ENH: Standardize and Cache Random Seed values used in filters (BlueQu…

…artzSoftware#745)

* Feature clustering seed updates
* Add Bad Data seed updates
* K Filter seed updates
* Uncertain grid seed updates
* Merge twins seed updates
* Reorganize scalar segment features feature ids randomization
* Initialize Data Randomness Rework and Seed Updates
* Standardize Id Randomization to use a consistent seed
* Point Sample Triangle Geometry Seeding Updates

Signed-off-by: Michael Jackson <[email protected]>

---------

Signed-off-by: Michael Jackson <[email protected]>
Co-authored-by: Michael Jackson <[email protected]>

src/Plugins/ComplexCore/src/ComplexCore/Filters/AddBadDataFilter.cpp

@@ -4,9 +4,11 @@
 
 #include "complex/DataStructure/DataPath.hpp"
 #include "complex/DataStructure/Geometry/ImageGeom.hpp"
+#include "complex/Filter/Actions/CreateArrayAction.hpp"
 #include "complex/Filter/Actions/EmptyAction.hpp"
 #include "complex/Parameters/ArraySelectionParameter.hpp"
 #include "complex/Parameters/BoolParameter.hpp"
+#include "complex/Parameters/DataObjectNameParameter.hpp"
 #include "complex/Parameters/GeometrySelectionParameter.hpp"
 #include "complex/Parameters/NumberParameter.hpp"
 
@@ -52,9 +54,12 @@ Parameters AddBadDataFilter::parameters() const
   Parameters params;
 
   // Create the parameter descriptors that are needed for this filter
-  params.insertSeparator(Parameters::Separator{"Optional Variables"});
+  params.insertSeparator(Parameters::Separator{"Seeded Randomness"});
   params.insertLinkableParameter(std::make_unique<BoolParameter>(k_UseSeed_Key, "Use Seed for Random Generation", "When true the user will be able to put in a seed for random generation", false));
-  params.insert(std::make_unique<NumberParameter<uint64>>(k_SeedValue_Key, "Seed", "The seed fed into the random generator", std::mt19937::default_seed));
+  params.insert(std::make_unique<NumberParameter<uint64>>(k_SeedValue_Key, "Seed Value", "The seed fed into the random generator", std::mt19937::default_seed));
+  params.insert(std::make_unique<DataObjectNameParameter>(k_SeedArrayName_Key, "Stored Seed Value Array Name", "Name of array holding the seed value", "AddBadData SeedValue"));
+
+  params.insertSeparator(Parameters::Separator{"Optional Variables"});
   params.insertLinkableParameter(std::make_unique<BoolParameter>(k_PoissonNoise_Key, "Add Random Noise", "If true the user may set the poisson volume fraction", false));
   params.insert(std::make_unique<Float32Parameter>(k_PoissonVolFraction_Key, "Volume Fraction of Random Noise", "A value between 0 and 1 inclusive that is compared against random generation", 0.0f));
   params.insertLinkableParameter(std::make_unique<BoolParameter>(k_BoundaryNoise_Key, "Add Boundary Noise", "If true the user may set the boundary volume fraction", false));
@@ -91,6 +96,7 @@ IFilter::PreflightResult AddBadDataFilter::preflightImpl(const DataStructure& da
   auto pBoundaryVolFractionValue = filterArgs.value<float32>(k_BoundaryVolFraction_Key);
   auto pGBEuclideanDistancesArrayPathValue = filterArgs.value<DataPath>(k_GBEuclideanDistancesArrayPath_Key);
   auto pImageGeometryPathValue = filterArgs.value<DataPath>(k_ImageGeometryPath_Key);
+  auto pSeedArrayNameValue = filterArgs.value<std::string>(k_SeedArrayName_Key);
 
   PreflightResult preflightResult;
   complex::Result<OutputActions> resultOutputActions;
@@ -118,6 +124,11 @@ IFilter::PreflightResult AddBadDataFilter::preflightImpl(const DataStructure& da
     return {MakeErrorResult<OutputActions>(-76237, fmt::format("Image geometry must have a valid cell Attribute Matrix"))};
   }
 
