dispyParameterVariationExample.py

You can view and download this file on Github: dispyParameterVariationExample.py

  1#+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
  2# This is an EXUDYN example
  3#
  4# Details:  This example performs a parameter variation of a simple
  5#           mass-spring-damper system; varying mass, spring, using the python module dispy
  6#           In this example, we return the computed sensor values to the ParameterVariation funtion
  7#           and visualize a set of results
  8#           NOTE: the real speedup using multiprocessing or cluster is about 3 on a 4-core machine, but not the value given by dispy!
  9#
 10# Author:   Stefan Holzinger, Johannes Gerstmayr
 11# Date:     2022-04-11
 12#
 13# Copyright:This file is part of Exudyn. Exudyn is free software. You can redistribute it and/or modify it under the terms of the Exudyn license. See 'LICENSE.txt' for more details.
 14#
 15#+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
 16
 17import numpy as np
 18
 19#this is the function that is called via dispy; parameterSet contains the dictionary of current parameters and
 20#optional 'functionData' with additional data passed to this function (e.g. numpy arrays)
 21#parameters are written into the 'P' structure, such that they can be overwritten easily
 22def ParameterFunction(parameterSet):
 23    # modules such as exudyn etc. need to be imported inside parameter function
 24    import socket
 25    import numpy as np
 26    import exudyn as exu
 27    import exudyn.itemInterface as eii
 28
 29    SC = exu.SystemContainer()
 30    mbs = SC.AddSystem()
 31
 32    #++++++++++++++++++++++++++++++++++++++++++++++
 33    #++++++++++++++++++++++++++++++++++++++++++++++
 34    #store default parameters in structure (all these parameters can be varied!)
 35    class P: pass #create emtpy structure for parameters; simplifies way to update parameters
 36
 37    #default values
 38    P.mass = 1.6          #mass in kg
 39    P.spring = 4000       #stiffness of spring-damper in N/m
 40    P.damper = 8          #damping constant in N/(m/s)
 41    P.u0=-0.08            #initial displacement
 42    P.v0=1                #initial velocity
 43    P.f =80               #force applied to mass
 44    P.L=0.5               #spring length (for drawing)
 45    P.computationIndex = 'Ref' #if computationIndex not provided
 46
 47    # #now update parameters with parameterSet (will work with any parameters in structure P)
 48    for key,value in parameterSet.items():
 49        setattr(P,key,value)
 50
 51     #++++++++++++++++++++++++++++++++++++++++++++++
 52     #++++++++++++++++++++++++++++++++++++++++++++++
 53     #START HERE: create parameterized model, using structure P, which is updated in every computation
 54
 55    # +++++++++++++++++++++++++++++++++++++++
 56    # create mechanical system
 57
 58    #mass-spring-damper system
 59    P.L=0.5               #spring length (for drawing)
 60
 61    x0=P.f/P.spring         #static displacement
 62
 63    #node for 3D mass point:
 64    n1=mbs.AddNode(eii.Point(referenceCoordinates = [P.L,0,0],
 65                                initialCoordinates = [P.u0,0,0],
 66                                initialVelocities= [P.v0,0,0]))
 67
 68
 69    #ground node
 70    nGround=mbs.AddNode(eii.NodePointGround(referenceCoordinates = [0,0,0]))
 71
 72    #add mass point (this is a 3D object with 3 coordinates):
 73    massPoint = mbs.AddObject(eii.MassPoint(physicsMass = P.mass, nodeNumber = n1))
 74
 75    #marker for ground (=fixed):
 76    groundMarker=mbs.AddMarker(eii.MarkerNodeCoordinate(nodeNumber= nGround, coordinate = 0))
 77    #marker for springDamper for first (x-)coordinate:
 78    nodeMarker  =mbs.AddMarker(eii.MarkerNodeCoordinate(nodeNumber= n1, coordinate = 0))
 79
 80    #spring-damper between two marker coordinates
 81    nC = mbs.AddObject(eii.CoordinateSpringDamper(markerNumbers = [groundMarker, nodeMarker],
 82                                                  stiffness = P.spring, damping = P.damper))
 83
 84    #add load:
 85    mbs.AddLoad(eii.LoadCoordinate(markerNumber = nodeMarker, load = P.f))
 86
 87    #add sensor:
 88    fileName = 'solution/paramVarDisplacement'+str(P.computationIndex)+'.txt'
 89    mbs.AddSensor(eii.SensorObject(objectNumber=nC, fileName=fileName,
 90                                outputVariableType=exu.OutputVariableType.Force))
 91
 92
 93    #print(mbs)
 94    mbs.Assemble()
 95
 96    steps = 1000000  #number of steps to show solution; use many steps to see speedup!
