Robust identification from partial frequency data

Being given noisy band-limited pointwise measurements of a function f belonging to the disc algebra, we provide a computational procedure to build an approximation which robustly converges to f on the range of frequencies (where the measurement points are assumed to be dense) as the number of data tends to ∞ and the l^∞ noise level goes to 0, while staying within some Lipschitz-continuous tolerance from a given behaviour outside the bandwidth, f being assumed to meet this tolerance