FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W174902875> ?p ?o ?g. }

• W174902875 abstract "Maximum likelihood (ML) estimates of the parameters of SDEs are consistent and asymptotically efficient, but unfortunately difficult to obtain if a closed form expression for the transitional probability density function (PDF) of the process is not available. One popular way to obtain the transitional PDF is to solve the Fokker-Planck equation numerically. However the treatment of the delta function initial condition and zero-flux boundary conditions, both of which are necessary to implement these numerical schemes, is not straightforward. By reformulating the problem in terms of the transitional cumulative distribution function (CDF), it is shown that these conditions are handled easily. The transitional PDF is subsequently computed by numerical differentiation of the transitional CDF and used to construct a likelihood function in the usual way. Consider the general one-dimensional, timehomogeneous stochastic differential equation (SDE) dX = μ(X; θ) dt + √ g(X; θ) dW where X is a stochastic Markov process, μ(x; θ) and g(x;θ) are respectively the instantaneous drift and instantaneous diffusion of X , dW is the differential of the Wiener process and θ is a vector of parameters to be estimated. The aim of ML estimation is to minimise the negative log-likelihood function with respect to the parameter vector θ. ML estimation relies on the fact that the transitional density of X at time t is the solution of the Fokker-Planck equation ∂f ∂t = ∂ ∂x [ 1 2 ∂(g(x; θ)f) ∂x − μ(x; θ)f ] satisfying a delta function initial condition and gradient-like boundary conditions. This paper is concerned with an equivalent statement of this problem in terms of the transitional CDF, F (x, t) , which is defined in terms of the transitional PDF, f(x, t), by F (x, t) = ∫ x f(u, t) du . When expressed in terms of F (x, t), the FokkerPlanck equation takes the form ∂F ∂t = 1 2 ∂ ∂x ( g ∂F ∂x ) − μ ∂x with a step function initial condition and Dirichlet boundary conditions. Both the PDF and CDF approaches to the solution of the Fokker-Planck equation are implemented using a finite-difference method. The latter is easier to implement than the former, because the initial condition is more amenable to numerical work and the gradient-like boundary conditions associated with the Fokker-Planck equation are replaced by Dirichlet boundary conditions in the modification. The efficacy of the numerical solution is evaluated by means of two Monte Carlo exercises based on simulating the CIR equation dX = α(β −X)dt + σ √ X dW with α = 0.2, β = 0.08 and σ = 0.1, using Milstein’s scheme with 1000 time steps of size 0.001 between observations. The first experiment compares the PDF and CDF approaches in terms of the accuracy of loglikelihood computation, while the second simulation experiment involves the estimation of the parameters of the underlying CIR model. The most significant finding is that the CDF approach using the step function initial condition can be implemented successfully in practice. By contrast, there is no equivalent result for the PDFbased procedure, because it is always necessary to approximate the initial condition. The main empirical result to emerge from the simulation exercises in this paper is that, given equivalent starting information, the CDF approach is always superior to the PDF approach in terms of the accuracy of likelihood evaluation." @default.
• W174902875 created "2016-06-24" @default.
• W174902875 creator A5012772168 @default.
• W174902875 creator A5027906794 @default.
• W174902875 creator A5087781489 @default.
• W174902875 date "2005-01-01" @default.
• W174902875 modified "2023-09-26" @default.
• W174902875 title "ML Estimation Of The Parameters Of SDE's By Numerical Solution Of The Fokker-Planck Equation" @default.
• W174902875 cites W1994908910 @default.
• W174902875 cites W2076057077 @default.
• W174902875 cites W3021444882 @default.
• W174902875 cites W3123173808 @default.
• W174902875 hasPublicationYear "2005" @default.
• W174902875 type Work @default.
• W174902875 sameAs 174902875 @default.
• W174902875 citedByCount "0" @default.
• W174902875 crossrefType "journal-article" @default.
• W174902875 hasAuthorship W174902875A5012772168 @default.
• W174902875 hasAuthorship W174902875A5027906794 @default.
• W174902875 hasAuthorship W174902875A5087781489 @default.
• W174902875 hasConcept C105795698 @default.
• W174902875 hasConcept C134306372 @default.
• W174902875 hasConcept C14036430 @default.
• W174902875 hasConcept C197055811 @default.
• W174902875 hasConcept C28826006 @default.
• W174902875 hasConcept C33923547 @default.
• W174902875 hasConcept C51955184 @default.
• W174902875 hasConcept C62354387 @default.
• W174902875 hasConcept C69123182 @default.
• W174902875 hasConcept C78045399 @default.
• W174902875 hasConcept C78458016 @default.
• W174902875 hasConcept C86803240 @default.
• W174902875 hasConceptScore W174902875C105795698 @default.
• W174902875 hasConceptScore W174902875C134306372 @default.
• W174902875 hasConceptScore W174902875C14036430 @default.
• W174902875 hasConceptScore W174902875C197055811 @default.
• W174902875 hasConceptScore W174902875C28826006 @default.
• W174902875 hasConceptScore W174902875C33923547 @default.
• W174902875 hasConceptScore W174902875C51955184 @default.
• W174902875 hasConceptScore W174902875C62354387 @default.
• W174902875 hasConceptScore W174902875C69123182 @default.
• W174902875 hasConceptScore W174902875C78045399 @default.
• W174902875 hasConceptScore W174902875C78458016 @default.
• W174902875 hasConceptScore W174902875C86803240 @default.
• W174902875 hasLocation W1749028751 @default.
• W174902875 hasOpenAccess W174902875 @default.
• W174902875 hasPrimaryLocation W1749028751 @default.
• W174902875 hasRelatedWork W128525024 @default.
• W174902875 hasRelatedWork W141383440 @default.
• W174902875 hasRelatedWork W1523816691 @default.
• W174902875 hasRelatedWork W1553760566 @default.
• W174902875 hasRelatedWork W1850530143 @default.
• W174902875 hasRelatedWork W1901424371 @default.
• W174902875 hasRelatedWork W1977770003 @default.
• W174902875 hasRelatedWork W1992794596 @default.
• W174902875 hasRelatedWork W2013209425 @default.
• W174902875 hasRelatedWork W2045124575 @default.
• W174902875 hasRelatedWork W2052203633 @default.
• W174902875 hasRelatedWork W2077254975 @default.
• W174902875 hasRelatedWork W2129189718 @default.
• W174902875 hasRelatedWork W2170193238 @default.
• W174902875 hasRelatedWork W2326306694 @default.
• W174902875 hasRelatedWork W2371917858 @default.
• W174902875 hasRelatedWork W2390366302 @default.
• W174902875 hasRelatedWork W2552630144 @default.
• W174902875 hasRelatedWork W291318382 @default.
• W174902875 hasRelatedWork W71082738 @default.
• W174902875 isParatext "false" @default.
• W174902875 isRetracted "false" @default.
• W174902875 magId "174902875" @default.
• W174902875 workType "article" @default.