Cairo pdf canvas size

Hello,

I want to save a pdf plot using Cairo, but the canvas of the saved  
file
seems too large when compared to the actual plotted area.

Is there a way to control the relation between the canvas size and  
the size
of actual plotting area?

Something like this:

u=seq(from=-pi, to=pi, length=1000)
f=sin(u)
Cairo("example.pdf", type="pdf",width=12,height=12,units="cm",dpi=300)
par(cex.axis=.6,col.axis="grey",ann=FALSE, lwd=.25,bty="n", las=1,  
tcl=-.2, mgp=c(3,.5,0))
xlim=c(-pi,pi)
ylim=round(c(min(f),max(f)))
plot(u,f,xlim,ylim,type="l",col="firebrick3", axes=FALSE)
axis(side=1, lwd=.25, col="darkgrey", at=seq(from=xlim[1],  
to=xlim[2], length=5))
axis(side=2, lwd=.25, col="darkgrey", at=seq(from=ylim[1],  
to=ylim[2], length=5))
abline(v=seq(from=xlim[1], to=xlim[2], length=5), lwd=. 
25,lty="dotted", col="grey")
abline(h=seq(from=ylim[1], to=ylim[2], length=5), lwd=. 
25,lty="dotted", col="grey")
dev.off()

Notice how the canvas' margins are relatively far from the plotting  
area.

Something like this:

u=seq(from=-pi, to=pi, length=1000)
f=sin(u)
Cairo("example.pdf", type="pdf",width=12,height=12,units="cm",dpi=300)
par(cex.axis=.6,col.axis="grey",ann=FALSE, lwd=.25,bty="n", las=1, tcl=-.2,
mgp=c(3,.5,0))
xlim=c(-pi,pi)
ylim=round(c(min(f),max(f)))
plot(u,f,xlim,ylim,type="l",col="firebrick3", axes=FALSE)
axis(side=1, lwd=.25, col="darkgrey", at=seq(from=xlim[1], to=xlim[2],
length=5))
axis(side=2, lwd=.25, col="darkgrey", at=seq(from=ylim[1], to=ylim[2],
length=5))
abline(v=seq(from=xlim[1], to=xlim[2], length=5), lwd=.25,lty="dotted",
col="grey")
abline(h=seq(from=ylim[1], to=ylim[2], length=5), lwd=.25,lty="dotted",
col="grey")
dev.off()

Thanks,

Eduardo

On Thu, Jan 6, 2011 at 1:00 AM, David Winsemius <dwinsemius at comcast.net>wrote:

On Jan 5, 2011, at 9:38 PM, Eduardo de Oliveira Horta wrote:

?Hello,

I want to save a pdf plot using Cairo, but the canvas of the saved file
seems too large when compared to the actual plotted area.

Is there a way to control the relation between the canvas size and the
size
of actual plotting area?

OS?, ?... example?

==

David Winsemius, MD
West Hartford, CT

? ? ? ?[[alternative HTML version deleted]]

Hi:

On Thu, Jan 6, 2011 at 5:36 AM, Eduardo de Oliveira Horta
<eduardo.oliveirahorta at gmail.com> wrote:

Peter,
thank you, that's what I was looking for!
David,?I forgot to tell you my OS. Sorry... it's Win7. I'm running a
RKWard session.
And this is strange:

Cairo("example.pdf", type="pdf",width=12,height=12,units="cm",dpi=300)

Error: could not find function "Cairo"
... maybe you're not using the Cairo
package??http://cran.r-project.org/web/packages/Cairo/Cairo.pdf

And Dennis, thanks for the code. It worked, and I'm considering to adopt
data frames in the near future. By the way, I'm working with functional time
series, so each observation is a function (or a vector representing that
function evaluated on a grid) indexed by time. Any insights on how to
implement data frames here?

I don't see a real issue. It would be easier to give you concrete
information if there were an artificial example that mimics your situation,
but it's not that hard.? I'd suggest looking into the zoo package to create
a series - it can handle both regular (zooreg()) and irregular (zoo())
series. Basically, a zoo object is a numeric vector with a time index. One
can create multiple series with a single index, individual series with
different indices that can be combined into data frames, etc. I've browsed
through some of the code that accompanies Ramsey, Hooker and Graves' FDA
book in R and Matlab, and occasionally they use the zoo package as well.

