ALSSM Output Sequence

UNDER CONSTRUCTION

  • ALSSM.eval_alssm() — TODO generates the trajectory of a CostSegment ALSSM model for one or multiple given initial state vectors xs (without considering the CostSegment’s borders or windows).

  • CostSegment.eval_alssm() — generates the trajectory of a CostSegment ALSSM model for one or multiple given initial state vectors xs (without considering the CostSegment’s borders or windows).

  • CompositeCost.eval_alssms() — generates the trajectory of the CompositeCost ALSSM models for one or multiple given initial state vectors xs (without considering the CostSegment’s borders or windows).

Trajectories

ALSSM trajectories denote the deterministic output vector of ALSSMs once the state vector is fixed.

  • CostSegment.trajectories() — generates a trajectory of a CostSegment’s ALSSM, which is the trajectory of the ALSSM bounded by the CostSegment’s window.

  • CompositeCost.trajectories() — generates the trajectory of the CompositeCost ALSSM models for one ore multiple given initial state vectors xs, but only considering the active ALSSMs in a Segment (determined by matrix F) and each bounded by its corresponding window.

  • map_trajectories() — maps a generated trajectory to a vector of length K.

Windows

A window weights a cost function by per-sample weights. The following functions explicitly generate these per-sample weights for a known cost function as represented by a CostSegment or a CompositeCost. Note that these explicit functions are commonly only used during the engineering and developing process, to visualize current windows.

  • CompositeSegment.window() — generates a vector with the per-sample window weights of a Segment (with is an essential part of a cost function as represented by a CostSegment).

  • CompositeCost.window() — generates a vector with the per-segment and/or per sample window weights of a cost function as represented by a CompositeCost.

  • map_window() — maps the generated windows of one or multiple Segment to a output vector of length K.