Researchers have established a fundamental connection between two leading paradigms in quantum functional programming: Quantum Signal Processing (QSP)-based methods and the Universal Hamiltonian Eigenvalue Transformation (UHET). Previously, the relationship between UHET, which transforms Hamiltonian dynamics, and QSP techniques, including Quantum Singular Value Transformation (QSVT), remained unclear despite their evident structural similarities. This new work resolves this gap, demonstrating that UHET can be interpreted as a (randomized) linearization of Generalized QSP (GQSP).

Building on this finding, the researchers have introduced a linearized variant of Hamiltonian-based QSVT, which they call Universal Hamiltonian Singular Value Transformation (UHSVT). This algorithm enables the efficient transformation of the singular values of any arbitrary matrix A, provided it is encoded in a block of a Hamiltonian whose dynamics are accessible as a black box. The transformation is performed by any sufficiently differentiable complex-valued function f.

A key advantage of UHSVT is that it only requires the function f to vanish at the origin. This contrasts with previous QSVT-based approaches, which imposed more restrictive conditions, such as a lower bound on the singular values of A or the ability to perform X-rotation gates on the induced two-dimensional 'qubitized' subspace. This simplification in requirements could broaden the applicability of quantum signal processing techniques in various contexts.