TU Wien, Bull and DataTeach.ai Propose Quantum-Assisted Counterfactual Regret Minimization for Imperfect-Information Games Without a Hardware Run, arXiv 2611.14182, 10 November 2026: TRL 2

Gabriel Maresch, Erik Schulze and Jnan Yalla of TU Wien, Bull GmbH and DataTeach.ai posted to arXiv on 10 November 2026 a framework that places quantum amplitude estimation inside counterfactual regret minimization, the solver behind the poker programs, and proved that the exploitability bound keeps its classical convergence term plus an additive quantum-estimation term. The query count per counterfactual value falls from order 1/ε² to order 1/ε, the circuits are sized in qubits and depth, and a noise model shows gate error capping the reachable precision after a few tens of layers. This post reads the paper as dual-use quantum-assisted game solving for security games and course-of-action analysis at TRL 2, explains why the authors report no simulator or hardware result by choice, and lists the Kuhn or Leduc poker benchmark, the concrete oracle and the device-specific crossover depth a program office should wait for.