Requires

1

• Requires
• Cost/reward matrix
• TPM

• Requires
• Cost/reward matrix
• TPM
• TTM
• If TTM is exponentially distd
• then SMDP CTMDP

If time is Exp distn then use e(-gt), else find
discount factors for each transition time
If you cannot obtain TPM and TTM then
approximate dynamic programming is needed (a.k.a
neuro-dynamic Programming, reinforcement
learning, simulation-based Optimization) OR 774
YES