private void SimulatedReadoutData()
{
ReadoutData rd = new ReadoutData(2);
}
public void AddHist(ReadoutData rd)
{
Channel0Hist.Add(rd.Channel[0].Energies());
Channel1Hist.Add(rd.Channel[1].Energies());
}
public void SetWaveform(ReadoutData rd)
{
if (rd.Channel[0].Event.Length > 0 && rd.Channel[0].Event[0].Waveform != null)
{
Channel0DigitalProbe1 = CopyByteArray(rd.Channel[0].Event[0].Waveform.DTrace1);
Channel0DigitalProbe2 = CopyByteArray(rd.Channel[0].Event[0].Waveform.DTrace2);
Channel0VirtualProbe1 = CopyUShortArray(rd.Channel[0].Event[0].Waveform.Trace1);
Channel0VirtualProbe2 = CopyUShortArray(rd.Channel[0].Event[0].Waveform.Trace2);
}
if (rd.Channel[1].Event.Length > 0 && rd.Channel[1].Event[0].Waveform != null)
{
Channel1DigitalProbe1 = CopyByteArray(rd.Channel[1].Event[0].Waveform.DTrace1);
Channel1DigitalProbe2 = CopyByteArray(rd.Channel[1].Event[0].Waveform.DTrace2);
Channel1VirtualProbe1 = CopyUShortArray(rd.Channel[1].Event[0].Waveform.Trace1);
Channel1VirtualProbe2 = CopyUShortArray(rd.Channel[1].Event[0].Waveform.Trace2);
}
}
/// <summary>
/// Simulation update
/// </summary>
/// <returns></returns>
public bool OnSimulationUpdate(ReadoutData rd)
{
//Debug.WriteLine("boooo {0}", rd.Channel[0].Energies().Count);
Log.MCAData.AddHist(rd);
Log.MCAData.SetWaveform(rd);
return true;
}
/// <summary>
/// Update histogram
/// </summary>
/// <param name="rd"></param>
/// <returns></returns>
public bool OnExperimentUpdate(ReadoutData rd)
{
Log.MCAData.AddHist(rd);
Log.MCAData.SetWaveform(rd);
return true;
}
/// <summary>
/// Computes the readout layer.
/// </summary>
/// <param name="predictors">The predictors.</param>
/// <param name="readoutData">The detailed computed data.</param>
/// <returns>An output vector of the computed and naturalized values.</returns>
public double[] Compute(double[] predictors, out ReadoutData readoutData)
{
//Check readyness
if (!Trained)
{
throw new InvalidOperationException($"Readout layer is not trained. Build function has to be called before Compute function can be used.");
}
//Normalize predictors
double[] nrmPredictors = NormalizePredictors(predictors);
//Compute all readout units
CompositeResult[] unitsResults = ComputeReadoutUnits(nrmPredictors, out double[] nrmOutputVector);
//Build readout units results
ReadoutData.ReadoutUnitData[] readoutUnitsData = new ReadoutData.ReadoutUnitData[unitsResults.Length];
for (int unitIdx = 0; unitIdx < readoutUnitsData.Length; unitIdx++)
{
readoutUnitsData[unitIdx] = new ReadoutData.ReadoutUnitData()
{
Name = _readoutUnitCollection[unitIdx].Name,
Index = _readoutUnitCollection[unitIdx].Index,
Task = _readoutUnitCollection[unitIdx].Task,
CompResult = unitsResults[unitIdx],
RawNrmDataValue = nrmOutputVector[unitIdx],
RawNatDataValue = _outputFeatureFilterCollection[unitIdx].ApplyReverse(nrmOutputVector[unitIdx]),
FinalNatDataValue = _outputFeatureFilterCollection[unitIdx].ApplyReverse(nrmOutputVector[unitIdx])
};
}
//Compute all "One Takes All" groups
ReadoutData.OneTakesAllGroupData[] groupsData = null;
if (_oneTakesAllGroupCollection != null)
{
groupsData = new ReadoutData.OneTakesAllGroupData[_oneTakesAllGroupCollection.Length];
for (int groupIdx = 0; groupIdx < _oneTakesAllGroupCollection.Length; groupIdx++)
{
int groupInnerWinnerIdx = _oneTakesAllGroupCollection[groupIdx].Compute(unitsResults, out CompositeResult groupResult, out double[] groupOutputVector);
int layerWinnerIdx = _oneTakesAllGroupCollection[groupIdx].MemberReadoutUnitIndexCollection[groupInnerWinnerIdx];
groupsData[groupIdx] = new ReadoutData.OneTakesAllGroupData()
{
GroupName = _oneTakesAllGroupCollection[groupIdx].Name,
WinningReadoutUnitName = _readoutUnitCollection[layerWinnerIdx].Name,
WinningReadoutUnitIndex = layerWinnerIdx,
MemberWinningGroupIndex = groupInnerWinnerIdx,
MemberReadoutUnitIndexes = _oneTakesAllGroupCollection[groupIdx].MemberReadoutUnitIndexCollection.ToArray(),
CompResult = groupResult,
MemberProbabilities = groupOutputVector
};
//Update nrmOuputVector
for (int i = 0; i < _oneTakesAllGroupCollection[groupIdx].MemberReadoutUnitIndexCollection.Count; i++)
{
if (i == groupInnerWinnerIdx)
{
nrmOutputVector[_oneTakesAllGroupCollection[groupIdx].MemberReadoutUnitIndexCollection[i]] = InternalDataRange.Max;
}
else
{
nrmOutputVector[_oneTakesAllGroupCollection[groupIdx].MemberReadoutUnitIndexCollection[i]] = InternalDataRange.Min;
}
//nrmOutputVector[_oneTakesAllGroupCollection[groupIdx].MemberReadoutUnitIndexCollection[i]] = groupOutputVector[i];
}
}
}
//Output data finalization
double[] natOuputVector = NaturalizeOutputs(nrmOutputVector);
for (int unitIdx = 0; unitIdx < readoutUnitsData.Length; unitIdx++)
{
readoutUnitsData[unitIdx].FinalNatDataValue = natOuputVector[unitIdx];
}
readoutData = new ReadoutData(nrmOutputVector, natOuputVector, readoutUnitsData, groupsData);
return(natOuputVector);
}
