Package org.apache.sysds.runtime.compress.colgroup.dictionary

Interface IDictionary

All Known Implementing Classes:
ACachingMBDictionary, ADictionary, AIdentityDictionary, DeltaDictionary, Dictionary, IdentityDictionary, IdentityDictionarySlice, MatrixBlockDictionary, PlaceHolderDict, QDictionary
public interface IDictionary

  • Field Details

    • LOG

      static final org.apache.commons.logging.Log LOG

  • Method Details

    • getValues

      double[] getValues()
      Get all the values contained in the dictionary as a linearized double array.
      Returns:
      linearized double array

    • getValue

      double getValue(int i)
      Get Specific value contained in the dictionary at index.
      Parameters:
      i - The index to extract the value from
      Returns:
      The value contained at the index

    • getValue

      double getValue(int r, int col, int nCol)
      Get Specific value contain in dictionary at index.
      Parameters:
      r - Row target
      col - Col target
      nCol - nCol in dictionary
      Returns:
      value

    • getInMemorySize

      long getInMemorySize()
      Returns the memory usage of the dictionary.
      Returns:
      a long value in number of bytes for the dictionary.

    • aggregate

      double aggregate(double init, Builtin fn)
      Aggregate all the contained values, useful in value only computations where the operation is iterating through all values contained in the dictionary.
      Parameters:
      init - The initial Value, in cases such as Max value, this could be -infinity
      fn - The Function to apply to values
      Returns:
      The aggregated value as a double.

    • aggregateWithReference

      double aggregateWithReference(double init, Builtin fn, double[] reference, boolean def)
      Aggregate all the contained values, with a reference offset.
      Parameters:
      init - The initial value, in cases such as Max value this could be -infinity.
      fn - The function to apply to the values
      reference - The reference offset to each value in the dictionary
      def - If the reference should be treated as an instance of only as reference
      Returns:
      The aggregated value as a double.

    • aggregateRows

      double[] aggregateRows(Builtin fn, int nCol)
      Aggregate all entries in the rows.
      Parameters:
      fn - The aggregate function
      nCol - The number of columns contained in the dictionary.
      Returns:
      Aggregates for this dictionary tuples.

    • aggregateRowsWithDefault

      double[] aggregateRowsWithDefault(Builtin fn, double[] defaultTuple)
      Aggregate all entries in the rows of the dictionary with a extra cell in the end that contains the aggregate of the given defaultTuple.
      Parameters:
      fn - The aggregate function
      defaultTuple - The default tuple to aggregate in last cell
      Returns:
      Aggregates for this dictionary tuples.

    • aggregateRowsWithReference

      double[] aggregateRowsWithReference(Builtin fn, double[] reference)
      Aggregate all entries in the rows with an offset value reference added.
      Parameters:
      fn - The aggregate function
      reference - The reference offset to each value in the dictionary
      Returns:
      Aggregates for this dictionary tuples.

    • aggregateCols

      void aggregateCols(double[] c, Builtin fn, IColIndex colIndexes)
      Aggregates the columns into the target double array provided.
      Parameters:
      c - The target double array, this contains the full number of columns, therefore the colIndexes for this specific dictionary is needed.
      fn - The function to apply to individual columns
      colIndexes - The mapping to the target columns from the individual columns

    • aggregateColsWithReference

      void aggregateColsWithReference(double[] c, Builtin fn, IColIndex colIndexes, double[] reference, boolean def)
      Aggregates the columns into the target double array provided.
      Parameters:
      c - The target double array, this contains the full number of columns, therefore the colIndexes for this specific dictionary is needed.
      fn - The function to apply to individual columns
      colIndexes - The mapping to the target columns from the individual columns
      reference - The reference offset values to add to each cell.
      def - If the reference should be treated as a tuple as well

    • applyScalarOp

      IDictionary applyScalarOp(ScalarOperator op)
      Allocate a new dictionary and applies the scalar operation on each cell of to then return the new dictionary.
      Parameters:
      op - The operator.
      Returns:
      The new dictionary to return.

    • applyScalarOpAndAppend

      IDictionary applyScalarOpAndAppend(ScalarOperator op, double v0, int nCol)
      Allocate a new dictionary with one extra row and applies the scalar operation on each cell of to then return the new dictionary.
      Parameters:
      op - The operator
      v0 - The new value to put into each cell in the new row
      nCol - The number of columns in the dictionary
      Returns:
      The new dictionary to return.

    • applyUnaryOp

      IDictionary applyUnaryOp(UnaryOperator op)
      Allocate a new dictionary and apply the unary operator on each cell.
      Parameters:
      op - The operator.
      Returns:
      The new dictionary to return.

