public static double AverageWidth = 0.0; // Average Width each iteration
public static void InitializeKmeans(double[][] PointPositionINPUT, ParallelOptions _parallelOptionsINPUT, string FileName, int ClusterPosition, int FirstClusterValue, int StartingPosition,
int Ncent_GlobalINPUT, int MaxNcent_GlobalINPUT, int Ncent_Global_ParallelINPUT, int ParameterVectorDimensionINPUT, double CenterChangeINPUT, int IterationCutINPUT)
{
Kmeans.PointPosition = PointPositionINPUT;
Kmeans._parallelOptions = _parallelOptionsINPUT;
Kmeans.Ncent_Global = Ncent_GlobalINPUT;
Kmeans.MaxNcent_Global = MaxNcent_GlobalINPUT;
Kmeans.Ncent_Global_Parallel = Ncent_Global_ParallelINPUT;
Kmeans.KmeansCenterChangeCut = CenterChangeINPUT;
Kmeans.KmeansIterationCut = IterationCutINPUT;
Kmeans.ParameterVectorDimension = ParameterVectorDimensionINPUT;
Kmeans.SetParallelCenterDecomposition();
Kmeans.InitialPointAssignment = new int[DAVectorUtility.PointCount_Process];
if (FileName.Length == 0)
{
return;
}
// Read Initial Assignments
DAVectorUtility.SALSAPrint(0, "Kmeans Read File " + FileName + " Points " + DAVectorUtility.PointCount_Global.ToString() + " Starting at position "
+ StartingPosition.ToString() + " Dimension " + Kmeans.ParameterVectorDimension.ToString() + " Cluster Position " + ClusterPosition.ToString() + " Initial value " + FirstClusterValue.ToString());
Kmeans.ReadDataFromFile(FileName, ClusterPosition, FirstClusterValue, StartingPosition);
} // End InitializeKmeans
} // End SetupKmeans
public static void RunKmeans(double[][] ClusterCenterINPUT, int[] ClusterSizeINPUT, double[] ClusteRadiusINPUT,
out int Ncent_GlobalFINAL, out double AverageWidthFINAL)
{
ArrayList KeepPCfractions = new ArrayList(200);
ArrayList KeepCCfractions = new ArrayList(200);
// Inherit Solution arrays
Kmeans.ClusterCenter = ClusterCenterINPUT;
Kmeans.ClusterSize = ClusterSizeINPUT;
Kmeans.ClusterRadius = ClusteRadiusINPUT;
Kmeans.ClusterWidth = new double[Kmeans.MaxNcent_Global];
// Set up TriangleInequality
KmeansTriangleInequality.SetExternalFunctions(GetClusterRadius, GetClusterCenters, FindClusterCenters);
KmeansTriangleInequality.InitializeTriangleInequality(Kmeans.PointPosition, Kmeans._parallelOptions, Kmeans.ClusterCenter,
Kmeans.Ncent_Global, Kmeans.MaxNcent_Global, Kmeans.Ncent_Global_Parallel, Kmeans.ParameterVectorDimension);
DAVectorUtility.SALSAPrint(0, "Start Kmeans ****** Number of Centers " + Kmeans.Ncent_Global.ToString() + " Max Number of Centers " + Kmeans.MaxNcent_Global.ToString()
+ " Center Limit for Parallelism " + Kmeans.Ncent_Global_Parallel.ToString() + " Vector Dimension " + Kmeans.ParameterVectorDimension.ToString());
Kmeans.FindClusterCenters(true, Kmeans.InitialPointAssignment, null, null);
Kmeans.CountKmeansIterations = 0;
bool StartStop = false;
int CountStops = 0;
while (Kmeans.CountKmeansIterations < Kmeans.KmeansIterationCut)
{
double save1 = KmeansTriangleInequality.NumberFullDistancesCalculatedCC;
double save2 = KmeansTriangleInequality.NumberFullDistancesCalculatedPC;
KmeansTriangleInequality.NextIteration();
++Kmeans.CountKmeansIterations;
bool WillStop = false;
if (!StartStop)
{
if (Kmeans.AverageCenterChange < Kmeans.AverageRadius * Kmeans.KmeansCenterChangeCut)
{
StartStop = true;
}
}
else
{
++CountStops;
if (CountStops > 10)
{
WillStop = true;
}
}
double tmp1 = (KmeansTriangleInequality.NumberFullDistancesCalculatedCC - save1) / (double)Kmeans.MaxNcent_Global;
