preconditioning made Gurobi around 100 times faster on a test problem

[Highland Canopy]

Rigetti has published a blog post on Medium about using its quantum hardware to speed up a commercial optimisation solver. The solver in question is Gurobi, which is widely used in industry for problems like planning delivery routes, balancing power networks and allocating resources. Rigetti calls its approach Quantum Preconditioning, and this work extends it to problems with hard constraints. The post is written by lead quantum researcher Maxime Dupont

The idea is not to replace the classical solver but to give it a better starting point. A quantum algorithm called QAOA runs first and produces information that guides Gurobi's search towards good solutions. On a benchmark called balanced graph bi-partitioning with 40 variables, the preconditioned version of Gurobi reached solutions within 1 percent of the best possible answer roughly 100 times faster than Gurobi on its own. The improvement was there even at the shallowest circuit depth tested, which matters because shallow circuits are less affected by noise

The work was done with Purdue University's SECQUOIA group, with funding from the US Department of Energy's Superconducting Quantum Materials and Systems Center. That gives it some academic weight beyond a company announcement. Still, it is a blog post rather than a peer reviewed paper, so the full details need checking

There are clear caveats. The tests only used a single hard equality constraint, and 40 variables is small by real world standards, where industrial problems can have thousands or millions. Rigetti says its next steps are problems with multiple and mixed constraints. It is also a little unclear how much of the gain would hold up as problems scale

Even so, this hybrid approach feels like a realistic path to useful quantum computing, with the quantum part helping rather than taking over. It fits with what Bain and others have said about AI and computing value coming from practical gains. Is guiding classical solvers the most promising near term use of quantum? Or will classical improvements keep moving the goalposts?

Quantum Boosting A Commercial Optimization Solver 2766Cba259D1medium.com