    • applyUnaryOpAndAppend

      IDictionary applyUnaryOpAndAppend(UnaryOperator op, double v0, int nCol)
      Allocate a new dictionary with one extra row and apply the unary operator on each cell.
      Parameters:
      op - The operator.
      v0 - The new value to put into each cell in the new row
      nCol - The number of columns in the dictionary
      Returns:
      The new dictionary to return.

    • applyScalarOpWithReference

      IDictionary applyScalarOpWithReference(ScalarOperator op, double[] reference, double[] newReference)
      Allocate a new dictionary and apply the scalar operation on each cell to then return a new dictionary. outValues[j] = op(this.values[j] + reference[i]) - newReference[i]
      Parameters:
      op - The operator to apply to each cell.
      reference - The reference value to add before the operator.
      newReference - The reference value to subtract after the operator.
      Returns:
      A New Dictionary.

    • applyUnaryOpWithReference

      IDictionary applyUnaryOpWithReference(UnaryOperator op, double[] reference, double[] newReference)
      Allocate a new dictionary and apply the scalar operation on each cell to then return a new dictionary. outValues[j] = op(this.values[j] + reference[i]) - newReference[i]
      Parameters:
      op - The unary operator to apply to each cell.
      reference - The reference value to add before the operator.
      newReference - The reference value to subtract after the operator.
      Returns:
      A New Dictionary.

    • binOpLeft

      IDictionary binOpLeft(BinaryOperator op, double[] v, IColIndex colIndexes)
      Apply binary row operation on the left side
      Parameters:
      op - The operation to this dictionary
      v - The values to use on the left hand side.
      colIndexes - The column indexes to consider inside v.
      Returns:
      A new dictionary containing the updated values.

    • binOpLeftAndAppend

      IDictionary binOpLeftAndAppend(BinaryOperator op, double[] v, IColIndex colIndexes)
      Apply binary row operation on the left side with one extra row evaluating with zeros.
      Parameters:
      op - The operation to this dictionary
      v - The values to use on the left hand side.
      colIndexes - The column indexes to consider inside v.
      Returns:
      A new dictionary containing the updated values.

    • binOpLeftWithReference

      IDictionary binOpLeftWithReference(BinaryOperator op, double[] v, IColIndex colIndexes, double[] reference, double[] newReference)
      Apply the binary operator such that each value is offset by the reference before application. Then put the result into the new dictionary, but offset it by the new reference. outValues[j] = op(v[colIndexes[i]], this.values[j] + reference[i]) - newReference[i]
      Parameters:
      op - The operation to apply on the dictionary values.
      v - The values to use on the left side of the operator.
      colIndexes - The column indexes to use.
      reference - The reference value to add before operator.
      newReference - The reference value to subtract after operator.
      Returns:
      A new dictionary.

    • binOpRight

      IDictionary binOpRight(BinaryOperator op, double[] v, IColIndex colIndexes)
      Apply binary row operation on the right side.
      Parameters:
      op - The operation to this dictionary
      v - The values to use on the right hand side.
      colIndexes - The column indexes to consider inside v.
      Returns:
      A new dictionary containing the updated values.

    • binOpRightAndAppend

      IDictionary binOpRightAndAppend(BinaryOperator op, double[] v, IColIndex colIndexes)
      Apply binary row operation on the right side with one extra row evaluating with zeros.
      Parameters:
      op - The operation to this dictionary
      v - The values to use on the right hand side.
      colIndexes - The column indexes to consider inside v.
      Returns:
      A new dictionary containing the updated values.

    • binOpRight

      IDictionary binOpRight(BinaryOperator op, double[] v)
      Apply binary row operation on the right side as with no columns to extract from v.
      Parameters:
      op - The operation to this dictionary
      v - The values to apply on the dictionary (same number of cols as the dictionary)
      Returns:
      A new dictionary containing the updated values.

    • binOpRightWithReference

      IDictionary binOpRightWithReference(BinaryOperator op, double[] v, IColIndex colIndexes, double[] reference, double[] newReference)
      Apply the binary operator such that each value is offset by the reference before application. Then put the result into the new dictionary, but offset it by the new reference. outValues[j] = op(this.values[j] + reference[i], v[colIndexes[i]]) - newReference[i]
      Parameters:
      op - The operation to apply on the dictionary values.
      v - The values to use on the right side of the operator.
      colIndexes - The column indexes to use.
      reference - The reference value to add before operator.
      newReference - The reference value to subtract after operator.
      Returns:
      A new dictionary.