+  {
+    auto createAction = std::make_unique<CreateArrayAction>(DataType::uint64, std::vector<usize>{1}, std::vector<usize>{1}, DataPath({pSeedArrayNameValue}));
+    resultOutputActions.value().appendAction(std::move(createAction));
+  }
+
   // Return both the resultOutputActions and the preflightUpdatedValues via std::move()
   return {std::move(resultOutputActions), std::move(preflightUpdatedValues)};
 }
@@ -126,6 +137,15 @@ IFilter::PreflightResult AddBadDataFilter::preflightImpl(const DataStructure& da
 Result<> AddBadDataFilter::executeImpl(DataStructure& dataStructure, const Arguments& filterArgs, const PipelineFilter* pipelineNode, const MessageHandler& messageHandler,
                                        const std::atomic_bool& shouldCancel) const
 {
+  auto seed = filterArgs.value<uint64>(k_SeedValue_Key);
+  if(!filterArgs.value<bool>(k_UseSeed_Key))
+  {
+    seed = static_cast<std::mt19937_64::result_type>(std::chrono::steady_clock::now().time_since_epoch().count());
+  }
+
+  // Store Seed Value in Top Level Array
+  dataStructure.getDataRefAs<UInt64Array>(DataPath({filterArgs.value<std::string>(k_SeedArrayName_Key)}))[0] = seed;
+
   AddBadDataInputValues inputValues;
 
   inputValues.PoissonNoise = filterArgs.value<bool>(k_PoissonNoise_Key);
@@ -134,7 +154,6 @@ Result<> AddBadDataFilter::executeImpl(DataStructure& dataStructure, const Argum
   inputValues.BoundaryVolFraction = filterArgs.value<float32>(k_BoundaryVolFraction_Key);
   inputValues.GBEuclideanDistancesArrayPath = filterArgs.value<DataPath>(k_GBEuclideanDistancesArrayPath_Key);
   inputValues.ImageGeometryPath = filterArgs.value<DataPath>(k_ImageGeometryPath_Key);
-  inputValues.UseSeed = filterArgs.value<bool>(k_UseSeed_Key);
   inputValues.SeedValue = filterArgs.value<uint64>(k_SeedValue_Key);
 
   return AddBadData(dataStructure, messageHandler, shouldCancel, &inputValues)();

src/Plugins/ComplexCore/src/ComplexCore/Filters/AddBadDataFilter.hpp

@@ -32,6 +32,7 @@ class COMPLEXCORE_EXPORT AddBadDataFilter : public IFilter
   static inline constexpr StringLiteral k_ImageGeometryPath_Key = "image_geometry_path";
   static inline constexpr StringLiteral k_UseSeed_Key = "use_seed";
   static inline constexpr StringLiteral k_SeedValue_Key = "seed_value";
+  static inline constexpr StringLiteral k_SeedArrayName_Key = "seed_array_name";
 
   /**
    * @brief Returns the name of the filter.

src/Plugins/ComplexCore/src/ComplexCore/Filters/Algorithms/AddBadData.cpp

@@ -1,7 +1,6 @@
 #include "AddBadData.hpp"
 
 #include "complex/DataStructure/DataArray.hpp"
-#include "complex/DataStructure/DataGroup.hpp"
 #include "complex/DataStructure/Geometry/ImageGeom.hpp"
 #include "complex/Utilities/DataGroupUtilities.hpp"
 #include "complex/Utilities/FilterUtilities.hpp"
@@ -46,14 +45,7 @@ Result<> AddBadData::operator()()
 {
   std::random_device randomDevice;        // Will be used to obtain a seed for the random number engine
   std::mt19937 generator(randomDevice()); // Standard mersenne_twister_engine seeded with rd()
-  std::mt19937::result_type seed = m_InputValues->SeedValue;
-
-  if(!m_InputValues->UseSeed)
-  {
-    seed = static_cast<std::mt19937::result_type>(std::chrono::steady_clock::now().time_since_epoch().count());
-  }
-
-  generator.seed(seed);
+  generator.seed(m_InputValues->SeedValue);
   std::uniform_real_distribution<float32> distribution(0.0F, 1.0F);
 
   auto& imgGeom = m_DataStructure.getDataRefAs<ImageGeom>(m_InputValues->ImageGeometryPath);

src/Plugins/ComplexCore/src/ComplexCore/Filters/Algorithms/AddBadData.hpp

@@ -19,7 +19,6 @@ struct COMPLEXCORE_EXPORT AddBadDataInputValues
   float32 BoundaryVolFraction;
   DataPath GBEuclideanDistancesArrayPath;
   DataPath ImageGeometryPath;
-  bool UseSeed;
   uint64 SeedValue;
 };
 

src/Plugins/ComplexCore/src/ComplexCore/Filters/Algorithms/FindFeatureClustering.cpp

