model

Type Members

1. case class Branch[A](left: Tree[A], right: Tree[A]) extends Tree[A] with Product with Serializable

   

2. case class BranchParameter(left: Parameters, right: Parameters) extends Parameters with Product with Serializable

   

3. case class BrownianParameter(m0: DenseVector[Double], c0: DenseMatrix[Double], mu: DenseVector[Double], sigma: DenseMatrix[Double]) extends SdeParameter with Product with Serializable

   

4. case class CredibleInterval(lower: Double, upper: Double) extends Product with Serializable

   

   Credible intervals from a set of samples in a distribution

   Credible intervals from a set of samples in a distribution

   lower

   the lower interval

   upper

   the upper interval

5. sealed trait Data extends AnyRef

   

   A single observation of a time series

6. trait DataService extends AnyRef

   

7. type Eta = Seq[Double]

   

8. case class Filter(mod: Model, resamplingScheme: Resample[State]) extends ParticleFilter with Product with Serializable

   

9. case class FilterLgcp(mod: Model, resamplingScheme: Resample[State], precision: Int) extends ParticleFilter with Product with Serializable

   

   In order to calculate Eta in the LGCP model, we need to merge the advance state and transform state functions

10. case class ForecastOut(t: Time, obs: Observation, obsIntervals: CredibleInterval, eta: Double, etaIntervals: CredibleInterval, state: State, stateIntervals: IndexedSeq[CredibleInterval]) extends Product with Serializable

    

    Forecast data

    Forecast data

    t

    the time of the observation

    obs

    an observation of the process

    obsIntervals

    the upper and lower credible intervals of the observation

    eta

    the transformed latent state

    etaIntervals

    the credible intervals of the transformed latent state

    state

    the untransformed latent state

    stateIntervals

    the intervals of the latent state

11. type Gamma = Double

    

12. case class Leaf[A](value: A) extends Tree[A] with Product with Serializable

    

13. case class LeafParameter(scale: Option[Double], sdeParam: SdeParameter) extends Parameters with Product with Serializable

    

14. type LogLikelihood = Double

    

15. case class MetropState(ll: LogLikelihood, params: Parameters, accepted: Int) extends Product with Serializable

    

    The state of the metropolis-hastings algorithms

    The state of the metropolis-hastings algorithms

    ll

    the log-likelihood of the observations given the latent state and the current parameters

    params

    the current set of parameters

    accepted

    the total number of accepted moves in the metropolis hastings algorithm

16. trait MetropolisHastings extends AnyRef

    

17. trait Model extends AnyRef

    

18. type Observation = Double

    

19. case class ObservationWithState(t: Time, observation: Observation, eta: Eta, gamma: Gamma, sdeState: State) extends Data with Product with Serializable

    

    A single observation of a time series, containing a realisation of the filtering state

    A single observation of a time series, containing a realisation of the filtering state

    observation

    pi(eta), the observation

    eta

    g(gamma), the latent state transformed by the linking-function

    gamma

    f(x_t), the latent state transformed by the linear transformation

    sdeState

    x_t

20. case class OrnsteinParameter(m0: DenseVector[Double], c0: DenseMatrix[Double], theta: DenseVector[Double], alpha: DenseVector[Double], sigma: DenseVector[Double]) extends SdeParameter with Product with Serializable

    

21. sealed trait Parameters extends AnyRef

    

22. trait ParticleFilter extends AnyRef

    

23. case class ParticleMetropolis(logLikelihood: (Parameters) ⇒ LogLikelihood, initialParams: Parameters, proposal: (Parameters) ⇒ Rand[Parameters], prior: (Parameters) ⇒ LogLikelihood) extends MetropolisHastings with Product with Serializable

    

    Implementation of the particle metropolis algorithm

    Implementation of the particle metropolis algorithm

    logLikelihood

    a function from parameters to LogLikelihood

    initialParams

    the starting parameters for the metropolis algorithm

    proposal

    a SYMMETRIC proposal distribution for the metropolis algorithm (eg. Gaussian)

24. case class ParticleMetropolisHastings(logLikelihood: (Parameters) ⇒ LogLikelihood, transitionProb: (Parameters, Parameters) ⇒ LogLikelihood, proposal: (Parameters) ⇒ Rand[Parameters], initialParams: Parameters, prior: (Parameters) ⇒ LogLikelihood) extends MetropolisHastings with Product with Serializable

    

    Implementation of the particle metropolis hastings algorithm specified prior distribution

    Implementation of the particle metropolis hastings algorithm specified prior distribution

    logLikelihood

    a function from parameters to LogLikelihood

    proposal

    a generic proposal distribution for the metropolis algorithm (eg. Gaussian)

    initialParams

    the starting parameters for the metropolis algorithm

25. case class PfOut(time: Time, observation: Option[Observation], eta: Double, etaIntervals: CredibleInterval, state: State, stateIntervals: IndexedSeq[CredibleInterval]) extends Product with Serializable

    

    A class representing a return type for the particle filter, containing the state and associated credible intervals

    A class representing a return type for the particle filter, containing the state and associated credible intervals

    time

    the time of the process

    observation

    an optional observation, note discretely observed processes cannot be seen at all time points continuously

    state

    the mean of the empirical filtering distribution at time 'time'

26. case class PfState(t: Time, observation: Option[Observation], particles: Seq[State], weights: Seq[LogLikelihood], ll: LogLikelihood) extends Product with Serializable

    

    Representation of the state of the particle filter, at each step the previous observation time, t0, and particle cloud, particles, is required to compute forward.

    Representation of the state of the particle filter, at each step the previous observation time, t0, and particle cloud, particles, is required to compute forward. The meanState and intervals are recorded in each step, so they can be outputted immediately without having to calculate these from the particle cloud after

27. type Resample[A] = (Seq[A], Seq[LogLikelihood]) ⇒ Seq[A]

    

28. trait Sde extends AnyRef

    

29. sealed trait SdeParameter extends AnyRef

    

30. case class SimulatedData(model: Model) extends DataService with Product with Serializable

    

31. type State = Tree[DenseVector[Double]]

    

32. case class StateSpace(time: Time, state: State) extends Product with Serializable

    

    Representing a realisation from a stochastic differential equation

33. type StepFunction = (SdeParameter) ⇒ (State, TimeIncrement) ⇒ Rand[State]

    

34. type Time = Double

    

35. type TimeIncrement = Double

    

36. case class TimedObservation(t: Time, observation: Observation) extends Data with Product with Serializable

    

    A single observation of a time series

    A single observation of a time series

    observation

    pi(eta), the observation

37. sealed trait Tree[A] extends AnyRef

    

    A binary tree implementation, to be used when combining models Hopefully this simplifies "zooming" into values and changing them

38. type UnparamModel = Kleisli[Id, Parameters, Model]

    

39. type UnparamSde = Kleisli[Id, SdeParameter, Sde]