double tmp2 = (KmeansTriangleInequality.NumberFullDistancesCalculatedPC - save2) / ((double)Kmeans.MaxNcent_Global * (double)DAVectorUtility.PointCount_Global);
double tmp3 = KmeansTriangleInequality.NumberFullDistancesCalculatedPC / ((double)Kmeans.MaxNcent_Global * (double)(DAVectorUtility.PointCount_Global * Kmeans.CountKmeansIterations));
double tmp4 = (KmeansTriangleInequality.NumberFullDistancesCalculatedPC + KmeansTriangleInequality.NumberFullDistancesCalculatedCC) / ((double)Kmeans.MaxNcent_Global * (double)(DAVectorUtility.PointCount_Global * Kmeans.CountKmeansIterations));
DAVectorUtility.SALSAPrint(0, "Iteration " + Kmeans.CountKmeansIterations.ToString() + " Average Center Change " + Kmeans.AverageCenterChange.ToString("E4")
+ " Average Radius " + Kmeans.AverageRadius.ToString("E4") + " Average Width " + Kmeans.AverageWidth.ToString("E4")
+ " CC calcs per C " + tmp1.ToString("F4") + " PC calcs per P&C " + tmp2.ToString("F6")
+ " Cumul PC / Max " + tmp3.ToString("F6") + " Cumul PC+CC / PC Max " + tmp4.ToString("F6"));
KeepPCfractions.Add(tmp2);
KeepCCfractions.Add(tmp1 / DAVectorUtility.PointCount_Global);
if (((Kmeans.CountKmeansIterations % 10) == 1) || WillStop)
{
string message = " Sizes";
for (int CenterIndex = 0; CenterIndex < Kmeans.Ncent_Global; CenterIndex++)
{
message += " " + Kmeans.ClusterSize[CenterIndex].ToString();
}
DAVectorUtility.SALSAPrint(0, message);
}
if (WillStop)
{
break;
}
}
DAVectorUtility.SALSAPrint(0, "End Kmeans Iterations " + Kmeans.CountKmeansIterations.ToString() + " Iteration Cut " + Kmeans.KmeansIterationCut.ToString() +
" Average Center Change " + Kmeans.AverageCenterChange.ToString("E4") + " Average Radius " + Kmeans.AverageRadius.ToString("E4") +
" Average Width " + Kmeans.AverageWidth.ToString("E4") + " Fractional Cut " + Kmeans.KmeansCenterChangeCut.ToString("F4"));
KmeansTriangleInequality.PrintDiagnostics();
string messagePC = "\nPC Calcs per Point iteration";
string messageCC = "\nCC Calcs per Point iteration";
int numPC = KeepPCfractions.Count;
for (int linecount = 0; linecount < numPC; linecount++)
{
messagePC += " " + ((double)KeepPCfractions[linecount]).ToString("F4") + ",";
messageCC += " " + ((double)KeepCCfractions[linecount]).ToString("F4") + ",";
}
DAVectorUtility.SALSAPrint(0, messagePC);
DAVectorUtility.SALSAPrint(0, messageCC);
Ncent_GlobalFINAL = Kmeans.Ncent_Global;
AverageWidthFINAL = Kmeans.AverageWidth;
// Print Histograms
if (KmeansTriangleInequality.UseTriangleInequality != 0)
{
KmeansTriangleInequality.PlotPointHistograms(Math.Sqrt(AverageWidthFINAL));
KmeansTriangleInequality.PlotCenterHistograms(Math.Sqrt(AverageWidthFINAL));
}
return;
} // End RunKmeans()
} // End FindClusterCenters(int[] NearestCentertoPoint, double[][] LastClusterCenter)
public static void ReadDataFromFile(string fname, int ClusterPosition, int FirstClustervalue, int StartPointPosition)
{
char[] _sep = new[] { ' ', ',', '\t' };
int FirstPointPosition = 0;
int TotalNumberPointstoRead = 0;
FirstPointPosition = DAVectorUtility.PointStart_Process;
TotalNumberPointstoRead = DAVectorUtility.PointCount_Process;
Random RandomObject = new Random(10101010 + DAVectorUtility.MPI_Rank);
if (ClusterPosition < 0)
{
DAVectorUtility.SALSAPrint(0, "Random Start 10101010 plus rank ******************* Option " + ClusterPosition.ToString());
}
int MinSplitSize = ClusterPosition + 1;
if (StartPointPosition >= 0)
{
MinSplitSize = Math.Max(MinSplitSize, StartPointPosition + Kmeans.ParameterVectorDimension);