    • clone

      IDictionary clone()
      Returns a deep clone of the dictionary.
      Returns:
      A deep clone

    • write

      void write(DataOutput out) throws IOException
      Write the dictionary to a DataOutput.
      Parameters:
      out - the output sink to write the dictionary to.
      Throws:
      IOException - if the sink fails.

    • getExactSizeOnDisk

      long getExactSizeOnDisk()
      Calculate the space consumption if the dictionary is stored on disk.
      Returns:
      the long count of bytes to store the dictionary.

    • getDictType

      IDictionary.DictType getDictType()
      Get the dictionary type this dictionary is.
      Returns:
      The Dictionary type this is.

    • getNumberOfValues

      int getNumberOfValues(int ncol)
      Get the number of distinct tuples given that the column group has n columns
      Parameters:
      ncol - The number of Columns in the ColumnGroup.
      Returns:
      the number of value tuples contained in the dictionary.

    • getNumberOfColumns

      int getNumberOfColumns(int nrow)
      Get the number of columns in this dictionary, provided you know the number of values, or rows.
      Parameters:
      nrow - The number of rows/values known inside this dictionary
      Returns:
      The number of columns

    • sumAllRowsToDouble

      double[] sumAllRowsToDouble(int nrColumns)
      Method used as a pre-aggregate of each tuple in the dictionary, to single double values. Note if the number of columns is one the actual dictionaries values are simply returned.
      Parameters:
      nrColumns - The number of columns in the ColGroup to know how to get the values from the dictionary.
      Returns:
      a double array containing the row sums from this dictionary.

    • sumAllRowsToDoubleWithDefault

      double[] sumAllRowsToDoubleWithDefault(double[] defaultTuple)
      Do exactly the same as the sumAllRowsToDouble but also sum the array given to a extra index in the end of the array.
      Parameters:
      defaultTuple - The default row to sum in the end index returned.
      Returns:
      a double array containing the row sums from this dictionary.

    • sumAllRowsToDoubleWithReference

      double[] sumAllRowsToDoubleWithReference(double[] reference)
      Method used as a pre-aggregate of each tuple in the dictionary, to single double values with a reference.
      Parameters:
      reference - The reference values to add to each cell.
      Returns:
      a double array containing the row sums from this dictionary.

    • sumAllRowsToDoubleSq

      double[] sumAllRowsToDoubleSq(int nrColumns)
      Method used as a pre-aggregate of each tuple in the dictionary, to single double values. Note if the number of columns is one the actual dictionaries values are simply returned.
      Parameters:
      nrColumns - The number of columns in the ColGroup to know how to get the values from the dictionary.
      Returns:
      a double array containing the row sums from this dictionary.

    • sumAllRowsToDoubleSqWithDefault

      double[] sumAllRowsToDoubleSqWithDefault(double[] defaultTuple)
      Method used as a pre-aggregate of each tuple in the dictionary, to single double values. But adds another cell to the return with an extra value that is the sum of the given defaultTuple.
      Parameters:
      defaultTuple - The default row to sum in the end index returned.
      Returns:
      a double array containing the row sums from this dictionary.

    • sumAllRowsToDoubleSqWithReference

      double[] sumAllRowsToDoubleSqWithReference(double[] reference)
      Method used as a pre-aggregate of each tuple in the dictionary, to single double values.
      Parameters:
      reference - The reference values to add to each cell.
      Returns:
      a double array containing the row sums from this dictionary.

    • productAllRowsToDouble

      double[] productAllRowsToDouble(int nrColumns)
      Method to product all rows to a column vector.
      Parameters:
      nrColumns - The number of columns in the ColGroup to know how to get the values from the dictionary.
      Returns:
      A row product

    • productAllRowsToDoubleWithDefault

      double[] productAllRowsToDoubleWithDefault(double[] defaultTuple)
      Method to product all rows to a column vector with a default value added in the end.
      Parameters:
      defaultTuple - The default row that aggregate to last cell
      Returns:
      A row product

    • productAllRowsToDoubleWithReference

      double[] productAllRowsToDoubleWithReference(double[] reference)
      Method to product all rows to a column vector with a reference values added to all cells, and a reference product in the end
      Parameters:
      reference - The reference row
      Returns:
      A row product

    • colSum

      void colSum(double[] c, int[] counts, IColIndex colIndexes)
      Get the column sum of the values contained in the dictionary
      Parameters:
      c - The output array allocated to contain all column groups output.
      counts - The counts of the individual tuples.
      colIndexes - The columns indexes of the parent column group, this indicate where to put the column sum into the c output.

