Description Usage Arguments Details Value Note Author(s) References See Also Examples

This function generates MCMC samples from a (posterior) density function f (not necessarily normalized) in search of a global minimum of f. It uses a simple Metropolis algorithm to generate the samples. Global monitors the mcmc samples and returns the minimum value of f, as well as a sample covariance (covm) that can be used as input for the Bhat function mymcmc.

1 |

`x` |
a list with components 'label' (of mode character), 'est' (the parameter vector with the initial guess), 'low' (vector with lower bounds), and 'upp' (vector with upper bounds) |

`nlogf` |
negative log of the density function (not necessarily normalized) |

`beta` |
'inverse temperature' parameter |

`mc` |
length of MCMC search run |

`scl` |
not used |

`skip` |
number of cycles skipped for graphical output |

`nfcn` |
number of function calls |

`plot` |
logical variable. If TRUE the chain and the negative log density (nlogf) is plotted |

standard output reports a summary of the acceptance fraction, the current values of nlogf and the parameters for every (100*skip) th cycle. Plotted chains show values only for every (skip) th cycle.

list with the following components:

`fmin ` |
minimum value of nlogf for the samples obtained |

`xmin ` |
parameter values at fmin |

`covm ` |
covariance matrix of differences between consecutive samples in chain |

This function is part of the Bhat package

E. Georg Luebeck (FHCRC)

too numerous to be listed here

`dfp`

, `newton`

, `logit.hessian`

`mymcmc`

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 | ```
# generate some Poisson counts on the fly
dose <- c(rep(0,50),rep(1,50),rep(5,50),rep(10,50))
data <- cbind(dose,rpois(200,20*(1+dose*.5*(1-dose*0.05))))
# neg. log-likelihood of Poisson model with 'linear-quadratic' mean:
nlogf <- function (x) {
ds <- data[, 1]
y <- data[, 2]
g <- x[1] * (1 + ds * x[2] * (1 - x[3] * ds))
return(sum(g - y * log(g)))
}
# initialize global search
x <- list(label=c("a","b","c"), est=c(10, 0.25, 0.05), low=c(0,0,0), upp=c(100,10,.1))
# samples from posterior density (~exp(-nlogf))) with non-informative
# (random uniform) priors for "a", "b" and "c".
out <- global(x, nlogf, beta = 1., mc=1000, scl=2, skip=1, nfcn = 0, plot=TRUE)
# start MCMC from some other point: e.g. try x$est <- c(16,.2,.02)
``` |

Bhat documentation built on May 29, 2017, 8:14 p.m.

Embedding an R snippet on your website

Add the following code to your website.

For more information on customizing the embed code, read Embedding Snippets.