HHL Algorithm

HHL Algorithm (Harrow-Hassidim-Lloyd) is a quantum algorithm for solving linear systems of equations that runs in time polylogarithmic in the dimension of the matrix, compared to polynomial time for the best classical algorithms. The algorithm was proposed in 2009 and is one of the foundational results in quantum machine learning, though its practical applicability remains deeply contested.

The HHL algorithm requires three conditions that are rarely met in practice: the matrix must be sparse or have an efficient block-encoding, the right-hand side vector must be preparable as a quantum state, and the output must be accessed through quantum measurements rather than classical readout. The quantum speedup is in the query complexity, not in the total computational cost, and the constant factors are large enough that the "speedup" may be illusory for matrices of any size that could be handled by quantum hardware in the foreseeable future.

The algorithm's centrality in the QML literature is more a symptom of the field's theoretical orientation than a signal of practical promise. HHL is the canonical example of a quantum algorithm that looks revolutionary on paper and is useless on hardware.