    • colSumSq

      void colSumSq(double[] c, int[] counts, IColIndex colIndexes)
      Get the column sum of the values contained in the dictionary
      Parameters:
      c - The output array allocated to contain all column groups output.
      counts - The counts of the individual tuples.
      colIndexes - The columns indexes of the parent column group, this indicate where to put the column sum into the c output.

    • colSumSqWithReference

      void colSumSqWithReference(double[] c, int[] counts, IColIndex colIndexes, double[] reference)
      Get the column sum of the values contained in the dictionary with an offset reference value added to each cell.
      Parameters:
      c - The output array allocated to contain all column groups output.
      counts - The counts of the individual tuples.
      colIndexes - The columns indexes of the parent column group, this indicate where to put the column sum into the c output.
      reference - The reference values to add to each cell.

    • sum

      double sum(int[] counts, int nCol)
      Get the sum of the values contained in the dictionary
      Parameters:
      counts - The counts of the individual tuples
      nCol - The number of columns contained
      Returns:
      The sum scaled by the counts provided.

    • sumSq

      double sumSq(int[] counts, int nCol)
      Get the square sum of the values contained in the dictionary
      Parameters:
      counts - The counts of the individual tuples
      nCol - The number of columns contained
      Returns:
      The square sum scaled by the counts provided.

    • sumSqWithReference

      double sumSqWithReference(int[] counts, double[] reference)
      Get the square sum of the values contained in the dictionary with a reference offset on each value.
      Parameters:
      counts - The counts of the individual tuples
      reference - The reference value
      Returns:
      The square sum scaled by the counts and reference.

    • getString

      String getString(int colIndexes)
      Get a string representation of the dictionary, that considers the layout of the data.
      Parameters:
      colIndexes - The number of columns in the dictionary.
      Returns:
      A string that is nicer to print.

    • sliceOutColumnRange

      IDictionary sliceOutColumnRange(int idxStart, int idxEnd, int previousNumberOfColumns)
      Modify the dictionary by removing columns not within the index range.
      Parameters:
      idxStart - The column index to start at.
      idxEnd - The column index to end at (not inclusive)
      previousNumberOfColumns - The number of columns contained in the dictionary.
      Returns:
      A dictionary containing the sliced out columns values only.

    • containsValue

      boolean containsValue(double pattern)
      Detect if the dictionary contains a specific value.
      Parameters:
      pattern - The value to search for
      Returns:
      true if the value is contained else false.

    • containsValueWithReference

      boolean containsValueWithReference(double pattern, double[] reference)
      Detect if the dictionary contains a specific value with reference offset.
      Parameters:
      pattern - The pattern/ value to search for
      reference - The reference double array.
      Returns:
      true if the value is contained else false.

    • getNumberNonZeros

      long getNumberNonZeros(int[] counts, int nCol)
      Calculate the number of non zeros in the dictionary. The number of non zeros should be scaled with the counts given. This gives the exact number of non zero values in the parent column group.
      Parameters:
      counts - The counts of each dictionary entry
      nCol - The number of columns in this dictionary
      Returns:
      The nonZero count

    • getNumberNonZerosWithReference

      long getNumberNonZerosWithReference(int[] counts, double[] reference, int nRows)
      Calculate the number of non zeros in the dictionary. Each value in the dictionary should be added to the reference value. The number of non zeros should be scaled with the given counts.
      Parameters:
      counts - The Counts of each dict entry.
      reference - The reference vector.
      nRows - The number of rows in the input.
      Returns:
      The NonZero Count.

    • addToEntry

      void addToEntry(double[] v, int fr, int to, int nCol)
      Adds the dictionary entry from this dictionary to the d dictionary
      Parameters:
      v - The target dictionary (dense double array)
      fr - The from index is the tuple index to copy from.
      to - The to index is the row index to copy into.
      nCol - The number of columns in both cases

    • addToEntry

      void addToEntry(double[] v, int fr, int to, int nCol, int rep)
      Adds the dictionary entry from this dictionary to the v dictionary rep times.
      Parameters:
      v - The target dictionary (dense double array)
      fr - The from index is the tuple index to copy from.
      to - The to index is the row index to copy into.
      nCol - The number of columns in both cases
      rep - The number of repetitions to apply (simply multiply do not loop)