 97    tEnd = 1     #end time of simulation
 98
 99    simulationSettings = exu.SimulationSettings()
100    simulationSettings.solutionSettings.writeSolutionToFile = False
101    simulationSettings.solutionSettings.sensorsWritePeriod = 5e-3  #output interval of sensors
102    simulationSettings.timeIntegration.numberOfSteps = steps
103    simulationSettings.timeIntegration.endTime = tEnd
104
105    simulationSettings.timeIntegration.generalizedAlpha.spectralRadius = 1 #no damping
106
107    #start solver:
108    mbs.SolveDynamic(simulationSettings)
109
110    # +++++++++++++++++++++++++++++++++++++++++++++++++++++
111    # evaluate difference between reference and optimized solution
112
113    # get reference solution
114    if 'functionData' in parameterSet: # load reference solution from function data
115        functionData = parameterSet['functionData']
116        refSol = functionData['refSol']
117    else: # load from file
118        refSol = np.loadtxt(fileName, comments='#', delimiter=',')
119
120    # computation solution from current simulation
121    sol = np.loadtxt(fileName, comments='#', delimiter=',')
122
123    # compute error
124    diff = refSol[:,1] - sol[:,1]
125    errorNorm = np.sqrt(np.dot(diff,diff))/steps*tEnd
126
127
128    return [errorNorm, sol]  #we can also return any pickle-able data in return value, which is then stored in values of ParameterVariation
129    #return errorNorm #for debug: (socket.gethostname(), errorNorm)
130
131
132
133#%%
134if __name__ == '__main__':
135    from exudyn.processing import ParameterVariation
136    import time
137
138
139    #++++++++++++++++++++++++++++++++++++++++++++++++
140    # generate reference solution
141    refval = ParameterFunction({}) # compute reference solution
142    #print("refval =", refval)
143
144    # add reference solution to function data
145    referenceSolution = np.loadtxt('solution/paramVarDisplacementRef.txt', comments='#', delimiter=',')
146    functionData = {'refSol': referenceSolution}
147
148
149
150    #++++++++++++++++++++++++++++++++++++++++++++++++
151    # process inputs to form array with each parameter values etc. see ParameterVariation()
152    useCluster = True # if false, multiprocessing is used
153    if useCluster: # specify nodes by providing host names or IPv4-addresses
154        if False: #put your dispy hosts here
155            activeHosts = ['111.112.113.114'] # enter list of IPv4 addresses or host names here
156        else: #just use local host (for tests only!);
157            import socket
158            hostName = socket.gethostname()
159            print('your hostname is: ',hostName)
160            activeHosts = [hostName]
161        clusterHostNames = activeHosts
162        useDispyWebMonitor = 'useDispyWebMonitor' # if given to ParameterVariation(), a web browser is started which can be used to manage the cluster computation
163
164    else:
165        clusterHostNames = []
166        useDispyWebMonitor = ''
167
168
169
170    #++++++++++++++++++++++++++++++++++++++++++++++++
171    # perform parameter variation
172    n = 2 # n*n = number of variations
173    start_time = time.time()
174    [pDict, values] = ParameterVariation(parameterFunction = ParameterFunction,
175                                         parameters = {'mass':(1,2,n),
176                                                       'spring':(2000,8000,n),
177                                                       },
178                                         parameterFunctionData = functionData,
179                                         debugMode = False,
180                                         addComputationIndex = True,
181                                         useMultiProcessing = True,
182                                         showProgress = True,
183                                         clusterHostNames = clusterHostNames, #not that there is a significant overhead, thus cluster only makes sense for computations that take longer than 1 second
184                                         #useDispyWebMonitor = useDispyWebMonitor, #slows down a little and waits 5 seconds in the end to finish
185                                         )
186    print("--- %s seconds ---" % (time.time() - start_time))
187
188
189
190
191#%% post processing of all computed variations
192    if True:
193
194        #extract solution from return values
195        valuesPlot = []
196        for item in values:
197            valuesPlot +=[item[0]]
198
199        #plot first 8 results:
200        for i in range(len(values)):
201            from exudyn.plot import PlotSensor
202            PlotSensor(0, sensorNumbers=values[i][1], labels='result'+str(i), newFigure=(i==0),
203                       colorCodeOffset=i, fontSize=12, closeAll=(i==0))