Category Archives: stochastic search
Sampling: Rejection, Reservoir, and Slice
An article by Suilou Huang for catatrophe modeler AIR-WorldWide of Boston about rejection sampling in CAT modeling got me thinking about pulling together some notes about sampling algorithms of various kinds. There are, of course, books written about this subject, … Continue reading
Six cases of models
The previous post included an attempt to explain land surface temperatures as estimated by the BEST project using a dynamic linear model including regressions on both quarterly CO2 concentrations and ocean heat content. The idea was to check the explanatory … Continue reading
“Grid shading by simulated annealing” [Martyn Plummer]
Source: Grid shading by simulated annealing (or what I did on my holidays), aka “fun with GCHQ job adverts”, by Martyn Plummer, developer of JAGS. Excerpt: I wanted to solve the puzzle but did not want to sit down with … Continue reading
high dimension Metropolis-Hastings algorithms
If attempting to simulate from a multivariate standard normal distribution in a large dimension, when starting from the mode of the target, i.e., its mean γ, leaving the mode γis extremely unlikely, given the huge drop between the value of the density at the mode γ and at likely realisations Continue reading
Generating supports for classification rules in black box regression models
Inspired by the extensive and excellent work in approximate Bayesian computation (see also), especially that done by Professors Christian Robert and colleagues (see also), and Professor Simon Wood (see also), it occurred to me that the complaints regarding lack of … Continue reading
reblog: “Tiny Data, Approximate Bayesian Computation and the Socks of Karl Broman”
It’s Rasmus Bååth, in a post and video of which I am very fond: http://www.sumsar.net/blog/2014/10/tiny-data-and-the-socks-of-karl-broman/.
Southern New England Meteorology Conference, 24th October 2015
I attending the 2015 edition of the Southern New England Meteorology Conference in Milton, MA, near the Blue Hill, and its Blue Hill Climatological Observatory, of which I am a member as we as of the American Meteorological Society. I … Continue reading
On differential localization of tumors using relative concentrations of ctDNA. Part 1.
Like most mammalian tissue, tumors often produce shards of DNA as a byproduct of cell death and fracture. This circulating tumor DNA is being studied as a means of detecting tumors or their resurgence after treatment. (See also a Q&A … Continue reading
On the Climate Club
But if the other advanced nations had a stick — a tariff of 4 percent on the imports from countries not in the “climate club” — the cost-benefit calculation for the United States would flip. Not participating in the club … Continue reading
“A vignette on Metropolis” (Christian Robert)
Originally posted on Xi'an's Og:
Over the past week, I wrote a short introduction to the Metropolis-Hastings algorithm, mostly in the style of our Introduction to Monte Carlo with R book, that is, with very little theory and…
Markov Chain Monte Carlo methods and logistic regression
This post could also be subtitled “Residual deviance isn’t the whole story.” My favorite book on logistic regression is by Dr Joseph Hilbe, Logistic Regression Models, CRC Press, 2009, Chapman & Hill. It is a solidly frequentist text, but its … Continue reading
Christian Robert on the amazing Gibbs sampler
Professor Christian Robert remarks on the amazing Gibbs sampler. Implicitly he’s also underscoring the power of properly done Bayesian computational analysis. For here we have a problem with a posterior distribution having two strong modes, so a point estimate, like … Continue reading
Christian Robert on Alan Turing
Alan Turing Institute. See Professor Robert’s earlier post on Turing, too.
engineering and understanding with stable models
Stable distributions or Lévy -stable models is a class of probability distributions which contains the Gaussian, the Cauchy (or Lorentz), and the Lévy distribution. They are parameterized by an which is . Values of of 1 or less give distributions … Continue reading
On nested equivalence classes of climate models, ordered by computational complexity
I’m digging into the internals of ABC, for professional and scientific reasons. I’ve linked a great tutorial elsewhere, and argued that this framework, advanced by Wood, and Wilkinson (Robert), and Wilkinson (Darren), and Hartig and colleagues, and Robert and colleagues, … Continue reading
struggling with problems already partly solved by others
Climate modelers and models see as their frontier the problem of dealing with spontaneous dynamics in systems such as atmosphere or ocean which are not directly forced by boundary conditions such as radiative forcing due to increased greenhouse gas (“GHG”) … Continue reading
Bayesian deconvolution of stick lengths
Consider trying to determine the length of a straight stick. Instead of the measurement errors being clustered about zero, suppose the errors are known to be always positive, that is, no measurement ever underestimates the length of the stick. Such … Continue reading
The dp-means algorithm of Kulis and Jordan in R and Python
dp-means algorithm. Think k-means but with the number of clusters calculated. By John Myles White, in R. (Github link off that page.) By Scott Hendrickson, in Python. (Github link off that page.)
Blind Bayesian recovery of components of residential solid waste tonnage from totals data
This is a sketch of how maths and statistics can do something called blind source separation, meaning to estimate the components of data given only their totals. Here, I use Bayesian techniques for the purpose, sometimes called Bayesian inversion, using … Continue reading
singingbanana does “The Lorenz Machine”
On the power of mathematics, and why 55:45 versus 50:50 matters.
“The joy and martyrdom of trying to be a Bayesian”
Bayesians have all been there. Some of us don’t depend upon producing publications to assure our pay, so we less have the pressure of pleasing peer reviewers. Nonetheless, it’s all reacting to “What the hell are you doing? I don’t … Continue reading
The zero-crossings trick for JAGS: Finding roots stochastically
BUGS has a “zeros trick” (Lund, Jackson, Best, Thomas, Spiegelhalter, 2013, pages 204-206; see also an online illustration) for specifying a new distribution which is not in the standard set. The idea is to couple an invented-for-the-moment Poisson density to … Continue reading
“Double Plus Big Data”
Big Data. All the rage. Why? Apart from distributed software folks strutting their stuff, something which is likely to be fleeting, especially when quantum computing comes around, what does it buy anyone? I can see four possibilities, which I consider … Continue reading
“Bayes’ theorem in the 21st century”
Professor Bradley Efron wrote a piece on “Bayes’ theorem in the 21st century” in Science for 7th June 2013 which, as always, offers his measured approach to the frequentist-Bayesian controversy (see B. Efron, “A 250 year argument: Belief, behavior, and the … Continue reading