Trees       Indices       Help   
6.01
Module sm :: Class SM
[frames] | no frames]

Class SM

Known Subclasses:

Generic superclass representing state machines. Don't instantiate this: make a subclass with definitions for the following methods:

optional:

See State Machines chapter in 6.01 Readings for detailed explanation.

Instance Methods
 
getStartState(self)
Handles the case that self.startState is a function.
 
getNextValues(self, state, inp)
Default version of this method.
 
done(self, state)
By default, machines don't terminate
 
isDone(self)
Should only be used by transduce.
 
start(self, traceTasks=[], verbose=False, compact=True, printInput=True)
Call before providing inp to a machine, or to reset it.
 
step(self, inp)
Execute one 'step' of the machine, by propagating inp through to get a result, then updating self.state.
 
transduce(self, inps, verbose=False, traceTasks=[], compact=True, printInput=True, check=False)
Start the machine fresh, and feed a sequence of values into the machine, collecting the sequence of outputs
 
run(self, n=10, verbose=False, traceTasks=[], compact=True, printInput=True, check=False)
For a machine that doesn't consume input (e.g., one made with feedback, for n steps or until it terminates.
 
transduceF(self, inpFn, n=10, verbose=False, traceTasks=[], compact=True, printInput=True)
Like transduce, but rather than getting inputs from a list of values, get them by calling a function with the input index as the argument.
 
guaranteeName(self)
Makes sure that this instance has a unique name that can be used for tracing.
 
printDebugInfo(self, depth, state, nextState, inp, out, debugParams)
Default method for printing out all of the debugging information for a primitive machine.
 
doTraceTasks(self, inp, state, out, debugParams)
Actually execute the trace tasks.
 
check(thesm, inps=None)
Run a rudimentary check on a state machine, using the list of inputs provided.
Class Variables
  startState = None
By default, startState is none
  legalInputs = []
By default, the space of legal inputs is not defined.
  name = None
Name used for tracing
Instance Variables
  state
Instance variable set by start, and updated by step; should not be managed by user
Method Details

getStartState(self)

 

Handles the case that self.startState is a function. Necessary for stochastic state machines. Ignore otherwise.

getNextValues(self, state, inp)

 

Default version of this method. If a subclass only defines getNextState, then we assume that the output of the machine is the same as its next state.

isDone(self)

 

Should only be used by transduce. Don't call this.

start(self, traceTasks=[], verbose=False, compact=True, printInput=True)

 

Call before providing inp to a machine, or to reset it. Sets self.state and arranges things for tracing and debugging.

Parameters:
  • traceTasks - list of trace tasks. See documentation for doTraceTasks for details
  • verbose - If True, print a description of each step of the machine
  • compact - If True, then if verbose = True, print a one-line description of the step; if False, print out the recursive substructure of the state update at each step
  • printInput - If True, then if verbose = True, print the whole input in each step, otherwise don't. Useful to set to False when the input is large and you don't want to see it all.

step(self, inp)

 

Execute one 'step' of the machine, by propagating inp through to get a result, then updating self.state. Error to call step if done is true.

Parameters:
  • inp - next input to the machine

transduce(self, inps, verbose=False, traceTasks=[], compact=True, printInput=True, check=False)

 

Start the machine fresh, and feed a sequence of values into the machine, collecting the sequence of outputs

For debugging, set the optional parameter check = True to (partially) check the representation invariance of the state machine before running it. See the documentation for the check method for more information about what is tested.

See documentation for the start method for description of the rest of the parameters.

Parameters:
  • inps - list of inputs appropriate for this state machine
Returns:
list of outputs

run(self, n=10, verbose=False, traceTasks=[], compact=True, printInput=True, check=False)

 

For a machine that doesn't consume input (e.g., one made with feedback, for n steps or until it terminates.

See documentation for the start method for description of the rest of the parameters.

Parameters:
  • n - number of steps to run
Returns:
list of outputs

doTraceTasks(self, inp, state, out, debugParams)

 

Actually execute the trace tasks. A trace task is a list consisting of three components:

  • name: is the name of the machine to be traced
  • mode: is one of 'input', 'output', or 'state'
  • fun: is a function

To do a trace task, we call the function fun on the specified attribute of the specified mahine. In particular, we execute it right now if its machine name equals the name of this machine.

check(thesm, inps=None)

 

Run a rudimentary check on a state machine, using the list of inputs provided. Makes sure that getNextValues is defined, and that it takes the proper number of input arguments (three: self, start, inp). Also print out the start state, and check that getNextValues provides a legal return value (list of 2 elements: (state,output)). And tries to check if getNextValues is changing either self.state or some other attribute of the state machine instance (it shouldn't: getNextValues should be a pure function).

Raises exception 'InvalidSM' if a problem is found.

Parameters:
  • thesm - the state machine instance to check
  • inps - list of inputs to test the state machine on (default None)
Returns:
none

   Trees       Indices       Help   
6.01
Generated by Epydoc 3.0.1 on Tue Apr 26 17:51:17 2011 http://epydoc.sourceforge.net