    • addToEntryVectorized

      void addToEntryVectorized(double[] v, int f1, int f2, int f3, int f4, int f5, int f6, int f7, int f8, int t1, int t2, int t3, int t4, int t5, int t6, int t7, int t8, int nCol)
      Vectorized add to entry, this call helps with a bit of locality for the cache.
      Parameters:
      v - The target dictionary (dense double array)
      f1 - From index 1
      f2 - From index 2
      f3 - From index 3
      f4 - From index 4
      f5 - From index 5
      f6 - From index 6
      f7 - From index 7
      f8 - From index 8
      t1 - To index 1
      t2 - To index 2
      t3 - To index 3
      t4 - To index 4
      t5 - To index 5
      t6 - To index 6
      t7 - To index 7
      t8 - To index 8
      nCol - Number of columns in the dictionary

    • subtractTuple

      IDictionary subtractTuple(double[] tuple)
      Allocate a new dictionary where the tuple given is subtracted from all tuples in the previous dictionary.
      Parameters:
      tuple - a double list representing a tuple, it is given that the tuple with is the same as this dictionaries.
      Returns:
      a new instance of dictionary with the tuple subtracted.

    • getMBDict

      MatrixBlockDictionary getMBDict(int nCol)
      Get this dictionary as a MatrixBlock dictionary. This allows us to use optimized kernels coded elsewhere in the system, such as matrix multiplication. Return null if the matrix is empty.
      Parameters:
      nCol - The number of columns contained in this column group.
      Returns:
      A Dictionary containing a MatrixBlock.

    • scaleTuples

      IDictionary scaleTuples(int[] scaling, int nCol)
      Scale all tuples contained in the dictionary by the scaling factor given in the int list.
      Parameters:
      scaling - The amount to multiply the given tuples with
      nCol - The number of columns contained in this column group.
      Returns:
      A New dictionary (since we don't want to modify the underlying dictionary)

    • preaggValuesFromDense

      IDictionary preaggValuesFromDense(int numVals, IColIndex colIndexes, IColIndex aggregateColumns, double[] b, int cut)
      Pre Aggregate values for Right Matrix Multiplication.
      Parameters:
      numVals - The number of values contained in this dictionary
      colIndexes - The column indexes that is associated with the parent column group
      aggregateColumns - The column to aggregate, this is preprocessed, to find remove consideration for empty columns
      b - The values in the right hand side matrix
      cut - The number of columns in b.
      Returns:
      A new dictionary with the pre aggregated values.

    • replace

      IDictionary replace(double pattern, double replace, int nCol)
      Make a copy of the values, and replace all values that match pattern with replacement value. If needed add a new column index.
      Parameters:
      pattern - The value to look for
      replace - The value to replace the other value with
      nCol - The number of columns contained in the dictionary.
      Returns:
      A new Column Group, reusing the index structure but with new values.

    • replaceWithReference

      IDictionary replaceWithReference(double pattern, double replace, double[] reference)
      Make a copy of the values, and replace all values that match pattern with replacement value. If needed add a new column index. With reference such that each value in the dict is considered offset by the values contained in the reference.
      Parameters:
      pattern - The value to look for
      replace - The value to replace the other value with
      reference - The reference tuple to add to all entries when replacing
      Returns:
      A new Column Group, reusing the index structure but with new values.

    • product

      void product(double[] ret, int[] counts, int nCol)
      Calculate the product of the dictionary weighted by counts.
      Parameters:
      ret - The result dense double array (containing one value)
      counts - The count of individual tuples
      nCol - Number of columns in the dictionary.

    • productWithDefault

      void productWithDefault(double[] ret, int[] counts, double[] def, int defCount)
      Calculate the product of the dictionary weighted by counts with a default value added .
      Parameters:
      ret - The result dense double array (containing one value)
      counts - The count of individual tuples
      def - The default tuple
      defCount - The count of the default tuple

    • productWithReference

      void productWithReference(double[] ret, int[] counts, double[] reference, int refCount)
      Calculate the product of the dictionary weighted by counts and offset by reference
      Parameters:
      ret - The result dense double array (containing one value)
      counts - The counts of each entry in the dictionary
      reference - The reference value.
      refCount - The number of occurrences of the ref value.

    • colProduct

      void colProduct(double[] res, int[] counts, IColIndex colIndexes)
      Calculate the column product of the dictionary weighted by counts.
      Parameters:
      res - The result vector to put the result into
      counts - The weighted count of individual tuples
      colIndexes - The column indexes.

    • colProductWithReference

      void colProductWithReference(double[] res, int[] counts, IColIndex colIndexes, double[] reference)
      Calculate the column product of the dictionary weighted by counts.
      Parameters:
      res - The result vector to put the result into
      counts - The weighted count of individual tuples
      colIndexes - The column indexes.
      reference - The reference value.

