Hierarchical statistical modeling software: Write models, MCMCs, particle filters, or other needs. Automatically compile them from R via code-generated C++.
Announcing nimble version 1.4.0 and new package nimbleQuad (1.4.0 to match nimble), both on CRAN (https://t.co/mhUqCHfx36, https://t.co/kTd3RYLzQy). nimbleQuad provides Laplace approximation, adaptive Gauss-Hermite quadrature, and INLA-like posterior approximations.
Announcing the new nimbleMacros package on CRAN, which provides more compact ways to specify linear model components or other model components in the nimble hierarchical modeling language. You can also write your own model macros. https://t.co/oUeBwXHaaM
New versions of nimble and nimbleHMC are available on CRAN. nimble now includes a Barker block sampler for MCMC, better implementations of Laplace approximation and adaptive Gauss-Hermite quadrature, and calling any user-provided optimization function. https://t.co/7Adq09ih7g
@Doofgradstudent@BenCAugustine Take a look at the variables in the uncompiled model after running the MCMC and see if any of them have changed size. That would indicate where uncompiled (R) execution is dynamically increasing a size, which won't happen in the compiled version.
🚨 The second article of my PhD is out 🥳 We integrated commute-time distance⚡ into dynamic occupancy models using @R_nimble to model carnivore recolonisation 🦦🐱. Curious why we're using hierarchical models for connectivity analyses? Check out the blog post!
@masonfidino Oh no! I'm sorry to hear that. We really try to respond to all questions, but sometimes one slips past us during busy times. I hope you'll give the list another try at some point.
We've updated nimbleEcology to use nimble's automatic differentiation features, allowing its occupancy, capture-recapture, HMM, and N-mixture models to work with HMC, Laplace approximation, and other AD algorithms. https://t.co/PKdadDZVeL @Ben_R_Goldstein
Minor update to nimbleHMC, with No-U-turn Hamiltonian Monte Carlo samplers for nimble models: https://t.co/LwIXp6AWio. Version 0.2.2 includes better diagnostic checking for AD (automatic differentiation) support in any parts of a model to be sampled by HMC.
nimble 1.2.0 is out! (details: https://t.co/Kbznhp9HCQ). Includes adaptive Gauss-Hermite quadrature, better Laplace approx, Pólya-gamma sampler, noncentered sampler, revamped MCEM, new ways to provide your own distributions and functions with internal data, and some speedups.
Currently struggling with MCMC simulation analysis where some data sets require centered RE parameterization, other noncentered.
Nimble just added a new sampler that does both 🤯🤯
https://t.co/rJBbIXeErC
We spent such a great time talking about ecological modeling and Bayesian Inference with @R_nimble at @IREC_CSIC_UCLM#IBER24! 🗺️🦌📊🐬📉Thanks Pepe Jimenez for your talk and all attendees for coming, we hope to repeat it soon! Materials (🇪🇸) at https://t.co/kMbbfUSdZN #rstats
Running these analyses are bit complicated, and I rely a lot on the amazing @R_nimble package to build my models. But to make things more user-friendly I developed baorista, a dedicated R package that put the most complicated things in the backend
https://t.co/xB8s9LNw84
Todo preparado para el taller sobre Inferencia Bayesiana en Ecología con R y @R_nimble! Tres días en el @IREC_CSIC_UCLM hablando sobre ecología y estadśitica 📊🦌📉🐰. Más info:
https://t.co/kMbbfUSdZN #dIBER#rstat#bayes @ValentinLauret Cheatsheet by @SoniaIllanas 😊
So we developed an alternative Bayesian approach, using the amazing @R_nimble R package. We come up with two solutions, a parametric approach based on the classic 's-shape' curve discussed in the literature and a more flexible non-parametric method.
— ABSTRACT SUBMISSION OPEN —
The #ISEC2024 conference will showcase what’s cool and exciting in #statisticalecology.
Want to be part of it? 🤓☔️
You can now submit your abstract for talk or poster. Just follow the link on our website:
https://t.co/Z8gaLQq8Vx