convergence.py

import math
import os.path
from os import path
 
class ReadStats:
 
    def __init__(self):
        self.generations = []
        self.times = []
        self.evals = []
        self.mins = []
        self.maxs = []
        self.avgs = []
 
    def readLogFile(self, fileName):
        try:
            logFile = open(fileName, 'r')
        except IOError:
            print("Couldn't open the file ", fileName)
            quit()
 
        generation = []
        eval = []
        time = []
        max = []
        min = []
        avg = []
        std = []
        inGeneration = False
        readMax = readMin = readAvg = readStd = False
        readEval = readTime = False
 
        for line in logFile:
            items = [word for word in line.split()]
            if(len(items) == 0):
                continue
            if(items[0] == "Generation:"):
                generation.append(int(items[1]))
                inGeneration = True
            if(not readEval and items[0] == "Evaluations:"):
                eval.append(int(items[1]))
                readEval = True
            if(not readTime and items[0] == "Elapsed"):
                time.append(int(items[2]))
                readTime = True
            if(not readMax and items[0] == "max:"):
                max.append(float(items[1]))
                readMax = True
            if(not readMin and items[0] == "min:"):
                min.append(float(items[1]))
                readMin = True
            if(not readAvg and items[0] == "avg:"):
                avg.append(float(items[1]))
                readAvg = True
            if(not readStd and items[0] == "stdev:"):
                std.append(float(items[1]))
                readStd = True
            if(inGeneration and readMax and readMin and readAvg and readStd):
                inGeneration = readMax = readMin = readAvg = readStd = False
                readEval = readTime = False
 
        logFile.close()
        self.generations.append(generation)
        self.times.append(time)
        self.evals.append(eval)
        self.mins.append(min)
        self.maxs.append(max)
        self.avgs.append(avg)
#       print("eval:", eval)
#       print("time:", time)
#       print("max:", max)
#       print("min:", min)
#       print(len(self.evals))
 
        
    def getPartialLog(self, percentage):
        stats = []
        for i in range(len(self.generations)):
            logNo = int((len(self.generations[i]) - 1) * percentage);
            stat = []
            stat.append(self.mins[i][logNo])
            stat.append(self.maxs[i][logNo]);
            stat.append(self.avgs[i][logNo]);
            stat.append(self.generations[i][logNo]);
            stat.append(self.evals[i][logNo]);
            stat.append(self.times[i][logNo]);
            stats.append(stat);
        return stats;
 
    def getMean(self, values):
        sum = float(0)
        for i in range(len(values)):
            sum += values[i]
        return sum / len(values)
 
 
    def getStdDev(self, values):
        if len(values) == 0:
            return 0
        numerator = float(0)
        average = self.getMean(values)
        for i in range(len(values)):
            numerator += ((values[i] - average) * (values[i] - average));
        denominator = (len(values) - 1)
        return math.sqrt(numerator / denominator)
 
 
    def getMin(self, values):
        min = values[0]
        for i in range(len(values)):
            if min > values[i]:
                min = values[i]
        return min
 
 
    def getMax(self, values):
        max = values[0]
        for i in range(len(values)):
            if max < values[i]:
                max = values[i]
        return max
 
 
    def getMedian(self, values):
        values.sort()
        if len(values) % 2 == 0:
            return (values[int(len(values) / 2) - 1] + values[int(len(values) / 2)]) / 2
        else:
            return values[int(len(values) / 2)]
 
 
def printStr(FloatNumber, Precision):
    return "%0.*f" % (Precision, FloatNumber)
 
 
 
 
# main
 
# default vrijednosti
logBase = "log_";
logExt = ".txt";
outFile = "convergence.txt";
logStartNum = 1;
logEndNum = 0;
nSamples = 50;
nDigits = 2;
 
line = input(" -- Log filename base (enter - \"" + logBase + "\") : ")
if line != "":
    logBase = line
 
line = input(" -- Starting log number (enter - " + str(logStartNum) + ") : ")
if line != "":
    logStartNum = int(line)
 
line = input(" -- End log number (enter - read all): ")
if line != "":
    logEndNum = int(line)
 
line = input(" -- Log filename num of digits (enter - 2) : ")
if line != "":
    nDigits = int(line)
 
line = input(" -- Extension (enter - \"" + logExt + "\") : ")
if line != "":
    logExt = line
 
line = input(" -- Output file (enter - \"" + outFile + "\") : ")
if line != "":
    outFile = line
 
line = input(" -- Number of samples (enter - " + str(nSamples) + ") : ")
if line != "":
    nSamples = int(line)
 
 
logNum = logStartNum
rs = ReadStats()
 
while True:
    fileName = ""
    fileName += logBase
    fileName += str(logNum).zfill(nDigits) 
    fileName += logExt
 
    print("Log filename: ", fileName)
    opened = False
    if path.exists(fileName): 
        print("Found!")
        opened = True
        rs.readLogFile(fileName)
    else:
        print("File not found!")
 
    logNum += 1
    cont = False
    if logEndNum >= logNum or opened == True:
        cont = True
    if not cont:
        break
 
#rs.readLogFile(fileName)
 
vAvgMin, vAvgMax, vMedMin, vMedMax, vMaxMin, vMinMax, vMinMin, vMaxMax, vAvgAvg = ([] for i in range(9))
vHits = []
vGens = []
vEvals = []
sample = 0
nSamples = 10
while sample < nSamples:
    percentage = 1. * sample / nSamples
    stats = rs.getPartialLog(percentage)
    nRuns = len(stats)
    avg_min = 0.
    avg_max = 0.
    mins = []
    maxs = []   # vektor minimuma i maksimuma u ovom trenutku
    avgs = []   # vektor srednjih vrijednosti u ovom trenutku
    for runId in range(nRuns):
        curr_min = stats[runId][0]
        curr_max = stats[runId][1]
        mins.append(stats[runId][0])
        maxs.append(stats[runId][1])
        avgs.append(stats[runId][2])
 
    vAvgMin.append(rs.getMean(mins))
    vAvgMax.append(rs.getMean(maxs))
    vMedMin.append(rs.getMedian(mins))
    vMedMax.append(rs.getMedian(maxs))
    vMinMin.append(rs.getMin(mins))
    vMaxMin.append(rs.getMax(mins))
    vMinMax.append(rs.getMin(maxs))
    vMaxMax.append(rs.getMax(maxs))
    vAvgAvg.append(rs.getMean(avgs))
    vHits.append(0)
    vGens.append(stats[0][3])
    vEvals.append(stats[0][4])
    sample += 1
 
 
file = open(outFile,"w")
file.write("recId\tgen\teval\tmin_min\tavg_min\tmed_min\tmax_min\tmin_max\tavg_max\tmed_max\tmax_max\thits\tavg_avg\n")
for sample in range(nSamples):
    output = str(sample) + "\t" + str(vGens[sample]) + "\t" + str(vEvals[sample]) + "\t"
    output += str(vMinMin[sample]) + "\t" + str(vAvgMin[sample]) + "\t" + str(vMedMin[sample]) + "\t" + str(vMaxMin[sample]) + "\t"
    output += str(vMinMax[sample]) + "\t" + str(vAvgMax[sample]) + "\t" + str(vMedMax[sample]) + "\t" + str(vMaxMax[sample]) + "\t"
    output += str(vHits[sample]) + "\t" + str(vAvgAvg[sample]) + "\n"
    file.write(output)
file.close()