JIJ's SQOA solver lands in IBM's Functions Catalog

Started by CyberWarden49, Yesterday at 06:00 AM

Previous topic - Next topic

0 Members and 1 Guest are viewing this topic.

Topic: JIJ's SQOA solver lands in IBM's Functions Catalog   Views(Read 90 times)
Active members in this topic:
CyberWarden49(1)

CyberWarden49

A Japanese company has added an optimisation tool to IBM's quantum platform. Quantum Computing Report says Tokyo based JIJ has launched its SQOA solver, which combines variable compression with classical subspace diagonalisation

The approach avoids variational loops on the hardware. A Quantum Random Access Code relaxation fits up to three binary variables per qubit

On Heron r2 hardware, it tackled a 384 node MaxCut problem on 129 qubits with a 98.5 percent approximation ratio and 12 seconds of QPU time. A 200 node Maximum Independent Set problem on 67 qubits reached 100 percent

It accepts several common input formats, and its largest tier uses 48 CPUs and 768GB of RAM. It is based on a paper presented at IEEE Quantum Week 2025

Fitting three variables per qubit lets the solver tackle bigger problems than the hardware size suggests. Optimisation problems appear everywhere, from logistics to scheduling. Short QPU times also help keep costs down

Which real world problems should quantum optimisation tackle first? And are compression tricks like this the key to useful results?