Quantinuum has developed a new probabilistic quantum algorithm that moves beyond standard approaches to matrix calculations by leveraging mixed states. Unlike many current methods relying on block encodings, this algorithm directly encodes functions into mixed states, an area researchers say is comparatively less explored. The algorithm allows for flexible computation at each step, offering three distinct actions: returning the current state, applying a completely positive map that does not increase trace, or restarting the process. Researchers note that block encodings and mixed states are incomparable computational resources even when they represent the same data, suggesting potential advantages to this new approach for approximating the normalized inverse of a positive definite matrix and weighted sums or integrals. This work builds on previous results and introduces a deterministic stopping rule for efficient computation with a bounded number of oracle calls.
Probabilistic Algorithm for Lyapunov Equations & Matrix Inversion
This multi-path design allows for nuanced calculations when solving Lyapunov equations and inverting matrices, a capability previously limited by more rigid methodologies. The algorithm generates mixed states, which approximate matrix-valued weighted sums and integrals, expanding the scope of potential applications beyond standard matrix operations. The implications of this research extend to fields requiring complex system analysis, as demonstrated by references to work on open quantum systems and linear matrix inequalities, and could lead to more versatile quantum algorithms.
Mixed State Encoding & Block Encoding Comparison
Current approaches to quantum computation often rely on block encodings to represent data, but Quantinuum researchers explored an alternative path: encoding functions directly into mixed states, a comparatively less investigated area. Building on previous results by Zhang et al., the team created a system that moves beyond simple function representation, offering a nuanced approach to quantum computation. This work, published in Physics Applied on April 16, 2026, highlights a growing recognition that different encoding methods may be best suited for different computational challenges.
While block encodings of functions have received much attention in the literature, our work takes a step toward the less explored problem of encoding functions into mixed states.
Algorithms for solving linear systems were recently detailed in a publication from Quantinuum researchers, moving beyond traditional approaches reliant on block encodings. This multi-faceted design contrasts with simpler iterative methods and enables more nuanced calculations, particularly when addressing Lyapunov equations, which are critical in fields like control theory and stability analysis.
