/// <summary>
/// Sets up the Azure storage (Points and Centroids) for the first k-means iteration.
/// </summary>
public void InitializeStorage()
{
AzureHelper.LogPerformance(() =>
{
Random random = new Random();
if (jobData.Points == null)
{
// Initialize the points blob with N random ClusterPoints
Points = AzureHelper.CreateBlob(jobData.JobID.ToString(), AzureHelper.PointsBlob);
using (ObjectStreamWriter <ClusterPoint> stream = new ObjectStreamWriter <ClusterPoint>(Points, point => point.ToByteArray(), ClusterPoint.Size))
{
for (int i = 0; i < jobData.N; i++)
{
stream.Write(new ClusterPoint(
random.NextDouble() * 100 - 50,
random.NextDouble() * 100 - 50,
Guid.Empty));
}
}
}
else
{
// Use the given points blob
Points = AzureHelper.GetBlob(jobData.Points);
// Initialize N based on that
using (ObjectStreamReader <ClusterPoint> stream = new ObjectStreamReader <ClusterPoint>(Points, ClusterPoint.FromByteArray, ClusterPoint.Size))
{
jobData.N = (int)stream.Length;
}
}
// Initialize the centroids blob with K random Centroids
Centroids = AzureHelper.CreateBlob(jobData.JobID.ToString(), AzureHelper.CentroidsBlob);
using (ObjectStreamWriter <Centroid> stream = new ObjectStreamWriter <Centroid>(Centroids, point => point.ToByteArray(), Centroid.Size))
{
for (int i = 0; i < jobData.K; i++)
{
stream.Write(new Centroid(
Guid.NewGuid(),
random.Next(-PointRange, PointRange),
random.Next(-PointRange, PointRange)));
}
}
}, jobID: jobData.JobID.ToString(), methodName: "InitializeStorage", iterationCount: IterationCount, points: new Lazy <string>(() => Points.Uri.ToString()), centroids: new Lazy <string>(() => Centroids.Uri.ToString()), machineID: MachineID);
}
private void RecalculateCentroids()
{
AzureHelper.LogPerformance(() =>
{
// Initialize the output blob
CloudBlob writeBlob = AzureHelper.CreateBlob(jobData.JobID.ToString(), Guid.NewGuid().ToString());
// Do the mapping and write the new blob
using (ObjectStreamReader <Centroid> stream = new ObjectStreamReader <Centroid>(Centroids, Centroid.FromByteArray, Centroid.Size))
{
var newCentroids = stream.Select(c =>
{
Point newCentroidPoint;
if (totalPointsProcessedDataByCentroid.ContainsKey(c.ID) && totalPointsProcessedDataByCentroid[c.ID].NumPointsProcessed != 0)
{
newCentroidPoint = totalPointsProcessedDataByCentroid[c.ID].PartialPointSum
/ (double)totalPointsProcessedDataByCentroid[c.ID].NumPointsProcessed;
}
else
{
newCentroidPoint = new Point();
}
c.X = newCentroidPoint.X;
c.Y = newCentroidPoint.Y;
return(c);
});
using (ObjectStreamWriter <Centroid> writeStream = new ObjectStreamWriter <Centroid>(writeBlob, point => point.ToByteArray(), Centroid.Size))
{
foreach (Centroid c in newCentroids)
{
writeStream.Write(c);
}
}
}
// Copy the contents of the new blob back into the old blob
Centroids.CopyFromBlob(writeBlob);
System.Diagnostics.Trace.TraceInformation("[ServerRole] Finished RecalculateCentroids(). Total points changed: {0}", TotalNumPointsChanged);
ResetPointChangedCounts();
}, jobData.JobID.ToString(), methodName: "RecalculateCentroids", iterationCount: IterationCount, points: Points.Uri.ToString(), centroids: Centroids.Uri.ToString(), machineID: MachineID);
}