    • centralMoment

      CmCovObject centralMoment(ValueFunction fn, int[] counts, int nRows)
      Central moment function to calculate the central moment of this column group. MUST be on a single column dictionary.
      Parameters:
      fn - The value function to apply
      counts - The weight of individual tuples
      nRows - The number of rows in total of the column group
      Returns:
      The central moment Object

    • centralMoment

      CmCovObject centralMoment(CmCovObject ret, ValueFunction fn, int[] counts, int nRows)
      Central moment function to calculate the central moment of this column group. MUST be on a single column dictionary.
      Parameters:
      ret - The Central Moment object to be modified and returned
      fn - The value function to apply
      counts - The weight of individual tuples
      nRows - The number of rows in total of the column group
      Returns:
      The central moment Object

    • centralMomentWithDefault

      CmCovObject centralMomentWithDefault(ValueFunction fn, int[] counts, double def, int nRows)
      Central moment function to calculate the central moment of this column group with a default offset on all missing tuples. MUST be on a single column dictionary.
      Parameters:
      fn - The value function to apply
      counts - The weight of individual tuples
      def - The default values to offset the tuples with
      nRows - The number of rows in total of the column group
      Returns:
      The central moment Object

    • centralMomentWithDefault

      CmCovObject centralMomentWithDefault(CmCovObject ret, ValueFunction fn, int[] counts, double def, int nRows)
      Central moment function to calculate the central moment of this column group with a default offset on all missing tuples. MUST be on a single column dictionary.
      Parameters:
      ret - The Central Moment object to be modified and returned
      fn - The value function to apply
      counts - The weight of individual tuples
      def - The default values to offset the tuples with
      nRows - The number of rows in total of the column group
      Returns:
      The central moment Object

    • centralMomentWithReference

      CmCovObject centralMomentWithReference(ValueFunction fn, int[] counts, double reference, int nRows)
      Central moment function to calculate the central moment of this column group with a reference offset on each tuple. MUST be on a single column dictionary.
      Parameters:
      fn - The value function to apply
      counts - The weight of individual tuples
      reference - The reference values to offset the tuples with
      nRows - The number of rows in total of the column group
      Returns:
      The central moment Object

    • centralMomentWithReference

      CmCovObject centralMomentWithReference(CmCovObject ret, ValueFunction fn, int[] counts, double reference, int nRows)
      Central moment function to calculate the central moment of this column group with a reference offset on each tuple. MUST be on a single column dictionary.
      Parameters:
      ret - The Central Moment object to be modified and returned
      fn - The value function to apply
      counts - The weight of individual tuples
      reference - The reference values to offset the tuples with
      nRows - The number of rows in total of the column group
      Returns:
      The central moment Object

    • rexpandCols

      IDictionary rexpandCols(int max, boolean ignore, boolean cast, int nCol)
      Rexpand the dictionary (one hot encode)
      Parameters:
      max - the tuple width of the output
      ignore - If we should ignore zero and negative values
      cast - If we should cast all double values to whole integer values
      nCol - The number of columns in the dictionary already (should be 1)
      Returns:
      A new dictionary

    • rexpandColsWithReference

      IDictionary rexpandColsWithReference(int max, boolean ignore, boolean cast, int reference)
      Rexpand the dictionary (one hot encode)
      Parameters:
      max - the tuple width of the output
      ignore - If we should ignore zero and negative values
      cast - If we should cast all double values to whole integer values
      reference - A reference value to add to all tuples before expanding
      Returns:
      A new dictionary

    • getSparsity

      double getSparsity()
      Get the sparsity of the dictionary.
      Returns:
      a sparsity between 0 and 1

    • multiplyScalar

      void multiplyScalar(double v, double[] ret, int off, int dictIdx, IColIndex cols)
      Multiply the v value with the dictionary entry at dictIdx and add it to the ret matrix at the columns specified in the int array.
      Parameters:
      v - Value to multiply
      ret - Output dense double array location
      off - Offset into the ret array that the "row" output starts at
      dictIdx - The dictionary entry to multiply.
      cols - The columns to multiply into of the output.

