internal HeuristicData(string heuristicName, string heuristicValue, RawData clone, Features.Mode featureMode)
: base(clone)
{
HeuristicName = heuristicName;
HeuristicValue = heuristicValue;
_featureMode = featureMode;
Data.Columns.Add("Makespan", typeof (int));
Data.Columns.Add("BestFoundMakespan", typeof(int));
Data.Columns.Add(heuristicName, typeof (string));
}
public RetraceSet(string distribution, string dimension, Trajectory track, int iter, bool extended,
int numFeat, int model, string stepwiseBias, Features.Mode featureMode, DirectoryInfo data)
: base(distribution, dimension, track, iter, extended, numFeat, model, stepwiseBias, data)
{
Read();
FeatureMode = featureMode;
if (FeatureMode != Features.Mode.Local)
FileInfo =
new FileInfo(FileInfo.FullName.Replace(Features.Mode.Local.ToString(), FeatureMode.ToString()));
}
internal HeuristicData(string distribution, string dimension, DataSet set, bool extended, string heuristicName,
string heuristicValue, DirectoryInfo data, Features.Mode featureMode)
: base(distribution, dimension, set, extended, data)
{
HeuristicName = heuristicName;
HeuristicValue = heuristicValue;
_featureMode = featureMode;
Data.Columns.Add("Makespan", typeof(int));
Data.Columns.Add("BestFoundMakespan", typeof(int));
Data.Columns.Add(heuristicName, typeof(string));
}
public PreferenceSet(string distribution, string dimension, Trajectory track, int iter,
bool extended, int numFeat, int model, string stepwiseBias, Ranking rank, Features.Mode featMode, DirectoryInfo data)
: base(distribution, dimension, track, iter, extended, numFeat, model, stepwiseBias, featMode, data)
{
FileInfo trainingFileInfo = new FileInfo(FileInfo.FullName);
this.FileInfo =
new FileInfo(string.Format(
@"{0}\Training\{1}.diff.{2}.csv", data.FullName,
FileInfo.Name.Substring(0, FileInfo.Name.Length - FileInfo.Extension.Length), (char) rank));
Data.Columns.Add("Rank", typeof (int));
var ranking = rank;
switch (ranking)
{
case Ranking.All:
_rankingFunction = AllRankings;
break;
case Ranking.Basic:
_rankingFunction = BasicRanking;
break;
case Ranking.FullPareto:
_rankingFunction = FullParetoRanking;
break;
case Ranking.PartialPareto:
_rankingFunction = PartialParetoRanking;
break;
}
if (FeatureMode == Features.Mode.Local)
ApplyAll(Retrace, null, null);
else
Read(trainingFileInfo);
_diffData = new List<Preference>[NumInstances, NumDimension];
for (int pid = 1; pid <= AlreadySavedPID; pid++)
for (int step = 0; step < NumDimension; step++)
_diffData[pid - 1, step] = new List<Preference>();
}
// commits dispatch!
public Features Dispatch1(int job, Features.Mode mode, LinearModel model)
{
Dispatch dispatch;
int slot = FindDispatch(job, out dispatch);
int time = _prob.Procs[job, dispatch.Mac];
int arrivalTime, slotReduced;
Features phi = new Features();
switch (mode)
{
case Features.Mode.Equiv:
phi.GetEquivPhi(job, this);
break;
}
_macs[dispatch.Mac].Update(dispatch.StartTime, time, slot, out slotReduced);
_jobs[job].Update(dispatch.StartTime, time, dispatch.Mac, out arrivalTime);
Sequence.Add(dispatch);
if (_jobs[job].MacCount == _prob.NumMachines)
ReadyJobs.Remove(job);
Makespan = _macs.Max(x => x.Makespan);
if (mode == Features.Mode.None) return null;
phi.GetLocalPhi(_jobs[job], _macs[dispatch.Mac], _prob.Procs[job, dispatch.Mac],
_jobs.Sum(p => p.WorkRemaining), _macs.Sum(p => p.TotSlack), Makespan, Sequence.Count,
dispatch.StartTime, arrivalTime, slotReduced, _totProcTime);
if (mode == Features.Mode.Global)
phi.GetGlobalPhi(this, model);
return phi;
}
public Features Difference(Features other)
{
Features diff = new Features();
