Namespace Microsoft.ML.Probabilistic.Math
Classes
AllZeroException
Exception type thrown when probability vector = (0,0,0,...,0).
ApproximateSparseVector
A one-dimensional vector of double values, optimised for the case where many of the elements share a common value (which need not be zero) within some tolerance.
BFGS
This implementation of BFGS is based on Algorithm 6.1 from Nocedal and Wright (Second edition, 2006).
BranchAndBound
ContinuedFraction
A class for evaluating continued fractions
DenseVector
1-dimensional dense container of double precision data that supports vector operations.
DerivativeChecker
Class used to check analytic derivatives using finite difference approximation
ExtendedDouble
Represents a number as Mantissa * exp(Exponent).
LBFGS
Implements the LBFGS compact Quasi-Newton solver.
LBFGSArray
Implements the LBFGS compact Quasi-Newton solver on an array of Vectors (which may be sparse)
LineSearch
This line search algorithm is algorithm 3.5/3.6 from Nocedal and Wright (Second edition, 2006). It provides a step length that satisfies the strong Wolfe conditions.
LowerTriangularMatrix
Class for lower triangular matrices
LuDecomposition
Class for calculating and doing operations with an LU decomposition
Matrix
Two-dimensional container of doubles.
MatrixMeanVarianceAccumulator
Class for accumulating weighted noisy matrix observations, and computing sample count, mean matrix, and covariance matrix
MatrixSingularException
Exception thrown when a singular matrix is encountered.
MeanAccumulator
Class for accumulating weighted scalar observations and computing sample count and mean
MeanVarianceAccumulator
Class for accumulating weighted noisy scalar observations, and computing sample count, mean, and variance
MeanVarianceAccumulator2
Class for accumulating weighted noisy scalar observations, and computing sample count, mean, and variance
MeanVarianceAccumulatorSkipNaNs
Decorator of MeanVarianceAccumulator that does not add if any input is NaN.
MMath
This class provides mathematical constants and special functions, analogous to System.Math. It cannot be instantiated and consists of static members only.
OptimiserIterationEventArgs
Optimiser iteration event
PiecewiseVector
A one-dimensional vector of double values, optimised for the case where many contiguous ranges of elements have the same value.
PositiveDefiniteMatrix
A subclass of Matrix with extra methods appropriate to positive-definite matrices.
PositiveDefiniteMatrixException
Exception thrown when a matrix is not positive definite.
Quadrature
Quadrature nodes and weights
Rand
This class provides a source of non-uniform random numbers. It cannot be instantiated and consists of only static functions.
Region
Represents a hyper-rectangle in arbitrary dimensions.
SparseVector
A one-dimensional vector of double values, optimised for the case where many of the elements share a common value (which need not be zero).
Sparsity
Defines sparsity settings for vectors. The sparsity handling has been designed to deal with large dimensional distributions such as Discrete and Dirichlet distributions.
UpperTriangularMatrix
Upper triangular matrix class
Vector
Base class for vectors. DenseVector, SparseVector, and ApproximateSparseVector all inherit from this base class.
VectorMeanVarianceAccumulator
Class for accumulating weighted noisy vector observations, and computing sample count, mean vector, and covariance matrix
Structs
ConstantVector
A vector which has a constant value between its start and end indices.
Interfaces
CanSetAllElementsTo<T>
Supports setting all elements to duplicates of the same value
Diffable
Supports calculating the maximum difference between this instance and another object (not necessarily of the same type)
SettableTo<T>
Supports setting an instance to a value
SettableToPower<T>
Supports setting an instance to a value raised to a power
SettableToProduct<T>
Supports setting an instance to the product of two values of the same type
SettableToProduct<T, U>
Supports setting an instance to the product of two values of different types
SettableToRatio<T>
Supports setting an instance to the ratio of two values of the same type
SettableToRatio<T, U>
Supports setting an instance to the ratio of two values of different types
SettableToWeightedSum<T>
Supports setting an instance to the weighted sum of two values of the same type
SettableToWeightedSumExact<T>
Indicates that a distribution can represent weighted sum
of distributions of type T exactly.
Enums
BFGS.ConvergenceCriteria
Convergence criteria:
- Gradient: |grad F|/sqrt(dimensions) <= eps
- Objective: |F(k+1)-F(k)|<=eps*max{|F(k)|,|F(k+1)|,1}
StorageType
The type of storage used in a vector, which is specified as part of the Sparsity class.
Delegates
FunctionEval
Delegate type for function evaluation
IterationEventHandler
Event delegate for handling iteration event
LBFGSArray.FunctionEvalArray
Delegate type for function evaluation
LineSearchEval
Delegate type for function evaluation
PiecewiseVector.RangeFunc
An action that takes a range and two values.