@@ -12,7 +12,7 @@ namespace
 {
 // -----------------------------------------------------------------------------
 std::vector<float32> GenerateRandomDistribution(float32 minDistance, float32 maxDistance, int32 numBins, const std::array<float32, 3>& boxDims, const std::array<float32, 3>& boxRes,
-                                                bool useSeedFromUser, uint64 userSeedValue)
+                                                uint64 userSeedValue)
 {
   std::vector<float32> freq(numBins, 0);
   std::vector<float32> randomCentroids;
@@ -36,12 +36,7 @@ std::vector<float32> GenerateRandomDistribution(float32 minDistance, float32 max
 
   std::random_device randomDevice;           // Will be used to obtain a seed for the random number engine
   std::mt19937_64 generator(randomDevice()); // Standard mersenne_twister_engine seeded with rd()
-  std::mt19937::result_type seed = userSeedValue;
-  if(!useSeedFromUser)
-  {
-    seed = static_cast<std::mt19937::result_type>(std::chrono::steady_clock::now().time_since_epoch().count());
-  }
-  generator.seed(seed);
+  generator.seed(userSeedValue);
   std::uniform_real_distribution<double> distribution(0.0, 1.0);
 
   randomCentroids.resize(largeNumber * 3);
@@ -270,7 +265,7 @@ Result<> FindFeatureClustering::operator()()
 
   randomRDF.resize(currentNumBins + 1);
   // Call this function to generate the random distribution, which is normalized by the total number of distances
-  randomRDF = GenerateRandomDistribution(min, max, m_InputValues->NumberOfBins, boxDims, boxRes, m_InputValues->UseSeed, m_InputValues->SeedValue);
+  randomRDF = GenerateRandomDistribution(min, max, m_InputValues->NumberOfBins, boxDims, boxRes, m_InputValues->SeedValue);
 
   // Scale the random distribution by the number of distances in this particular instance
   const float32 normFactor = totalPptFeatures * (totalPptFeatures - 1);

src/Plugins/ComplexCore/src/ComplexCore/Filters/Algorithms/FindFeatureClustering.hpp

@@ -16,7 +16,6 @@ struct COMPLEXCORE_EXPORT FindFeatureClusteringInputValues
   int32 NumberOfBins;
   int32 PhaseNumber;
   bool RemoveBiasedFeatures;
-  bool UseSeed;
   uint64 SeedValue;
   DataPath EquivalentDiametersArrayPath;
   DataPath FeaturePhasesArrayPath;

src/Plugins/ComplexCore/src/ComplexCore/Filters/Algorithms/PointSampleTriangleGeometry.cpp

@@ -67,8 +67,7 @@ Result<> PointSampleTriangleGeometry::operator()()
   auto tupleShape = {static_cast<usize>(m_Inputs->pNumberOfSamples)};
   vertex.getVertexAttributeMatrix()->resizeTuples(tupleShape);
 
-  std::mt19937_64::result_type seed = static_cast<std::mt19937_64::result_type>(std::chrono::steady_clock::now().time_since_epoch().count());
-  std::mt19937_64 generator(seed);
+  std::mt19937_64 generator(m_Inputs->Seed);
   std::uniform_real_distribution<> distribution(0.0f, 1.0f);
 
   // Create the Triangle Ids vector and fill it from zero to size-1 incrementally

src/Plugins/ComplexCore/src/ComplexCore/Filters/Algorithms/PointSampleTriangleGeometry.hpp

@@ -69,6 +69,7 @@ struct COMPLEXCORE_EXPORT PointSampleTriangleGeometryInputs
   DataPath pVertexGeometryPath;
   DataPath pVertexGroupDataPath;
   MultiArraySelectionParameter::ValueType pCreatedDataArrayPaths;
+  uint64 Seed;
 };
 
 /**

src/Plugins/ComplexCore/src/ComplexCore/Filters/Algorithms/ScalarSegmentFeatures.cpp