    • TSMMWithScaling

      void TSMMWithScaling(int[] counts, IColIndex rows, IColIndex cols, MatrixBlock ret)
      Transpose self matrix multiplication with a scaling factor on each pair of values.
      Parameters:
      counts - The scaling factor
      rows - The row indexes
      cols - The col indexes
      ret - The output matrix block

    • MMDict

      void MMDict(IDictionary right, IColIndex rowsLeft, IColIndex colsRight, MatrixBlock result)
      Matrix multiplication of dictionaries note the left is this, and it is supposed to be treated as if it is transposed while it is not physically transposed.
      Parameters:
      right - Right hand side of multiplication
      rowsLeft - Offset rows on the left
      colsRight - Offset cols on the right
      result - The output matrix block

    • MMDictScaling

      void MMDictScaling(IDictionary right, IColIndex rowsLeft, IColIndex colsRight, MatrixBlock result, int[] scaling)
      Matrix multiplication of dictionaries, note the left is this, and it is supposed to be treated as if it is transposed while it is not physically transposed.
      Parameters:
      right - Right hand side of multiplication (not transposed)
      rowsLeft - Offset rows on the left
      colsRight - Offset cols on the right
      result - The output matrix block
      scaling - The scaling

    • MMDictDense

      void MMDictDense(double[] left, IColIndex rowsLeft, IColIndex colsRight, MatrixBlock result)
      Matrix multiplication of dictionaries left side dense and transposed. The right side is this.
      Parameters:
      left - Dense left side (treat as if it is transposed but it is physically not)
      rowsLeft - Offset rows on the left
      colsRight - Offset cols on the right
      result - The output matrix block

    • MMDictScalingDense

      void MMDictScalingDense(double[] left, IColIndex rowsLeft, IColIndex colsRight, MatrixBlock result, int[] scaling)
      Matrix multiplication of dictionaries left side dense and transposed. The Right side is this, the scaling factor is used to multiply each element with.
      Parameters:
      left - Dense left side (Dense dictionary)
      rowsLeft - Offset rows on the left (That dictionaries column indexes)
      colsRight - Offset cols on the right (This dictionaries column indexes)
      result - The output matrix block, guaranteed to be allocated as dense.
      scaling - The scaling factor to multiply each entry with.

    • MMDictSparse

      void MMDictSparse(SparseBlock left, IColIndex rowsLeft, IColIndex colsRight, MatrixBlock result)
      Matrix multiplication of dictionaries left side sparse and transposed right side is this.
      Parameters:
      left - Sparse left side
      rowsLeft - Offset rows on the left
      colsRight - Offset cols on the right
      result - The output matrix block

    • MMDictScalingSparse

      void MMDictScalingSparse(SparseBlock left, IColIndex rowsLeft, IColIndex colsRight, MatrixBlock result, int[] scaling)
      Matrix multiplication of dictionaries left side sparse and transposed right side is this.
      Parameters:
      left - Sparse left side
      rowsLeft - Offset rows on the left
      colsRight - Offset cols on the right
      result - The output matrix block
      scaling - The scaling

    • TSMMToUpperTriangle

      void TSMMToUpperTriangle(IDictionary right, IColIndex rowsLeft, IColIndex colsRight, MatrixBlock result)
      Matrix multiplication but allocate output in upper triangle and twice if on diagonal, note this is left
      Parameters:
      right - Right side
      rowsLeft - Offset rows on the left
      colsRight - Offset cols on the right
      result - The output matrix block

    • TSMMToUpperTriangleDense

      void TSMMToUpperTriangleDense(double[] left, IColIndex rowsLeft, IColIndex colsRight, MatrixBlock result)
      Matrix multiplication but allocate output in upper triangle and twice if on diagonal, note this is right
      Parameters:
      left - Dense left side
      rowsLeft - Offset rows on the left
      colsRight - Offset cols on the right
      result - The output matrix block

    • TSMMToUpperTriangleSparse

      void TSMMToUpperTriangleSparse(SparseBlock left, IColIndex rowsLeft, IColIndex colsRight, MatrixBlock result)
      Matrix multiplication but allocate output in upper triangle and twice if on diagonal, note this is right
      Parameters:
      left - Sparse left side
      rowsLeft - Offset rows on the left
      colsRight - Offset cols on the right
      result - The output matrix block

    • TSMMToUpperTriangleScaling

      void TSMMToUpperTriangleScaling(IDictionary right, IColIndex rowsLeft, IColIndex colsRight, int[] scale, MatrixBlock result)
      Matrix multiplication but allocate output in upper triangle and twice if on diagonal, note this is left
      Parameters:
      right - Right side
      rowsLeft - Offset rows on the left
      colsRight - Offset cols on the right
      scale - Scale factor
      result - The output matrix block

