|
|
Post by darwin on Jun 25, 2014 15:42:01 GMT -8
I am looking for sample java code for automatically generating a propnet which can calculate the initial state, legal moves, next state etc. An example in any language would be helpful really, but particularly java because that is what I am using to build my player. THe ggb-base has a "SamplePropNetStateMachine.java" but large sections of the player are marked todo. From the coursera course O understand why a propnet could be highly efficient, especially if the size of the representation of the game can be dramatically reduced (i.e. to a vector of 1s and 0s), but I am fairly lost on how to actually auto generate the propnet.
|
|
rxe
Junior Member
Posts: 61
|
Post by rxe on Jul 3, 2014 13:37:08 GMT -8
It is hard to give a sample version, as propnet statemachines can vary in many ways. The code for galvanise is public : bitbucket.org/rxe/galvanise-dev. There are some sample propnets in there ggp-base/src/org/ggp/base/util/statemachine/implementation/glv/. Running org.ggp.base.util.statemachine.implementation.glv.Standalone <kif file> will run and compare times with respect to prover. The code wasn't especially written with other people in mind... I would also head along to alloyggp.blogspot.co.uk and sanchoggp.blogspot.co.uk for general advice. Many of the ideas of galvanise's propnet statemachine implementation can be credited to the Sancho crew. |
|
wat
New Member
Posts: 32 
|
Post by wat on Jul 7, 2014 10:27:43 GMT -8
OptimizingPropNetFactory.create does 99% of the work for you. But you still need to map between MachineState/Move <-> Proposition and perform propagation of values between propositions.
I built one a while ago, it was runnnig about 5x faster than ProverStateMachine.
|
|