@@ -6,8 +6,6 @@
 #include "complex/Parameters/BoolParameter.hpp"
 #include "complex/Utilities/DataArrayUtilities.hpp"
 
-#include <chrono>
-
 using namespace complex;
 
 #define CX_DEFAULT_CONSTRUCTORS(className)                                                                                                                                                             \
@@ -210,11 +208,6 @@ Result<> ScalarSegmentFeatures::operator()()
   // Generate the random voxel indices that will be used for the seed points to start a new grain growth/agglomeration
   auto totalPoints = inputDataArray->getNumberOfTuples();
 
-  const int64 rangeMin = 0;
-  const int64 rangeMax = totalPoints - 1;
-  Int64Distribution distribution;
-  initializeVoxelSeedGenerator(distribution, rangeMin, rangeMax);
-
   //  // Add compare function to arguments
   //  Arguments newArgs = args;
   //  newArgs.insert(k_CompareFunctKey, compare.get());
@@ -232,9 +225,15 @@ Result<> ScalarSegmentFeatures::operator()()
   activeArray->getDataStore()->fill(1);
   (*activeArray)[0] = 0;
 
-  // By default we randomize grains
+  // Randomize the feature Ids for purely visual clarify. Having random Feature Ids
+  // allows users visualizing the data to better discern each grain otherwise the coloring
+  // would look like a smooth gradient. This is a user input parameter
   if(m_InputValues->pShouldRandomizeFeatureIds)
   {
+    const int64 rangeMin = 0;
+    const int64 rangeMax = totalPoints - 1;
+    Int64Distribution distribution;
+    initializeStaticVoxelSeedGenerator(distribution, rangeMin, rangeMax);
     totalPoints = gridGeom->getNumberOfCells();
     randomizeFeatureIds(m_FeatureIdsArray, totalPoints, totalFeatures, distribution);
   }

src/Plugins/ComplexCore/src/ComplexCore/Filters/Algorithms/SharedFeatureFace.cpp

@@ -65,11 +65,10 @@ class SharedFeatureFaceImpl
 using SeedGenerator = std::mt19937_64;
 using Int64Distribution = std::uniform_int_distribution<int64>;
 // -----------------------------------------------------------------------------
-SeedGenerator initializeVoxelSeedGenerator(Int64Distribution& distribution, const int64 rangeMin, const int64 rangeMax)
+SeedGenerator initializeStaticVoxelSeedGenerator(Int64Distribution& distribution, const int64 rangeMin, const int64 rangeMax)
 {
-  auto seed = static_cast<SeedGenerator::result_type>(std::chrono::steady_clock::now().time_since_epoch().count());
   SeedGenerator generator;
-  generator.seed(seed);
+  generator.seed(SeedGenerator::default_seed);
   distribution = std::uniform_int_distribution<int64>(rangeMin, rangeMax);
 
   return generator;
@@ -81,7 +80,7 @@ void RandomizeFaceIds(complex::Int32Array& featureIds, uint64 totalFeatures, Int
   // Generate an even distribution of numbers between the min and max range
   const int64 rangeMin = 1;
   const int64 rangeMax = totalFeatures - 1;
-  auto generator = initializeVoxelSeedGenerator(distribution, rangeMin, rangeMax);
+  auto generator = initializeStaticVoxelSeedGenerator(distribution, rangeMin, rangeMax);
 
   DataStructure tmpStructure;
   auto rndNumbers = Int64Array::CreateWithStore<DataStore<int64>>(tmpStructure, std::string("_INTERNAL_USE_ONLY_NewFeatureIds"), std::vector<usize>{totalFeatures}, std::vector<usize>{1});
@@ -216,7 +215,7 @@ Result<> SharedFeatureFace::operator()()
     const int64 rangeMin = 0;
     const int64 rangeMax = static_cast<int64>(surfaceMeshFeatureFaceNumTriangles.getNumberOfTuples() - 1);
     Int64Distribution distribution;
-    initializeVoxelSeedGenerator(distribution, rangeMin, rangeMax);
+    initializeStaticVoxelSeedGenerator(distribution, rangeMin, rangeMax);
     ::RandomizeFaceIds(surfaceMeshFeatureFaceIds, index, distribution);
   }
   return {};

src/Plugins/ComplexCore/src/ComplexCore/Filters/Algorithms/UncertainRegularGridSampleSurfaceMesh.cpp

