paleoTS (version 0.3-1)

opt.RW.Mult: Functions to analyze multiple time-series jointly

Description

These functions are used to estimate parameters of the general random walk over two or more sequences. Three different models may be fit: the same parameters are present in all time series (opt.RW.mult); the same mean step but different step variances in each series (opt.RW.SameMs); the same step variance but different mean steps in each series (opt.RW.SameVs).

Usage

logL.Mult(p, y, model = c("GRW", "URW"))
logL.SameMs(p, y)
logL.SameVs(p, y)
opt.RW.Mult(y, cl=list(fnscale=-1), model=c("GRW", "URW"), pool=TRUE, meth="L-BFGS-B", hess=FALSE)
opt.RW.SameMs(y, cl=list(fnscale=-1), pool=TRUE, meth="L-BFGS-B", hess=FALSE)
opt.RW.SameVs(y, cl=list(fnscale=-1), pool=TRUE, meth="L-BFGS-B", hess=FALSE)

Arguments

p

a vector of parameter values

y

a emph{list} of paleoTS objects.

model

RW for general random walk, RWu for unbiased random walk

pool

logical, if TRUE variances are pooled across samples

cl

optimization option, passed to optim

meth

optimization option, passed to optim

hess

optimization option, passed to optim

Value

• Varies by function, see corresponding functions for the analysis of single sequences for more information.

Details

These functions work just as their conterparts for the analysis of single sequences; see those help functions for more detail.

References

Hunt, G. 2006. Fitting and comparing models of phyletic evolution: random walks and beyond. Paleobiology32:578--601.

See Also

logL.GRW, opt.GRW, fit3models

Examples

## create two sequences, with different parameter values
 y1<- sim.GRW(ns=20, ms=0, vs=1)
 y2<- sim.GRW(ns=20, ms=0, vs=0.2)

 ## fit some models with at least some shared dynamics across sequences
 m1<- opt.RW.Mult(list(y1,y2), model="GRW")
 m2<- opt.RW.SameMs(list(y1, y2))
 m3<- opt.RW.SameVs(list(y1,y2))

 ## fit separate models to each sequence
 msep1<- opt.GRW(y1)
 msep2<- opt.GRW(y2)
 sep.logL<- msep1$value + msep2$value   # total logL sums across two solutions
 sep.aic <- -2*sep.logL + 2*4     # AIC of separate solutions

 ## check out resulting log-likelihoods and AIC values
 mres<- array(dim=c(4,2))
 row.names(mres)<- c("m1", "m2", "m3", "msep")
 colnames(mres)<- c("logL", "AIC")
 mres[1,]<- c(m1$value, m1$AIC)
 mres[2,]<- c(m2$value, m2$AIC)
 mres[3,]<- c(m3$value, m3$AIC)
 mres[4,]<- c(sep.logL, sep.aic)
 print (mres)