for (int i = 0; i < LocalCount; i++)
diff.PhiLocal[i] = PhiLocal[i] - other.PhiLocal[i];
for (int i = 0; i < GlobalCount; i++)
diff.PhiGlobal[i] = PhiGlobal[i] - other.PhiGlobal[i];
for (int i = 0; i < SDRData.SDRCount; i++)
diff.Equiv[i] = Equiv[i] == other.Equiv[i];
diff.RND = null;
return diff;
}
public Preference(Schedule.Dispatch dispatch, Features features)
{
Dispatch = dispatch;
Feature = features;
}
public double PriorityIndex(Features phi)
{
var step = TimeIndependent ? 0 : phi.XiExplanatory[(int) Features.Explanatory.step] - 1;
double index = 0;
for (var i = 0; i < Features.LocalCount; i++)
index += LocalWeights[i][step]*phi.PhiLocal[i];
if (FeatureMode != Features.Mode.Global) return index;
for (var i = 0; i < Features.GlobalCount; i++)
index += GlobalWeights[i][step]*phi.PhiGlobal[i];
return index;
}
protected LinearModel(FileInfo file, Features.Mode featureMode, int timeDependentSteps, int numFeatures,
int modelID, string distribution, string dimension, Model type)
: this(file, featureMode, numFeatures, modelID, timeDependentSteps == 1, distribution, dimension, type)
{
for (int i = 0; i < Features.LocalCount; i++)
LocalWeights[i] = new double[timeDependentSteps];
if (featureMode != Features.Mode.Global) return;
for (int i = 0; i < Features.GlobalCount; i++)
GlobalWeights[i] = new double[timeDependentSteps];
}
protected LinearModel(FileInfo file, Features.Mode featureMode, int numFeatures, int modelID,
bool timeIndependent, string distribution, string dimension, Model type)
{
FileInfo = file;
FeatureMode = featureMode;
_numFeatures = numFeatures;
_modelID = modelID;
Name = String.Format("F{0}.M{1}", _numFeatures, _modelID);
TimeIndependent = timeIndependent;
Distribution = distribution;
Dimension = dimension;
Type = type;
}
public LinearModel(string distribution, string dimension, TrainingSet.Trajectory track, bool extended,
PreferenceSet.Ranking rank, bool timedependent, DirectoryInfo dataDir,
int numFeatures, int modelID, string stepwiseBias, int iter = -1, Features.Mode featMode = Features.Mode.Local)
: this(null, featMode, numFeatures, modelID, !timedependent, distribution, dimension, Model.PREF)
{
switch (track)
{
case TrainingSet.Trajectory.ILFIXSUP:
case TrainingSet.Trajectory.ILUNSUP:
case TrainingSet.Trajectory.ILSUP:
LinearModel model;
Iteration = GetImitationLearningFile(out model, distribution, dimension, track, extended,
numFeatures, modelID, dataDir.FullName, stepwiseBias, timedependent, iter);
FileInfo = model.FileInfo;
LocalWeights = model.LocalWeights;
return;
default:
string pat = String.Format("\\b(exhaust|full)\\.{0}.{1}.{2}.{3}{4}.{5}.{6}weights.{7}.csv",
distribution, dimension, (char) rank,
track, extended ? "EXT" : "", stepwiseBias,
FeatureMode == Features.Mode.Global ? "(Global|SDR)" : "",
timedependent ? "timedependent" : "timeindependent");
DirectoryInfo dir = new DirectoryInfo(String.Format(@"{0}\PREF\weights", dataDir.FullName));
Regex reg = new Regex(pat);
var files = dir.GetFiles("*.csv").Where(path => reg.IsMatch(path.ToString())).ToList();
if (files.Count <= 0)
throw new Exception(String.Format("Cannot find any weights belonging to {0}!", pat));
foreach (var file in files)
{
LinearModel[] logWeights = ReadLoggedLinearWeights(file, distribution, dimension, Model.PREF);
FileInfo = file;
foreach (
var w in logWeights.Where(w => w._numFeatures == _numFeatures && w._modelID == _modelID))
{
LocalWeights = w.LocalWeights;
GlobalWeights = w.GlobalWeights;
return;
}
}
throw new Exception(String.Format("Cannot find weights {0} to user requirements from {1}!", Name,
files[0].Name));
}
}