@@ -1,7 +1,6 @@
 #include "UncertainRegularGridSampleSurfaceMesh.hpp"
 
 #include "complex/DataStructure/DataArray.hpp"
-#include "complex/DataStructure/DataGroup.hpp"
 
 #include <random>
 
@@ -39,14 +38,7 @@ void UncertainRegularGridSampleSurfaceMesh::generatePoints(std::vector<Point3Df>
 
   std::random_device randomDevice;        // Will be used to obtain a seed for the random number engine
   std::mt19937 generator(randomDevice()); // Standard mersenne_twister_engine seeded with rd()
-  std::mt19937::result_type seed = m_InputValues->SeedValue;
-
-  if(!m_InputValues->UseSeed)
-  {
-    seed = static_cast<std::mt19937::result_type>(std::chrono::steady_clock::now().time_since_epoch().count());
-  }
-
-  generator.seed(seed);
+  generator.seed(m_InputValues->SeedValue);
   std::uniform_real_distribution<float32> distribution(0.0F, 1.0F);
 
   for(usize k = 0; k < dims[2]; k++)

src/Plugins/ComplexCore/src/ComplexCore/Filters/Algorithms/UncertainRegularGridSampleSurfaceMesh.hpp

@@ -16,7 +16,6 @@ namespace complex
 {
 struct COMPLEXCORE_EXPORT UncertainRegularGridSampleSurfaceMeshInputValues
 {
-  bool UseSeed;
   uint64 SeedValue;
   VectorUInt64Parameter::ValueType Dimensions;
   VectorFloat32Parameter::ValueType Spacing;

src/Plugins/ComplexCore/src/ComplexCore/Filters/FindFeatureClusteringFilter.cpp

@@ -61,9 +61,13 @@ Parameters FindFeatureClusteringFilter::parameters() const
   params.insert(std::make_unique<Int32Parameter>(k_NumberOfBins_Key, "Number of Bins for RDF", "Number of bins to split the RDF", 1));
   params.insert(std::make_unique<Int32Parameter>(k_PhaseNumber_Key, "Phase Index", "Ensemble number for which to calculate the RDF and clustering list", 1));
   params.insertLinkableParameter(std::make_unique<BoolParameter>(k_RemoveBiasedFeatures_Key, "Remove Biased Features", "Remove the biased features", false));
+
+  params.insertSeparator(Parameters::Separator{"Seeded Randomness"});
   params.insertLinkableParameter(
       std::make_unique<BoolParameter>(k_SetRandomSeed_Key, "Set Random Seed", "When checked, allows the user to set the seed value used to randomly generate the points in the RDF", true));
   params.insert(std::make_unique<UInt64Parameter>(k_SeedValue_Key, "Seed Value", "The seed value used to randomly generate the points in the RDF", std::mt19937::default_seed));
+  params.insert(std::make_unique<DataObjectNameParameter>(k_SeedArrayName_Key, "Stored Seed Value Array Name", "Name of array holding the seed value", "FindFeatureClustering SeedValue"));
+
   params.insertSeparator(Parameters::Separator{"Cell Feature Data"});
   params.insert(std::make_unique<ArraySelectionParameter>(k_EquivalentDiametersArrayPath_Key, "Equivalent Diameters", "Diameter of a sphere with the same volume as the Feature", DataPath{},
                                                           ArraySelectionParameter::AllowedTypes{DataType::float32}, ArraySelectionParameter::AllowedComponentShapes{{1}}));
@@ -83,6 +87,7 @@ Parameters FindFeatureClusteringFilter::parameters() const
   params.insertSeparator(Parameters::Separator{"Created Cell Ensemble Data"});
   params.insert(std::make_unique<DataObjectNameParameter>(k_RDFArrayName_Key, "Radial Distribution Function", "A histogram of the normalized frequency at each bin", "RDF"));
   params.insert(std::make_unique<DataObjectNameParameter>(k_MaxMinArrayName_Key, "Max and Min Separation Distances", "The max and min distance found between Features", "RDFMaxMinDistances"));
+
   // Associate the Linkable Parameter(s) to the children parameters that they control
   params.linkParameters(k_RemoveBiasedFeatures_Key, k_BiasedFeaturesArrayPath_Key, true);
   params.linkParameters(k_SetRandomSeed_Key, k_SeedValue_Key, true);
@@ -109,6 +114,7 @@ IFilter::PreflightResult FindFeatureClusteringFilter::preflightImpl(const DataSt
   auto pClusteringListArrayNameValue = filterArgs.value<std::string>(k_ClusteringListArrayName_Key);
   auto pRDFArrayNameValue = filterArgs.value<std::string>(k_RDFArrayName_Key);
   auto pMaxMinArrayNameValue = filterArgs.value<std::string>(k_MaxMinArrayName_Key);
+  auto pSeedArrayNameValue = filterArgs.value<std::string>(k_SeedArrayName_Key);
 
