fixing fluid.normalize json reloading issue (#262)

* a simple line... * proper initialisation * remove the change of scaling post fitting (as it was in the RT version in SC)
flucoma · Feb 23, 2024 · 8e07590 · 8e07590
1 parent 3335b40
commit 8e07590
Show file tree

Hide file tree

Showing 2 changed files with 3 additions and 6 deletions.
diff --git a/include/algorithms/public/Normalization.hpp b/include/algorithms/public/Normalization.hpp
@@ -49,6 +49,8 @@ class Normalization
  using namespace _impl;
  mMin = min;
  mMax = max;
+ mRange = mMax - mMin;
+ handleZerosInScale(mRange);
  mDataMin = asEigen<Array>(dataMin);
  mDataMax = asEigen<Array>(dataMax);
  mDataRange = mDataMax - mDataMin;
@@ -132,6 +134,7 @@ class Normalization
  {
  mMin = 0;
  mMax = 1.0;
+ mRange = 1.0;
  mDataMin.setZero();
  mDataMax.setZero();
  mDataRange.setZero();

diff --git a/include/clients/nrt/NormalizeClient.hpp b/include/clients/nrt/NormalizeClient.hpp
@@ -139,8 +139,6 @@ class NormalizeClient : public FluidBaseClient,
  StringVector ids{srcDataSet.getIds()};
  RealMatrix data(srcDataSet.size(), srcDataSet.pointSize());
  if (!mAlgorithm.initialized()) return Error(NoDataFitted);
- mAlgorithm.setMin(get<kMin>());
- mAlgorithm.setMax(get<kMax>());
  mAlgorithm.process(srcDataSet.getData(), data, invert);
  FluidDataSet<string, double, 1> result(ids, data);
  destPtr->setDataSet(result);
@@ -166,8 +164,6 @@ class NormalizeClient : public FluidBaseClient,
  RealVector src(mAlgorithm.dims());
  RealVector dest(mAlgorithm.dims());
  src <<= BufferAdaptor::ReadAccess(in.get()).samps(0, mAlgorithm.dims(), 0);
- mAlgorithm.setMin(get<kMin>());
- mAlgorithm.setMax(get<kMax>());
  mAlgorithm.processFrame(src, dest, invert);
  outBuf.samps(0, mAlgorithm.dims(), 0) <<= dest;
  return OK();
@@ -239,8 +235,6 @@ class NormalizeQuery : public FluidBaseClient, ControlIn, ControlOut
  RealVector dest(algorithm.dims());
  src <<= BufferAdaptor::ReadAccess(get<kInputBuffer>().get())
  .samps(0, algorithm.dims(), 0);
-// algorithm.setMin(get<kMin>());
-// algorithm.setMax(get<kMax>());
  algorithm.processFrame(src, dest, get<kInvert>() == 1);
  outBuf.samps(0, algorithm.dims(), 0) <<= dest;
  }