DynaML v1.5 Release Notes
Release Date: 2017-08-15 // over 6 years ago-
โ Additions
๐ฆ Package
dynaml.algebra
โ Added support for dual numbers.
//Zero Dualval zero = DualNumber.zero[Double] val dnum = DualNumber(1.5, -1.0) val dnum1 = DualNumber(-1.5, 1.0) //Algebraic operations: multiplication and addition/subtractiondnum1\*dnum2 dnum1 - dnum dnum\*zero
๐ฆ Package
dynaml.probability
- โ Added support for mixture distributions and mixture random variables.
MixtureRV
,ContinuousDistrMixture
for random variables andMixtureDistribution
for constructing mixtures of breeze distributions.
๐ฆ Package
dynaml.optimization
- โ Added
ModelTuner[T, T1]
trait as a super trait toGlobalOptimizer[T]
GridSearch
andCoupledSimulatedAnnealing
now extendAbstractGridSearch
andAbstractCSA
respectively.- โ Added
ProbGPMixtureMachine
: constructs a mixture model after a CSA or grid search routine by calculating the mixture probabilities of members of the final hyper-parameter ensemble.
Stochastic Mixture Models
๐ฆ Package
dynaml.models
- โ Added
StochasticProcessMixtureModel
as top level class for stochastic mixture models. - โ Added
GaussianProcessMixture
: implementation of gaussian process
mixture models. - โ Added
MVTMixture
: implementation of mixture model over
multioutput matrix T processes.
Kulback-Leibler Divergence
๐ฆ Package
dynaml.probability
- โ Added method
KL()
toprobability
package object, to calculate
the Kulback Leibler divergence between two continuous random
variables backed by breeze distributions.
Adaptive Metropolis Algorithms.
AdaptiveHyperParameterMCMC which
adapts the exploration covariance with each sample.HyperParameterSCAM adapts
the exploration covariance for each hyper-parameter independently.Splines and B-Spline Generators
๐ฆ Package
dynaml.analysis
- B-Spline generators
- ๐ Bernstein and Cardinal b-spline generators.
- Arbitrary spline functions can be created using the
SplineGenerator
class.
Cubic Spline Interpolation Kernels
๐ฆ Package
dynaml.kernels
- ๐ Added cubic spline interpolation kernel
CubicSplineKernel
and its ARD analogueCubicSplineARDKernel
Gaussian Process Models for Linear Partial Differential Equations
Based on a legacy ICML 2003 paper by Graepel. DynaML now ships with capability of performing PDE forward and inverse inference using the Gaussian Process API.
๐ฆ Package
dynaml.models.gp
GPOperatorModel
: models a quantity of interest which is governed by a linear PDE in space and time.
๐ฆ Package
dynaml.kernels
LinearPDEKernel
: The core kernel primitive accepted by theGPOperatorModel
class.GenExpSpaceTimeKernel
: a kernel of the exponential family which can serve as a handy base kernel forLinearPDEKernel
class.Basis Function Gaussian Processes
๐ DynaML now supports GP models with explicitly incorporated basis
functions as linear mean/trend functions.๐ฆ Package
dynaml.models.gp
GPBasisFuncRegressionModel
can
be used to create GP models with trends incorporated as a linear
combination of basis functions.
๐ฒ Log Gaussian Processes
- LogGaussianProcessModel represents
a stochastic process whose natural logarithm follows a gaussian process.
๐ Improvements
๐ฆ Package
dynaml.probability
- ๐ Changes to
RandomVarWithDistr
: made type parameterDist
covariant. - Reform to
IIDRandomVar
hierarchy.
๐ฆ Package
dynaml.probability.mcmc
- ๐ Bug-fixes to the
HyperParameterMCMC
class.
General
- DynaML now ships with Ammonite
v1.0.0
.
๐ Fixes
๐ฆ Package
dynaml.optimization
- Corrected energy calculation in
CoupledSimulatedAnnealing
; added
๐ฒ log likelihood due to hyper-prior.
๐ฆ Package
dynaml.optimization
- Corrected energy calculation in
CoupledSimulatedAnnealing
; added
๐ฒ log likelihood due to hyper-prior.
- โ Added support for mixture distributions and mixture random variables.