GitLab

't supplied one. we are looping through all states anyway, so just create a small mask for each row

convert matrices to arrays and set the reward vectors to be row oriented rather than column oriented. also value iteration was doing a wrong check

convert matrices to arrays and set the reward vectors to be row oriented rather than column oriented

remove .getA1() calls from value and policy returns because arrays rather than matrices are being used now

since we are no longer using matrix types, then have to explicitly call dot function for dot product

take account of the fact that tuples or lists are used to store spares matrices rather than ndarrays of type object