On 06/07/2010 10:54 AM, Paul Johnson wrote:
> Here's another example of my plotmath whipping boy, the Normal
> distribution.
>
> A colleague asks for a Normal plotted above a series of axes that
> represent various other distributions (T, etc).
>
> I want to use vectors of equations in plotmath to do this, but have
> run into trouble. Now I've isolated the problem down to a relatively
> small piece of working example code (below). If you would please run
> this, I think you will see the problem. When plotmath meets one
> vector of expressions, it converts all but one to math, so in the
> figure output i get, in LaTeX speak
>
> b1 $\mu-1.0 \sigma$ $\mu$
>
> All values except the first come out correctly.
>
> This happens only when I try to use bquote or substitute to get
> variables to fill in where the 1.96, 1.0, and so forth should be. In
> the figure output, you should see a second axis where all of the
> symbols are resolved correctly.
>
> As usual, thanks in advance for your help, sorry if I've made an
> obvious mistake or overlooked a manual.
>
> ### Filename: plotMathProblem.R
> ### Paul Johnson July 5, 2010
> ### email me <paulj...@ku.edu>
>
> sigma <- 10.0
> mu <- 4.0
>
> myx <- seq( mu - 3.5*sigma, mu+ 3.5*sigma, length.out=500)
>
> myDensity <- dnorm(myx,mean=mu,sd=sigma)
>
> ### xpd needed to allow writing outside strict box of graph
> ### Need big bottom margin to add several x axes
> par(xpd=TRUE, ps=10, mar=c(18,2,2,2))
>
> plot(myx, myDensity, type="l", xlab="", ylab="Probability Density ",
> main=myTitle1, axes=FALSE)
> axis(2, pos= mu - 3.6*sigma)
> axis(1, pos=0)
>
> lines(c(myx[1],myx[length(myx)]),c(0,0)) ### closes off axes
>
>
> addInteriorLine <- function(x, m, sd){
> for (i in 1:(length(x))){
> lines( c(x[i],x[i]), c(0, dnorm(x[i],m=m,sd=sd)), lty= 14, lwd=.2)
> }
}
>
>
> dividers <- c(qnorm(0.025), -1, 0, 1, qnorm(0.975))
> addInteriorLine(mu+sigma*dividers, mu,sigma)
>
> # bquote creates an expression that text plotters can use
> t1 <- bquote( mu== .(mu))
> mtext(bquote( mu == .(mu)), 1, at=mu, line=-1)
>
>
> addInteriorLabel <- function(pos1, pos2, m, s){
> area <- abs(100*( pnorm(m+pos1*s,m,s)-pnorm(m+pos2*s, m,s)))
> mid <- m+0.5*(pos1+pos2)*s
> text(mid, 0.5*dnorm(mid,m,s),label=paste(round(area,2),"%"))
}
>
>
> addInteriorLabel(dividers[1],dividers[2], mu, sigma)
> addInteriorLabel(dividers[2],dividers[3], mu, sigma)
> addInteriorLabel(dividers[3],dividers[4], mu, sigma)
> addInteriorLabel(dividers[4],dividers[5], mu, sigma)
>
>
>
>
> ### Following is problem point: axis will
> ### end up with correct labels, except for first point,
> ### where we end up with "b1" instead of "mu - 1.96*sigma".
> b1 <- substitute( mu - d*sigma, list(d=round(dividers[1],2)) )
> b2 <- substitute( mu - sigma )
> b3 <- substitute( mu )
> b4 <- substitute( mu + sigma )
> b5 <- substitute( mu + d*sigma, list(d=round(dividers[5],2)) )
> ## plot(-20:50,-20:50,type="n",axes=F)
> axis(1, line=4,at=mu+dividers*sigma,
> labels=c(expression(b1),b2,b3,b4,b5), padj=-1)
>
You want "as.expression(b1)", not "expression(b1)". The latter means "the
expression consisting of the symbol b1". The former means "take the object
stored in b1, and convert it to an expression.".
It's not perfect, because you'll end up with "mu - -1.96sigma" (i.e. two
minus signs), but it's closer than what you had.
