Big picture: combining off-line and on-line approaches maybe the best way to tackle large POMDPs.

Try planning:

• only where we are
• only where we can reach

Take into account three factors:

1. uncertainty in the value function
2. reachability from the current belief
3. actions that are likely optimal

It allows policy improvement on any base policy.

Setup

Discrete POMDPs:

• \(L\), lower bound
• \(U\), upper-bound
• \(b_0\): current belief

Two main phases: the algorithm

eansoetuhaosneu

Other Factoids

• charged
• flux

Chapters

• coulomb’s law
• superposition
• electric field
• Gauss’ Law
• electric potential
• current
• Ohm’s Law
• resistor
• kirchoff’s laws
• Capacitor
• Dynamic RC Circuits
• magnetism
• faraday’s law

Things to do

AP Phys C EM Things to Do

• Review all the names of units, and their SI conversions
• Review all eqns of time constants
• “Amperian Loop”