Many resources are needed to download a project. Please understand that we have to compensate our server costs. Thank you in advance.
Project price only 1 $
You can buy this project and download/modify it how often you want.
scalismo.sampling.proposals.MHProductProposal.scala Maven / Gradle / Ivy
package scalismo.sampling.proposals
import scalismo.sampling.{MHProposalGenerator, MHSample, ProposalGenerator, TransitionProbability}
object MHProductProposal {
def apply[A, B](a: MHProposalGenerator[A], b: MHProposalGenerator[B]): MHProductProposal2[A, B] =
new MHProductProposal2(a, b)
def apply[A, B, C](a: MHProposalGenerator[A],
b: MHProposalGenerator[B],
c: MHProposalGenerator[C]): MHProductProposal3[A, B, C] = new MHProductProposal3(a, b, c)
def apply[A, B, C, D](a: MHProposalGenerator[A],
b: MHProposalGenerator[B],
c: MHProposalGenerator[C],
d: MHProposalGenerator[D]): MHProductProposal4[A, B, C, D] = new MHProductProposal4(a, b, c, d)
}
class MHProductProposal2[A, B](proposal1: MHProposalGenerator[A], proposal2: MHProposalGenerator[B])
extends MHProposalGenerator[(A, B)] {
self =>
/** rate of transition from to (log value) */
override def logTransitionProbability(from: MHSample[(A, B)], to: MHSample[(A, B)]): Double = {
val transitionProbability1 = proposal1.logTransitionProbability(MHSample(from.parameters._1, from.generatedBy),
MHSample(to.parameters._1, to.generatedBy))
val transitionProbability2 = proposal2.logTransitionProbability(MHSample(from.parameters._2, from.generatedBy),
MHSample(to.parameters._2, to.generatedBy))
if (transitionProbability1.isInfinite || transitionProbability2.isInfinite) {
Double.NegativeInfinity
} else {
transitionProbability1 + transitionProbability2
}
}
/** draw a sample from this proposal distribution, may depend on current state */
override def propose(current: MHSample[(A, B)]): MHSample[(A, B)] = {
val sample1 = proposal1.propose(current.copy(parameters = current.parameters._1))
val sample2 = proposal2.propose(current.copy(parameters = current.parameters._2))
MHSample(
(sample1.parameters, sample2.parameters),
generatedBy = s"(${sample1.generatedBy}, ${sample2.generatedBy})"
)
}
}
class MHProductProposal3[A, B, C](proposal1: MHProposalGenerator[A],
proposal2: MHProposalGenerator[B],
proposal3: MHProposalGenerator[C])
extends MHProposalGenerator[(A, B, C)] {
self =>
/** rate of transition from to (log value) */
override def logTransitionProbability(from: MHSample[(A, B, C)], to: MHSample[(A, B, C)]): Double = {
val transitionProbability1 = proposal1.logTransitionProbability(MHSample(from.parameters._1, from.generatedBy),
MHSample(to.parameters._1, to.generatedBy))
val transitionProbability2 = proposal2.logTransitionProbability(MHSample(from.parameters._2, from.generatedBy),
MHSample(to.parameters._2, to.generatedBy))
val transitionProbability3 = proposal3.logTransitionProbability(MHSample(from.parameters._3, from.generatedBy),
MHSample(to.parameters._3, to.generatedBy))
if (transitionProbability1.isInfinite || transitionProbability2.isInfinite || transitionProbability3.isInfinite) {
Double.NegativeInfinity
} else {
transitionProbability1 + transitionProbability2 + transitionProbability3
}
}
/** draw a sample from this proposal distribution, may depend on current state */
override def propose(current: MHSample[(A, B, C)]): MHSample[(A, B, C)] = {
val sample1 = proposal1.propose(current.copy(parameters = current.parameters._1))
val sample2 = proposal2.propose(current.copy(parameters = current.parameters._2))
val sample3 = proposal3.propose(current.copy(parameters = current.parameters._3))
MHSample(
(sample1.parameters, sample2.parameters, sample3.parameters),
generatedBy = s"(${sample1.generatedBy}, ${sample2.generatedBy}, ${sample3.generatedBy})"
)
}
}
class MHProductProposal4[A, B, C, D](proposal1: MHProposalGenerator[A],
proposal2: MHProposalGenerator[B],
proposal3: MHProposalGenerator[C],
proposal4: MHProposalGenerator[D])
extends MHProposalGenerator[(A, B, C, D)] {
self =>
/** rate of transition from to (log value) */
override def logTransitionProbability(from: MHSample[(A, B, C, D)], to: MHSample[(A, B, C, D)]): Double = {
val transitionProbability1 = proposal1.logTransitionProbability(MHSample(from.parameters._1, from.generatedBy),
MHSample(to.parameters._1, to.generatedBy))
val transitionProbability2 = proposal2.logTransitionProbability(MHSample(from.parameters._2, from.generatedBy),
MHSample(to.parameters._2, to.generatedBy))
val transitionProbability3 = proposal3.logTransitionProbability(MHSample(from.parameters._3, from.generatedBy),
MHSample(to.parameters._3, to.generatedBy))
val transitionProbability4 = proposal4.logTransitionProbability(MHSample(from.parameters._4, from.generatedBy),
MHSample(to.parameters._4, to.generatedBy))
if (transitionProbability1.isInfinite || transitionProbability2.isInfinite || transitionProbability3.isInfinite || transitionProbability4.isInfinite) {
Double.NegativeInfinity
} else {
transitionProbability1 + transitionProbability2 + transitionProbability3 + transitionProbability4
}
}
/** draw a sample from this proposal distribution, may depend on current state */
override def propose(current: MHSample[(A, B, C, D)]): MHSample[(A, B, C, D)] = {
val sample1 = proposal1.propose(current.copy(parameters = current.parameters._1))
val sample2 = proposal2.propose(current.copy(parameters = current.parameters._2))
val sample3 = proposal3.propose(current.copy(parameters = current.parameters._3))
val sample4 = proposal4.propose(current.copy(parameters = current.parameters._4))
MHSample(
(sample1.parameters, sample2.parameters, sample3.parameters, sample4.parameters),
generatedBy = s"(${sample1.generatedBy}, ${sample2.generatedBy}, ${sample3.generatedBy}, ${sample4.generatedBy})"
)
}
}