Here's an example, but I expect that someone will show how to convert the
zoo series to data frames much more efficiently for use in ggplot2...

library(zoo)
library(ggplot2)
library(lattice)
# Generate three daily series with different start times and lengths
a <- zoo(rnorm(450), as.Date("2005-01-01") + 0:449)
b <- zoo(rnorm(600, 1, 2), as.Date('2005-06-01') + 0:599)
d <- zoo(rnorm(300, 2, 1), as.Date('2004-09-01') + 0:299)

# Convert to data frame, make time index a variable and make sure it's a
Date object
A <- as.data.frame(a)
B <- as.data.frame(b)
D <- as.data.frame(d)
A$Date <- as.Date(rownames(A))
B$Date <- as.Date(rownames(B))
D$Date <- as.Date(rownames(D))
# Give all three series the same name
names(A)[1] <- names(B)[1] <- names(D)[1] <- 'y'
# Stack the three data frames and create a series ID variable
comb <- rbind(A, B, D)
comb$Series <- rep(c('A', 'B', 'D'), c(nrow(A), nrow(B), nrow(D)))
str(comb)??? # make sure that Date is a Date object

# ggplot of the three series
ggplot(comb, aes(x = Date, y = y, color = Series)) + geom_path()
# Stacked individual plots (faceted)
last_plot() + facet_grid(Series ~ .)

# lattice version
xyplot(y ~ Date, data = comb, groups = Series, type = 'l', col.line = 1:3)
# Stacked individual series
xyplot(y ~ Date | Series, data = comb, type = 'l', layout = c(1, 3))

If you need the grid coordinates, use expand.grid() - it can be used when
creating a data frame, too.

As Bert noted the other night in another thread, one can use xyplot directly
on zoo objects, but I don't have any direct experience with that yet so will
defer to others if they wish to contribute. ?xyplot.zoo provides some
examples.

Hope this gives you some idea of what can be done,
Dennis

Best regards,
Eduardo
On Thu, Jan 6, 2011 at 1:47 AM, Peter Langfelder
<peter.langfelder at gmail.com> wrote:

On Wed, Jan 5, 2011 at 7:35 PM, Eduardo de Oliveira Horta
<eduardo.oliveirahorta at gmail.com> wrote:

Something like this:

u=seq(from=-pi, to=pi, length=1000)
f=sin(u)
Cairo("example.pdf", type="pdf",width=12,height=12,units="cm",dpi=300)
par(cex.axis=.6,col.axis="grey",ann=FALSE, lwd=.25,bty="n", las=1,
tcl=-.2,
mgp=c(3,.5,0))
xlim=c(-pi,pi)
ylim=round(c(min(f),max(f)))
plot(u,f,xlim,ylim,type="l",col="firebrick3", axes=FALSE)
axis(side=1, lwd=.25, col="darkgrey", at=seq(from=xlim[1], to=xlim[2],
length=5))
axis(side=2, lwd=.25, col="darkgrey", at=seq(from=ylim[1], to=ylim[2],
length=5))
abline(v=seq(from=xlim[1], to=xlim[2], length=5), lwd=.25,lty="dotted",
col="grey")
abline(h=seq(from=ylim[1], to=ylim[2], length=5), lwd=.25,lty="dotted",
col="grey")
dev.off()

Wow, you must like light colors :)

To the point, just set margins, for example

par(mar = c(2,2,0.5, 0.5))

(margins are bottom, left, top, right)

after the Cairo command.

BTW, Cairo doesn't work for me either... but I tried your example by
plotting to the screen.

Peter




?Notice how the canvas' margins are relatively far from the plotting
area.

Thanks,

Eduardo

On Thu, Jan 6, 2011 at 1:00 AM, David Winsemius
<dwinsemius at comcast.net>wrote:

On Jan 5, 2011, at 9:38 PM, Eduardo de Oliveira Horta wrote:

?Hello,

I want to save a pdf plot using Cairo, but the canvas of the saved
file
seems too large when compared to the actual plotted area.

Is there a way to control the relation between the canvas size and
the
size
of actual plotting area?

OS?, ?... example?

==

David Winsemius, MD
West Hartford, CT

? ? ? ?[[alternative HTML version deleted]]