   const DataPath clusteringListPath = pFeaturePhasesArrayPathValue.getParent().createChildPath(pClusteringListArrayNameValue);
 
@@ -129,7 +135,7 @@ IFilter::PreflightResult FindFeatureClusteringFilter::preflightImpl(const DataSt
   }
 
   const auto& cellEnsembleAM = dataStructure.getDataRefAs<AttributeMatrix>(pCellEnsembleAttributeMatrixNameValue);
-  const std::vector<usize> tupleShape = cellEnsembleAM.getShape();
+  const std::vector<usize>& tupleShape = cellEnsembleAM.getShape();
   {
     auto createArrayAction = std::make_unique<CreateArrayAction>(DataType::float32, tupleShape, std::vector<usize>{static_cast<usize>(pNumberOfBinsValue)},
                                                                  pCellEnsembleAttributeMatrixNameValue.createChildPath(pRDFArrayNameValue));
@@ -146,21 +152,34 @@ IFilter::PreflightResult FindFeatureClusteringFilter::preflightImpl(const DataSt
     resultOutputActions.value().appendAction(std::move(createArrayAction));
   }
 
+  {
+    auto createAction = std::make_unique<CreateArrayAction>(DataType::uint64, std::vector<usize>{1}, std::vector<usize>{1}, DataPath({pSeedArrayNameValue}));
+    resultOutputActions.value().appendAction(std::move(createAction));
+  }
+
   return {std::move(resultOutputActions), std::move(preflightUpdatedValues)};
 }
 
 //------------------------------------------------------------------------------
 Result<> FindFeatureClusteringFilter::executeImpl(DataStructure& dataStructure, const Arguments& filterArgs, const PipelineFilter* pipelineNode, const MessageHandler& messageHandler,
                                                   const std::atomic_bool& shouldCancel) const
 {
+  auto seed = filterArgs.value<uint64>(k_SeedValue_Key);
+  if(!filterArgs.value<bool>(k_SetRandomSeed_Key))
+  {
+    seed = static_cast<std::mt19937_64::result_type>(std::chrono::steady_clock::now().time_since_epoch().count());
+  }
+
+  // Store Seed Value in Top Level Array
+  dataStructure.getDataRefAs<UInt64Array>(DataPath({filterArgs.value<std::string>(k_SeedArrayName_Key)}))[0] = seed;
+
   FindFeatureClusteringInputValues inputValues;
 
   inputValues.ImageGeometryPath = filterArgs.value<DataPath>(k_SelectedImageGeometry_Key);
   inputValues.NumberOfBins = filterArgs.value<int32>(k_NumberOfBins_Key);
   inputValues.PhaseNumber = filterArgs.value<int32>(k_PhaseNumber_Key);
   inputValues.RemoveBiasedFeatures = filterArgs.value<bool>(k_RemoveBiasedFeatures_Key);
-  inputValues.UseSeed = filterArgs.value<bool>(k_SetRandomSeed_Key);
-  inputValues.SeedValue = filterArgs.value<uint64>(k_SeedValue_Key);
+  inputValues.SeedValue = seed;
   inputValues.EquivalentDiametersArrayPath = filterArgs.value<DataPath>(k_EquivalentDiametersArrayPath_Key);
   inputValues.FeaturePhasesArrayPath = filterArgs.value<DataPath>(k_FeaturePhasesArrayPath_Key);
   inputValues.CentroidsArrayPath = filterArgs.value<DataPath>(k_CentroidsArrayPath_Key);