Class TernaryOp

  • All Implemented Interfaces:
    ParseInfo

    public class TernaryOp
    extends MultiThreadedHop
    Primary use cases for now, are quantile (<n-1-matrix>, <n-1-matrix>, <literal>): quantile (A, w, 0.5) quantile (<n-1-matrix>, <n-1-matrix>, <scalar>): quantile (A, w, s) interquantile (<n-1-matrix>, <n-1-matrix>, <scalar>): interquantile (A, w, s) Keep in mind, that we also have binaries for it w/o weights. quantile (A, 0.5) quantile (A, s) interquantile (A, s) Note: this hop should be called AggTernaryOp in consistency with AggUnaryOp and AggBinaryOp; however, since there does not exist a real TernaryOp yet - we can leave it as is for now. CTABLE op takes 2 extra inputs with target dimensions for padding and pruning.
    • Field Detail

      • ALLOW_CTABLE_SEQUENCE_REWRITES

        public static boolean ALLOW_CTABLE_SEQUENCE_REWRITES
    • Method Detail

      • checkArity

        public void checkArity()
        Description copied from class: Hop
        Check whether this Hop has a correct number of inputs. (Some Hops can have a variable number of inputs, such as DataOp, DataGenOp, ParameterizedBuiltinOp, ReorgOp, TernaryOp, QuaternaryOp, MultipleOp, DnnOp, and SpoofFusedOp.) Parameterized Hops (such as DataOp) can check that the number of parameters matches the number of inputs.
        Specified by:
        checkArity in class Hop
      • setDisjointInputs

        public void setDisjointInputs​(boolean flag)
      • isGPUEnabled

        public boolean isGPUEnabled()
        Description copied from class: Hop
        In memory-based optimizer mode (see OptimizerUtils.isMemoryBasedOptLevel()), the exectype is determined by checking this method as well as memory budget of this Hop. Please see findExecTypeByMemEstimate for more detail. This method is necessary because not all operator are supported efficiently on GPU (for example: operations on frames and scalar as well as operations such as table).
        Specified by:
        isGPUEnabled in class Hop
        Returns:
        true if the Hop is eligible for GPU Exectype.
      • allowsAllExecTypes

        public boolean allowsAllExecTypes()
        Specified by:
        allowsAllExecTypes in class Hop
      • refreshSizeInformation

        public void refreshSizeInformation()
        Description copied from class: Hop
        Update the output size information for this hop.
        Specified by:
        refreshSizeInformation in class Hop
      • compare

        public boolean compare​(Hop that)
        Specified by:
        compare in class Hop
      • isMatrixIgnoreZeroRewriteApplicable

        public boolean isMatrixIgnoreZeroRewriteApplicable()
        Used for (1) constructing CP lops (hop-lop rewrite), and (2) in order to determine if dag split after removeEmpty necessary (#2 is precondition for #1).
        Returns:
        true if ignore zero rewrite