|
Figure 1
Hybrid reverse Monte Carlo of a 2D Huffman-like array, simplified to the case where only the merit factor Mf and spectral flatness dF are to be optimized. A randomly chosen zero of the iteratively evolving 1D Huffman sequence defined by Huffman's P(z) polynomial is moved in (a) by a small distance along a randomly chosen radial direction in the complex plane, as shown by the arrow. The corresponding conjugated zero (open circle vertically below in the bottom half plane) is accordingly moved in a mirrored fashion to ensure that the Huffman-like sequence remains real-valued. The curved arrow in (b) shows the change in the weighted 2D array autocorrelation metrics induced by the Markov event in (a), whereby the sum squared error χ2 slightly increases. |