Double-Bracket Algorithmic Cooling with Multi-Color Drives on Superconducting Qubits

Resetting a qubit with arbitrarily high fidelity remains a central challenge for scalable quantum computing. Algorithmic cooling offers an alternative by redistributing entropy through unitary operations. Here, we introduce and experimentally implement Double-Bracket Algorithmic Cooling (DBAC) - a unitary reset protocol that suppresses quantum coherence of pure states by simulating quantum imaginary-time evolution through recursive unitary synthesis. DBAC employs density-matrix exponentiation as a subroutine, making it a concrete realization of a dynamic quantum algorithm that processes information encoded in copies of the input state. Implemented on a superconducting qubit processor using native ZZ interactions, DBAC performs state-independent, measurement-free cooling, extending algorithmic cooling from entropy redistribution to coherence suppression. These preliminary results suggest that DBAC could provide a coherence-preserving primitive for on-demand qubit reset and point toward the potential of dynamic quantum algorithms for foundational operations in quantum thermodynamics.