Improved quantum lower and upper bounds for matrix scaling

Matrix scaling is a simple to state, yet widely applicable linear-algebraic problem: the goal is to scale the rows and columns of a given non-negative matrix such that the rescaled matrix has prescribed row and column sums. Motivated by recent results on first-order quantum algorithms for matrix scaling, we investigate the possibilities for quantum speedups for classical second-order algorithms, which comprise the state-of-the-art in the classical setting. We first show that there can be essentially no quantum speedup in terms of the input size in the high-precision regime: any quantum algorithm that solves the matrix scaling problem for n × n matrices with at most m non-zero entries and with ℓ2-error ε = ̃Θ(1/m) must make ̃Ω (m) queries to the matrix, even when the success probability is exponentially small in n. Additionally, we show that for ε ∈ [1/n, 1/2], any quantum algorithm capable of producing ε 100 -ℓ1-approximations of the row-sum vector of a (dense) normalized matrix uses Ω(n/ε) queries, and that there exists a constant ε0 > 0 for which this problem takes Ω(n1.5) queries. To complement these results we give improved quantum algorithms in the low-precision regime: with quantum graph sparsification and amplitude estimation, a box-constrained Newton method can be sped up in the large-ε regime, and outperforms previous quantum algorithms. For entrywise- positive matrices, we find an ε-ℓ1-scaling in time ̃O(n1.5/ε2), whereas the best previously known bounds were ̃O(n2polylog(1/ε)) (classical) and ̃O(n1.5/ε3) (quantum)