def testStochasticHillClimbingSearch(self):
from StochasticAlgorithms.StochasticHillClimbing import search
# Problem Configuration
numBits = 64
# Algorithm Configuration
maxIterations = 1000
# Execute the SHC algorithm
result = search(numBits, maxIterations)
print "Stochastic Hill Climbing Search Results : "
print "*" * 20
print "SHC Iteration "
print "*" * 20
print result["iteration"]
print "*" * 20
print "*" * 20
print "Initial One Max Count"
print "*" * 20
print result["initialCost"]
print "*" * 20
print "Final One Max Count"
print "*" * 20
print result["cost"]
print "*" * 20
print "*" * 20
print "*" * 20
print "Search FormattedOutput : (Final- Initial)/Iteration"
print "*" * 20
efficacy = round(
float(result["cost"] - result["initialCost"]) /
float(result["iteration"]), 2)
print efficacy
def testTabuSearch(self):
from StochasticAlgorithms.TabuSearch import search
# Problem Configuration
# Use Berlin52 instance of TSPLIB
# Algorithm Configuration
maxIterations = 100
maxTabuCount = 15
maxCandidates = 50
# Execute the algorithm
result = search(self.TSPLIB, maxIterations, maxTabuCount,
maxCandidates)
tspResult = TSPResult(7542, "Tabu Search Results")
print tspResult.FormattedOutput(result)
def testIteratedLocalSearch(self):
from StochasticAlgorithms.IteratedLocalSearch import search
# Problem Configuration
# Use Berlin52 instance of TSPLIB
# Algorithm Configuration
maxIterations = 100
maxNoImprove = 50
searchHistoryToKeep = 1 # since we don't plan to use it now
# Execute the Algorithm
result = search(self.TSPLIB, maxIterations, maxNoImprove,
searchHistoryToKeep)
tspResult = TSPResult(7542, "Iterated Local Search Results")
print tspResult.FormattedOutput(result)
def testReactiveTabuSearch(self):
from StochasticAlgorithms.ReactiveTabuSearch import search
# Problem Configuration
# Use Berlin52 instance of TSPLIB
# Algorithm Configuration
maxIterations = 100
maxCandidates = 50
increase = 1.3
decrease = 0.9
#Execute the algorithm
result = search(self.TSPLIB, maxIterations, increase, decrease,
maxCandidates)
tspResult = TSPResult(7542, "Reactive Tabu Search")
print tspResult.FormattedOutput(result)
def testRandomSearch(self):
from StochasticAlgorithms.RandomSearch import search
# Problem Configuration
searchVector = [-5, 5]
size = 2
# Algorithm Configuration
iterations = 10000
# Execute the random search algorithm
# Outputs a tuple containing the best cost and best input values
result = search(searchVector, iterations, size)
self.assertTrue(
basinFunction(result["vector"]) == result["cost"],
"Failed to get correct best basin value.")
basin = BasinResult("Random Search")
print basin.FormattedOutput(result)
def testGuidedLocalSearch(self):
from StochasticAlgorithms.GuidedLocalSearch import search
# Problem Configuration
# Use Berlin52 instance of TSPLIB
# Algorithm Configuration
maxIterations = 150
maxNoImprove = 20
localSearchOptima = 12000.0
alpha = 0.3
scalingFactor = alpha * (localSearchOptima / float(len(self.TSPLIB)))
# Execute the algorithm
result = search(self.TSPLIB, maxIterations, maxNoImprove,
scalingFactor)
tspResult = TSPResult(7542, "Guided Local Search Results")
print tspResult.FormattedOutput(result)
def testGreedyRandomizedAdaptiveSearch(self):
from StochasticAlgorithms.GreedyRandomizedAdaptiveSearch import search
# Problem Configuration
# Use Berlin52 instance of TSPLIB
# Algorithm Configuration
maxNoImprove = 50
maxIterations = 50
greedinessFactor = 0.3 # should be in the range [0,1]. 0 is more greedy and 1 is more generalized
# Execute the algorithm
result = search(self.TSPLIB, maxIterations, maxNoImprove,
greedinessFactor)
tspResult = TSPResult(7542,
"Greedy Randomized Adaptive Search Results")
print tspResult.FormattedOutput(result)
def testVariableNeighborhoodSearch(self):
from StochasticAlgorithms.VariableNeighborhoodSearch import search
# Problem Configuration
# Use Berlin52 instance of TSPLIB
# Algorithm Configuration
maxNoImprove = 50
maxNoImproveLocal = 70
neighborhoods = range(
1, 21) # since we want 20 runs for neighborhood starting with 1
# Execute the algorithm
result = search(self.TSPLIB, neighborhoods, maxNoImprove,
maxNoImproveLocal)
tspResult = TSPResult(7542, "Variable Neighborhood Search Results")
print tspResult.FormattedOutput(result)
def testAdaptiveRandomSearch(self):
from StochasticAlgorithms.AdaptiveRandomSearch import search
# Problem Configuration
searchVector = [-5, 5]
size = 2
# Algorithm Configuration
maxIterations = 10000
initFactor = 0.05
lFactor = 3.0
sFactor = 1.3
iterFactor = 10
maxNoChange = 25
# Execute the adaptive random search algorithm
# Outputs a tuple containing the best cost and best input values
result = search(maxIterations, size, searchVector, initFactor, sFactor,
lFactor, iterFactor, maxNoChange)
self.assertTrue(
basinFunction(result["vector"]) == result["cost"],
"Failed to get correct best basin value.")
basin = BasinResult("Adaptive Random Search")
print basin.FormattedOutput(result)
def testScatterSearch(self):
from StochasticAlgorithms.ScatterSearch import search
# Problem Configuration
searchVector = [-5, 5]
problemSize = 2
# Algorithm Configuration
maxIterations = 100
stepSize = (searchVector[1] -
searchVector[0]) * 0.05 # 5 percent of search space
maxNoImprove = 30
refSetSize = 10
diverseSetSize = 20
eliteCount = 5
# execute the algorithm
result = search(searchVector, problemSize, refSetSize, diverseSetSize,
maxIterations, maxNoImprove, eliteCount, stepSize)
self.assertTrue(
basinFunction(result["vector"]) == result["cost"],
"Failed to get correct best basin value.")
basin = BasinResult("Scatter Search")
print basin.FormattedOutput(result)