Given a finite-dimensional memoryless quantum channel N through an explicit rational/algebraic Choi matrix and a rational ε>0, does a single algorithm always terminate and output q satisfying ∣q−Q(N)∣≤ε? Here Q is the ordinary unassisted quantum capacity, in qubits per channel use.