}
else
{
Exception e = DAVectorUtility.SALSAError("Illegal Start Position on Points file " + fname + " Rank " + DAVectorUtility.MPI_Rank.ToString()
+ " POsition " + StartPointPosition.ToString() + " Number to Read " + TotalNumberPointstoRead.ToString());
throw (e);
}
bool success = false;
string line = " Unset";
int CountLinesinFile = 0;
try
{
StreamReader sr = null;
if (!string.IsNullOrEmpty(fname))
{
Stream stream = File.Open(fname, FileMode.Open, FileAccess.Read, FileShare.Read);
sr = new StreamReader(stream);
}
if (sr != null)
{
while (!sr.EndOfStream)
{
line = sr.ReadLine();
if (!string.IsNullOrEmpty(line))
{
string[] splits = line.Trim().Split(_sep, StringSplitOptions.RemoveEmptyEntries);
if (splits.Length < MinSplitSize)
{
DAVectorUtility.SALSAPrint(0, "Count " + CountLinesinFile.ToString() + " Illegal data length on Point file " + splits.Length.ToString()
+ " " + MinSplitSize.ToString() + " " + line);
continue;
} // Skip header lines
double junk;
if (!Double.TryParse(splits[StartPointPosition], out junk))
{
continue; // Skip header lines
}
if (CountLinesinFile < FirstPointPosition)
{
CountLinesinFile += 1;
continue;
}
int ActualPointPosition = CountLinesinFile - FirstPointPosition;
int label = 0;
Kmeans.PointPosition[ActualPointPosition][0] = double.Parse(splits[StartPointPosition]);
Kmeans.PointPosition[ActualPointPosition][1] = double.Parse(splits[StartPointPosition + 1]);
if (Kmeans.ParameterVectorDimension > 2)
{
for (int VectorIndex = 2; VectorIndex < Kmeans.ParameterVectorDimension; VectorIndex++)
{
Kmeans.PointPosition[ActualPointPosition][VectorIndex] = double.Parse(splits[VectorIndex + StartPointPosition]);
}
}
if (ClusterPosition >= 0)
{
if (!Int32.TryParse(splits[ClusterPosition], out label))
{
label = FirstClustervalue;
}
Kmeans.InitialPointAssignment[ActualPointPosition] = label - FirstClustervalue;
}
else
{
Kmeans.InitialPointAssignment[ActualPointPosition] = RandomObject.Next(Program.InitialNcent);
if (ClusterPosition == -2)
{ // Force each cluster to have one point
if (CountLinesinFile < Program.InitialNcent)
{
Kmeans.InitialPointAssignment[ActualPointPosition] = CountLinesinFile;
}
}
if (ClusterPosition == -3)
{
int divisor = Program.NumberDataPoints / Program.InitialNcent;
if (CountLinesinFile % divisor == 0)
{
Kmeans.InitialPointAssignment[ActualPointPosition] = CountLinesinFile / divisor;
}
}
if (ClusterPosition == -4)
{
int divisor = Program.NumberDataPoints / Program.InitialNcent;
Kmeans.InitialPointAssignment[ActualPointPosition] = CountLinesinFile / divisor;
}
}
++ActualPointPosition;
++CountLinesinFile;
if (CountLinesinFile >= (FirstPointPosition + TotalNumberPointstoRead))
{
break;
}
}
}
if (CountLinesinFile != (FirstPointPosition + TotalNumberPointstoRead))
{
Exception e = DAVectorUtility.SALSAError("Illegal count on Points file " + fname + " Rank " + DAVectorUtility.MPI_Rank.ToString()
+ " Lines in File " + CountLinesinFile.ToString() + " Number to Read " + TotalNumberPointstoRead.ToString());
throw (e);
}
success = true;
}
sr.Close();
}
catch (Exception e)
{
Console.WriteLine("Failed reading Points data " + DAVectorUtility.MPI_Rank.ToString() + " " + CountLinesinFile.ToString() + " Start "
+ FirstPointPosition.ToString() + " Number " + TotalNumberPointstoRead.ToString() + " " + line + e);
throw (e);
}
if (!success)
{
Exception e = DAVectorUtility.SALSAError("DA Vector File read error " + fname);
throw (e);
}
} // End ReadDataFromFile