Robust Bootstrap estimation of Subset ARX systems

Performs various robust estimations of subset ARX models, with block-bootstrap computation of their standard errors for identification.

ARX_roboot.m performs robust estimation with various criterions (e.g. winsorized OLS, least absolute deviations (LAD), least trimmed of squares (LTS), M-estimation with bisquare loss, etc.) of single-input subset ARX models. Inference and identification of the subset structure is performed with block-bootstrap computation of their standard error. ARX_roboot_demo.m illustrates the various methods on simulated series with comments.