    • TSMMToUpperTriangleDenseScaling

      void TSMMToUpperTriangleDenseScaling(double[] left, IColIndex rowsLeft, IColIndex colsRight, int[] scale, MatrixBlock result)
      Matrix multiplication but allocate output in upper triangle and twice if on diagonal, note this is right
      Parameters:
      left - Dense left side
      rowsLeft - Offset rows on the left
      colsRight - Offset cols on the right
      scale - Scale factor
      result - The output matrix block

    • TSMMToUpperTriangleSparseScaling

      void TSMMToUpperTriangleSparseScaling(SparseBlock left, IColIndex rowsLeft, IColIndex colsRight, int[] scale, MatrixBlock result)
      Matrix multiplication but allocate output in upper triangle and twice if on diagonal, note this is right
      Parameters:
      left - Sparse left side
      rowsLeft - Offset rows on the left
      colsRight - Offset cols on the right
      scale - Scale factor
      result - The output matrix block

    • cbind

      IDictionary cbind(IDictionary that, int nCol)
      Cbind this dictionary with that dictionary
      Parameters:
      that - the right hand side dictionary to cbind
      nCol - the right hand side number of columns
      Returns:
      The combined dictionary

    • equals

      boolean equals(Object o)
      Overrides:
      equals in class Object

    • equals

      boolean equals(double[] v)
      Indicate if this object is equal to the given array of doubles.
      Parameters:
      v - The list of double values
      Returns:
      If they are equal to this.

    • equals

      boolean equals(IDictionary o)
      Indicate if the other dictionary is equal to this.
      Parameters:
      o - The other object
      Returns:
      If it is equal

    • reorder

      IDictionary reorder(int[] reorder)
      Reorder the elements in the dictionary based on the reorder specification given.
      Parameters:
      reorder - The order to move to.
      Returns:
      A new Dictionary that is reordered.s

    • rightMMPreAggSparse

      IDictionary rightMMPreAggSparse(int numVals, SparseBlock b, IColIndex thisCols, IColIndex aggregateColumns, int nColRight)
      Pre-aggregate the given sparse block for right multiplication. The returned dictionary is the new column groups dictionary.
      Parameters:
      numVals - The number of values in this dictionary and in the output dictionary.
      b - The sparse block to pre aggregate, note this contains the entire right side matrix, not a reduced or sliced version.
      thisCols - The column indexes of this dictionary, these correspond to the rows to extract from the right side matrix
      aggregateColumns - The reduced column indexes of the right side, these are the number of columns in the output dictionary, and the columns to multiply this dictionary with.
      nColRight - The number of columns in the b sparse matrix.
      Returns:
      The pre-aggregate dictionary that can be used as the output dictionary for the right matrix multiplication

    • putSparse

      void putSparse(SparseBlock sb, int idx, int rowOut, int nCol, IColIndex columns)
      Put the row specified into the sparse block, via append calls.
      Parameters:
      sb - The sparse block to put into
      idx - The dictionary index to put in.
      rowOut - The row in the sparse block to put it into
      nCol - The number of columns in the dictionary
      columns - The columns to output into.

    • putDense

      void putDense(DenseBlock db, int idx, int rowOut, int nCol, IColIndex columns)
      Put the row specified into the sparse block, via append calls.
      Parameters:
      db - The dense block to put into
      idx - The dictionary index to put in.
      rowOut - The row in the sparse block to put it into
      nCol - The number of columns in the dictionary
      columns - The columns to output into.

    • append

      IDictionary append(double[] row)
      Return a new dictionary with the given row appended. If possible reuse as much of the old dictionary as possible.
      Parameters:
      row - The new row to append.
      Returns:
      A new dictionary with the appended row

    • getRow

      double[] getRow(int i, int nCol)
      Extract the values on a given row as a dense double array.
      Parameters:
      i - The row index to extract
      nCol - The number of columns in this column group
      Returns:
      The row extracted

    • countNNZZeroColumns

      int[] countNNZZeroColumns(int[] counts)
      Count the number of non zero values in each column of the dictionary, multiplied with the counts
      Parameters:
      counts - The counts to multiply with.
      Returns:
      The nonzero count of each column in the dictionary.

    • sliceColumns

      IDictionary sliceColumns(IntArrayList selectedColumns, int nCol)
      Slice out the selected columns given of this encoded group.
      Parameters:
      selectedColumns - The columns to slice out and return as a new matrix.
      nCol - The number of columns in this dictionary.
      Returns:
      The returned matrix

    • sort

      int[] sort()
      Sort the values of this dictionary via an index of how the values mapped previously. In practice this design means we can reuse the previous dictionary for the resulting column group
      Returns:
      The sorted index.