Not getting good convergence from jags models with uniformative priors without observation noise and arbitary starting postions. See examples:

  • fixed myers_jags run, loads knitr_defaults 11:18 am 2013/05/29
  • trouble with MCMC convergence for process-noise-only: Now with longer runs and better posterior estimator. Set for run on zero. 11:01 am 2013/05/29

Also a few updates:


  • Combine comment code into single file for both models: comment_reply.Rmd
  • Comment resubmitted. (repository tag: resubmission)
  • Ooh: tags provide a convenient way to make readable version-stable links (e.g.¬†as opposed to linking by the hash.)

handling scripts

Separated out my common knitr settings that usually take up space in my first code chunk.

# My preferred defaults (may be changed in individual chunks)
opts_chunk$set(tidy=FALSE, warning=FALSE, message=FALSE, cache=TRUE, 
               comment=NA, verbose=TRUE, fig.width=6, fig.height=4)
# Name the cache path and fig.path based on filename...
 opts_chunk$set(fig.path = paste("figure/",
                                 gsub(".Rmd", "", knitr:::knit_concord$get('infile')),
                                 "-", sep=""),
                                 cache.path = paste(gsub(".Rmd", "", knitr:::knit_concord$get('infile') ), 
                                 "/", sep=""))
# Set plotting to bw plot default, but with transparent background elements.  
# Note transparency requires the panel.background, plot.background, and device background all be set!
  theme_update(panel.background = element_rect(fill = "transparent", colour = NA),
               plot.background = element_rect(fill = "transparent", colour = NA))
# Set a color-blind friendly pallette
# adapted from
   cbPalette <- c("#000000", "#E69F00", "#56B4E9", "#009E73", 
                  "#F0E442", "#0072B2", "#D55E00", "#CC79A7")

also appears as gist:5600558

Saved script as ~/.knit_defaults.R and is sourced in by the first chunk instead.