Identification-assisted energy-structured residual learning for ROV trajectory tracking: simulation code and output

 Simulation code for the numerical study accompanying the manuscript Identification-Assisted Energy-Structured Residual Learning for Robust ROV Trajectory
  Tracking under Hydrodynamic Uncertainty and Tether Disturbances.

  The package contains a six-degree-of-freedom marine-craft benchmark with the true and nominal effective inertia of Table 1 and the damping and gain
  vectors of Appendix A; a finite-window Artificial Bee Colony with variable-neighborhood refinement for the per-axis identification of Section 3; the
  energy-structured residual learner of Section 4, whose dissipative matrix is fitted by least squares and projected onto the positive-semidefinite cone;
  the four controllers of Section 7.1; the sampled-data closed-loop driver; and the twenty-realization Monte Carlo study of Section 7.3.

  Section 7.1 requires every controller to be driven by the same realization, so the per-realization errors are paired. The Monte Carlo script therefore
  reports the paired t-test alongside the two-sample test, which is the statistic the paired design actually licenses.

  This is an independent re-implementation from the description in the manuscript, not the original development code, which was not available. README.md
  lists every value the manuscript fixes and every value it leaves open, and records the respects in which this implementation disagrees with the published
  numbers rather than tuning those disagreements away.