SimulateFrom
PURPOSE 
Simulates a POMDP from a given belief.
SYNOPSIS
function r=SimulateFrom(P,V,start,n)
DESCRIPTION
Simulates a POMDP from a given belief.
Simulates a POMDP from a given belief ('start') for 'n' steps and using
the given policy ('V') to decide the actions to execute.
The functions drawns a state at random from the 'start' distribution
and uses is as a hidden state in the simulation. This hidden state is
used to query the observation and reward models.
Returns the accumulated discounted reward fro the simulation.
This function is basically used to get statistics for the policy
derived after a planning session. As the elements in 'V' increase, the
simulation can get quite time demanding. Actually, this is the most
expensive step of the statistics collection that is more costly than
the planning itself.
See also GetPOMDPSolutionStatistics.CROSS-REFERENCE INFORMATION
This function calls:
- OptimalAction Returns the optimal action for a given belief.
- empty Checks if a policy is empty.
- rand Random state from a discrete belief.
- get Get for GBeliefs.
- rand Random state from a belief.
- get Get function for the GMixture object.
- rand Generates random points on a GMixture.
- get Gaussian object get function.
- rand Generates random ponts on a Gaussian.
- get Get function for CS_CO_CA_POMDPs.
- get Get function for CS_CO_DA_POMDPs.
- get Get function for CS_CO_POMDPs.
- get Get function for CS_DO_CA_POMDPs.
- get Get function for CS_DO_DA_POMDPs.
- get Get function for CS_POMDPs.
- get Get function for DS_CO_CA_POMDPs.
- get Get function for DS_CO_DA_POMDPs.
- get Get function for DS_DO_CA_POMDPs.
- get Get function for DS_DO_DA_POMDPs.
- SimulationStep Executes one step of a POMDP simulation.
- get Get functio for POMDPs.
- rand Random state from a continuous space.
- rand Random state from a discrete space.
- GetPOMDPSolutionStatistics Collects statistics with the outcome of an experiment.
SOURCE CODE
0001 function r=SimulateFrom(P,V,start,n) 0002 % Simulates a POMDP from a given belief. 0003 % 0004 % Simulates a POMDP from a given belief ('start') for 'n' steps and using 0005 % the given policy ('V') to decide the actions to execute. 0006 % 0007 % The functions drawns a state at random from the 'start' distribution 0008 % and uses is as a hidden state in the simulation. This hidden state is 0009 % used to query the observation and reward models. 0010 % 0011 % Returns the accumulated discounted reward fro the simulation. 0012 % 0013 % This function is basically used to get statistics for the policy 0014 % derived after a planning session. As the elements in 'V' increase, the 0015 % simulation can get quite time demanding. Actually, this is the most 0016 % expensive step of the statistics collection that is more costly than 0017 % the planning itself. 0018 % 0019 % See also GetPOMDPSolutionStatistics. 0020 0021 A=get(P,'ActionSpace'); 0022 noPolicy=empty(V); 0023 s=rand(start); 0024 b=start; 0025 r=0; 0026 gamma=1; 0027 for i=1:n 0028 if noPolicy 0029 a=rand(A); 0030 else 0031 a=OptimalAction(V,b); 0032 end 0033 [s b o r1]=SimulationStep(P,b,s,a); 0034 r=r+gamma*r1; 0035 gamma=gamma*P.gamma; 0036 end 0037