Programming With Models: Writing Statistical Algorithms for General Model Structures With NIMBLE

Valpine, Perry de; Turek, Daniel; Paciorek, Christopher J.; Anderson-Bergman, Clifford; Lang, Duncan Temple; Bodík, Rastislav

doi:10.1080/10618600.2016.1172487

Cited by 761 publications

(530 citation statements)

References 40 publications

Supporting

Mentioning

525

Contrasting

Unclassified

Order By: Relevance

“…NIMBLE extends the Bayesian inference Using Gibbs Sampling [ Gilks et al ., ] language and is flexible to implement complex model specifications as in our study. We use a Metropolis‐Hastings adaptive random‐walk sampler available in NIMBLE, and the proposal distribution is automatically adapted to a target acceptance rate (e.g., 0.4) depending on the sampler [ de Valpine et al ., ] to ensure optimal MCMC mixing [ Roberts et al ., ]. For VOC:CH 4 ratios (i.e.,

f_{X}^{*}

), for example,

f_{C_{2} H_{6_{N G}}}^{*}

and

f_{C_{2} H_{6_{normalPL}}}^{*}

are the C 2 H 6 :CH 4 emission ratios for the NG and PL sectors, respectively, which are optimized based on the VOC measurements during the inversion.…”

Section: Methodscontrasting

confidence: 50%

See 1 more Smart Citation

Estimating methane emissions from biological and fossil‐fuel sources in the San Francisco Bay Area

Jeong

Cui

Blake

et al. 2017

Geophysical Research Letters

View full text Add to dashboard Cite

We present the first sector‐specific analysis of methane (CH4) emissions from the San Francisco Bay Area (SFBA) using CH4 and volatile organic compound (VOC) measurements from six sites during September – December 2015. We apply a hierarchical Bayesian inversion to separate the biological from fossil‐fuel (natural gas and petroleum) sources using the measurements of CH4 and selected VOCs, a source‐specific 1 km CH4 emission model, and an atmospheric transport model. We estimate that SFBA CH4 emissions are 166–289 Gg CH4/yr (at 95% confidence), 1.3–2.3 times higher than a recent inventory with much of the underestimation from landfill. Including the VOCs, 82 ± 27% of total posterior median CH4 emissions are biological and 17 ± 3% fossil fuel, where landfill and natural gas dominate the biological and fossil‐fuel CH4 of prior emissions, respectively.

show abstract

f_{X}^{*}

), for example,

f_{C_{2} H_{6_{N G}}}^{*}

and

f_{C_{2} H_{6_{normalPL}}}^{*}

are the C 2 H 6 :CH 4 emission ratios for the NG and PL sectors, respectively, which are optimized based on the VOC measurements during the inversion.…”

Section: Methodscontrasting

confidence: 50%

“…To build MCMC samplers (sample size = 20,000), the NIMBLE package (version 0.5; [ de Valpine et al ., ]) is used together with the R statistical language (https://cran.r-project.org/). NIMBLE extends the Bayesian inference Using Gibbs Sampling [ Gilks et al ., ] language and is flexible to implement complex model specifications as in our study.…”

Section: Methodsmentioning

confidence: 99%

Estimating methane emissions from biological and fossil‐fuel sources in the San Francisco Bay Area

Jeong

Cui

Blake

et al. 2017

Geophysical Research Letters

View full text Add to dashboard Cite

show abstract

“…There are also alternative software platforms not tested here, such as nimble (de Valpine et al . ) and ensemble sampling (Goodman & Weare ), and future work comparing these to jags and Stan would be worthwhile. Stan is also not the only platform coupling automatic differentiation and HMC that is used by ecologists.…”

Section: Discussionmentioning

confidence: 99%

Faster estimation of Bayesian models in ecology using Hamiltonian Monte Carlo

2016

View full text Add to dashboard Cite

Summary1. Bayesian inference is a powerful tool to better understand ecological processes across varied subfields in ecology, and is often implemented in generic and flexible software packages such as the widely used BUGS family (BUGS, WinBUGS, OpenBUGS and JAGS). However, some models have prohibitively long run times when implemented in BUGS. A relatively new software platform called Stan uses Hamiltonian Monte Carlo (HMC), a family of Markov chain Monte Carlo (MCMC) algorithms which promise improved efficiency and faster inference relative to those used by BUGS. Stan is gaining traction in many fields as an alternative to BUGS, but adoption has been slow in ecology, likely due in part to the complex nature of HMC. 2. Here, we provide an intuitive illustration of the principles of HMC on a set of simple models. We then compared the relative efficiency of BUGS and Stan using population ecology models that vary in size and complexity. For hierarchical models, we also investigated the effect of an alternative parameterization of random effects, known as non-centering. 3.. For small, simple models there is little practical difference between the two platforms, but Stan outperforms BUGS as model size and complexity grows. Stan also performs well for hierarchical models, but is more sensitive to model parameterization than BUGS. Stan may also be more robust to biased inference caused by pathologies, because it produces diagnostic warnings where BUGS provides none. Disadvantages of Stan include an inability to use discrete parameters, more complex diagnostics and a greater requirement for hands-on tuning. 4. Given these results, Stan is a valuable tool for many ecologists utilizing Bayesian inference, particularly for problems where BUGS is prohibitively slow. As such, Stan can extend the boundaries of feasible models for applied problems, leading to better understanding of ecological processes. Fields that would likely benefit include estimation of individual and population growth rates, meta-analyses and cross-system comparisons and spatiotemporal models.

show abstract

“…The model and MCMC algorithm were written and executed in NIMBLE 6.10 (de Valpine et al. ) within R 3.4.1 (R Development Core Team ). One hundred MCMC chains of 600,000 iterations were run, the first 100,000 iterations were removed as burn‐in and samples were saved each 50 iterations.…”

Section: Methodsmentioning

confidence: 99%

A framework for estimating the effects of sequential reproductive barriers: Implementation using Bayesian models with field data from cryptic species

2018

View full text Add to dashboard Cite

Determining how reproductive barriers modulate gene flow between populations represents a major step toward understanding the factors shaping the course of speciation. Although many indices quantifying reproductive isolation (RI) have been proposed, they do not permit the quantification of cross-direction-specific RI under varying species frequencies and over arbitrary sequences of barriers. Furthermore, techniques quantifying associated uncertainties are lacking, and statistical methods unrelated to biological process are still preferred for obtaining confidence intervals and P-values. To address these shortcomings, we provide new RI indices that model changes in gene flow for both directions of hybridization, and we implement them in a Bayesian model. We use this model to quantify RI between two species of the psyllid Cacopsylla pruni based on field genotypic data for mating individuals, inseminated spermatophores and progeny. The results showed that preinsemination isolation was strong, mildly asymmetric, and indistinguishably different between study sites despite large differences in species frequencies; that postinsemination isolation strongly affected the more common hybrid type; and that cumulative isolation was close to complete. In the light of these results, we discuss how these developments can strengthen comparative RI studies.

show abstract

Programming With Models: Writing Statistical Algorithms for General Model Structures With NIMBLE

Cited by 761 publications

References 40 publications

Estimating methane emissions from biological and fossil‐fuel sources in the San Francisco Bay Area

Estimating methane emissions from biological and fossil‐fuel sources in the San Francisco Bay Area

Faster estimation of Bayesian models in ecology using Hamiltonian Monte Carlo

A framework for estimating the effects of sequential reproductive barriers: Implementation using Bayesian models with field data from cryptic species

Contact Info

Product

Resources

About