images/sm_SimPy_Logo.png

SimulationRTManual

Authors:
Version:1.0
Date:2003-December-18
SimPy version:1.4
Web-site:http://simpy.sourceforge.net/
Python-Version:2.2, 2.3
Created:2003-December-18

A Manual for SimulationRT

This manual describes SimulationRT, a SimPy module which supports synchronizing the execution of simulation models with real (wallclock) time.

Acknowledgement

SimulationRT is based on an idea by Geoff Jarrad of CSIRO (Australia). He contributed a lot to its development and testing on Windows and Unix.

Introduction

SimulationRT allows synchronizing simulation time and real (wallclock) time. This capability can be used to implement e.g. interactive game applications or to demonstrate a model's execution in real time.

Over and above the capabilities provided by Simulation, SimulationStep caters for executing SimPy models at user-controlled speed. This is done by setting a parameter in the call to simulateSimulationStep is a derivative of the Simulation module. It is identical to Simulation, except for the *simulate function which takes an additional parameter controlling execution speed.

Here is an example:

## RealTimeFireworks.py
from __future__ import generators
from SimPy.SimulationRT  import *
from random import *
import time
class Launcher(Process):
    def launch(self):
        while True:
            print "Launch at %.1f; wallclock: %s"%(now(),time.clock()-startTime)
            yield hold,self,uniform(1,maxFlightTime)
            print "Boom!!! Aaaah!! at %.1f; wallclock: %s"%(now(),time.clock()-startTime)
def model():
    initialize()
    for i in range(nrLaunchers):
        lau=Launcher()
        activate(lau,lau.launch())
    simulate(real_time=True,rel_speed=1,until=20) ##unit sim time = 1 sec clock
nrLaunchers=2
maxFlightTime=5.0 
startTime=time.clock()
model()

rels_speed=1 sets the synchronization so that 1 simulation time unit is executed approximately in 1 second wallclock time. Run under Python 2.2.2 on a Windows XP-box (1.7 GHz), this output resulted over about 18 seconds of wallclock time:

Launch at 0.0; wallclock: 0.000195555580376
Launch at 0.0; wallclock: 0.00190107960634
Boom!!! Aaaah!! at 1.8; wallclock: 1.78082661344
Launch at 1.8; wallclock: 1.78274501368
Boom!!! Aaaah!! at 2.8; wallclock: 2.84245930698
Launch at 2.8; wallclock: 2.84435982785
Boom!!! Aaaah!! at 4.1; wallclock: 4.08443978215
Launch at 4.1; wallclock: 4.09004328762
Boom!!! Aaaah!! at 5.2; wallclock: 5.14561822801
Launch at 5.2; wallclock: 5.14878203794
Boom!!! Aaaah!! at 7.0; wallclock: 6.99845622838
Launch at 7.0; wallclock: 7.00175357483
Boom!!! Aaaah!! at 7.4; wallclock: 7.39919794276
Launch at 7.4; wallclock: 7.40245282571
Boom!!! Aaaah!! at 9.7; wallclock: 9.69250728794
Launch at 9.7; wallclock: 9.69912935862
Boom!!! Aaaah!! at 10.6; wallclock: 10.5938587167
Launch at 10.6; wallclock: 10.6006140445
Boom!!! Aaaah!! at 13.8; wallclock: 13.8082362423
Launch at 13.8; wallclock: 13.8134877477
Boom!!! Aaaah!! at 14.1; wallclock: 14.1385670525
Launch at 14.1; wallclock: 14.1438146468
Boom!!! Aaaah!! at 16.4; wallclock: 16.411963811
Launch at 16.4; wallclock: 16.4172373863
Boom!!! Aaaah!! at 17.1; wallclock: 17.1429980626
Launch at 17.1; wallclock: 17.1482308506
Boom!!! Aaaah!! at 18.1; wallclock: 18.0742063586
Launch at 18.1; wallclock: 18.0794469688

Clearly, the wallclock time does not deviate significantly from the simulation time.

Limitations

This module works much better under Windows than under Unix or Linux, i.e., it gives much closer synchronization. Unfortunately, the handling of time in Python is not platform-independent at all. Here is a quote from the documentation of the time module:

"clock()
On Unix, return the current processor time as a floating point number expressed in seconds. 
The precision, and in fact the very definition of the meaning of ``processor time'' , depends 
on that of the C function of the same name, but in any case, this is the function to use for 
benchmarking Python or timing algorithms.

On Windows, this function returns wall-clock seconds elapsed since the first call to this 
function, as a floating point number, based on the Win32 function QueryPerformanceCounter(). 
The resolution is typically better than one microsecond. 
"

The SimulationRT API

Structure

Basically, SimulationStep has the same API as Simulation, but with a change in the definition of simulate.

simulate

Executes the simulation model.

Call:

simulate(<optional parameters>)
Mandatory parameters:
None.
Optional parameters:
  • until=0 : the maximum simulation (end) time (positive floating point number; default: 0)
  • real_time=False : flag to switch real time synchronization on or off (boolean; default: False, meaning no synchronization)
  • rel_speed=1 : ratio simulation time over wallclock time; example: rel_speed=200 executes 200 units of simulation time in about one second (positive floating point number; default: 1, i.e. 1 sec of simulation time is executed in about 1 sec of wallclock time)
Return value:
Simulation status at exit.

$Revision: 1.0 $ $Date: 2004-05-15 11:09:27+02 $ kgm