Home > marl > replay.m

replay

PURPOSE ^

Replays the agents' current policies

SYNOPSIS ^

function [world, agents, stats] = replay(world, agents, speed, maxiter)

DESCRIPTION ^

Replays the agents' current policies
  [WORLD, AGENTS[, STATS]] = REPLAY(WORLD, AGENTS[, SPEED[, MAXITER]])
  Replays the agents' policies over a single trial. Assumes that the
  agents have already been correctly setup during the learning process.

  Parameters:
   WORLD       - the world where the agents live
   AGENTS      - the (possibly heterogeneous) cell array of agents
   SPEED       - (optional) the speed of the replay. Ranges between 1 and
               10, 1 is (roughly) one second per iteration, 10 is full
               speed. Default is 7.
   MAXITER    - (optional) how much to allow the policy to run at most.
               Defaults to 1000 iterations. May be -1 ('run forever').
  Returns:
   WORLD       - the world, altered by the agents' actions during replay
   AGENTS      - the possibly altered agents.
   STATS       - replay statistics such as:
       'iter'      - how many iterations it took to terminate the task
       'stuck'     - boolean flag signaling whether the agents got stuck


  See also learn

CROSS-REFERENCE INFORMATION ^

This function calls: This function is called by:
Generated on Wed 04-Aug-2010 16:55:08 by m2html © 2005