def getRealAction_real(self, state, actVars=None):
Program.execute(state, self.registers, self.program.instructions[:, 0],
self.program.instructions[:, 1],
self.program.instructions[:, 2],
self.program.instructions[:, 3])
return self.registers[:self.actionLength]
def bid(self, state, memMatrix):
Program.execute(state, self.registers, self.program.instructions[:, 0],
self.program.instructions[:, 1],
self.program.instructions[:, 2],
self.program.instructions[:, 3], memMatrix,
memMatrix.shape[0], memMatrix.shape[1])
return self.registers[0]
def getRealAction_real_mem(self, state, actVars=None):
Program.execute(state, self.registers, self.program.instructions[:, 0],
self.program.instructions[:, 1],
self.program.instructions[:, 2],
self.program.instructions[:, 3], actVars["memMatrix"],
actVars["memMatrix"].shape[0],
actVars["memMatrix"].shape[1],
Program.memWriteProbFunc)
return self.registers[:self.actionLength]
def bid_def(self, state, actVars=None):
# exit early if we already got bidded this frame
if self.frameNum == actVars["frameNum"]:
return self.registers[0]
self.frameNum = actVars["frameNum"]
Program.execute(state, self.registers,
self.program.instructions[:,0], self.program.instructions[:,1],
self.program.instructions[:,2], self.program.instructions[:,3])
return self.registers[0]
def bid_mem(self, state, actVars=None):
# exit early if we already got bidded this frame
if self.frameNum == actVars["frameNum"]:
return self.registers[0]
self.frameNum = actVars["frameNum"]
Program.execute(state, self.registers,
self.program.instructions[:,0], self.program.instructions[:,1],
self.program.instructions[:,2], self.program.instructions[:,3],
actVars["memMatrix"], actVars["memMatrix"].shape[0], actVars["memMatrix"].shape[1],
Program.memWriteProbFunc)
return self.registers[0]
def getLearnersInsOuts(self, learners, clearStates=True):
inputs = []
outputs = []
for lrnr in learners:
lrnrInputs = []
lrnrOutputs = []
for state in lrnr.states:
regs = np.zeros(len(lrnr.registers))
Program.execute(state, regs, lrnr.program.modes,
lrnr.program.operations,
lrnr.program.destinations,
lrnr.program.sources)
lrnrInputs.append(state)
lrnrOutputs.append(regs[0])
if clearStates: # free up some space
lrnr.states = []
inputs.append(lrnrInputs)
outputs.append(lrnrOutputs)
return inputs, outputs
def bid(self, state):
Program.execute(state, self.registers, self.program.modes,
self.program.operations, self.program.destinations,
self.program.sources)
return self.registers[0]