R help - Jul 2017 - axis() after image.plot() does not work except if points() is inserted between

Thanks for the proposition. As you see bellow, par("usr") is the same 
before and after the points() (the full code is bellow):
....
 > par("usr")
[1] -0.2500000  1.2500000 -0.1666667  1.1666667
 > # if you remove this points() function, axis will show nothing.
 >
 > points(1.5, 1.5, type="p")
 > p2 <- par(no.readonly=TRUE)
 > par("usr")
[1] -0.2500000  1.2500000 -0.1666667  1.1666667
...

I can reproduce it in Ubuntu and MacosX R Gui and Rstudio (R 3.4.1).

Marc

Here is the code:
library(fields)
par(mar=c(5,4.5,4,7))
D <- matrix(c(10, 20, 25, 30, 12, 22, 32, 35, 13, 25, 38, 40), nrow=3)

p0 <- par(no.readonly=TRUE)
image.plot(D, col=rev(heat.colors(128)),bty="n",
xlab="Lines",
            ylab="Columns", cex.lab = 0.5, zlim=c(min(D,
na.rm=TRUE),max(D,
na.rm=TRUE)),
            las=1, axes=FALSE)
p1 <- par(no.readonly=TRUE)

par("usr")
par("xpd")

# if you remove this points() function, axis will show nothing.

points(1.5, 1.5, type="p")
p2 <- par(no.readonly=TRUE)
par("usr")
par("xpd")

##########
axis(1, at=seq(from=0, to=1, length=nrow(D)), labels=0:2, cex.axis=0.5)
axis(2, at=seq(from=0, to=1, length=ncol(D)), labels=0:3, las=1,
      cex.axis=0.5)

identical(p1, p2)




Le 24/07/2017 ? 05:17, Jim Lemon a ?crit :
> Hi marc,
> Try:
>
> par("usr")
>
> before and after the call to points and see if it changes.
>
> Jim
>
>
> On Sat, Jul 22, 2017 at 12:05 AM, Marc Girondot via R-help
> <r-help at r-project.org> wrote:
>> It is known (several discussions on internet) that axis() cannot be
used
>> after fields:::image.plot() (axis() shows nothing).
>>
>> However, if points(1.5, 1.5, type="p") is inserted before the
axis()
>> finctions, it works.
>>
>> I have investigated what points(1.5, 1.5, type="p") is doing
to allow axis
>> to work and I don't find a solution. par() options are identical
(p1 and p2
>> are identical).
>>
>> Has someone a solution ?
>>
>> Thanks
>>
>> Marc
>>
>> library(fields)
>> p0 <- par(no.readonly=TRUE)
>> image.plot(D, col=rev(heat.colors(128)),bty="n",
xlab="Lines",
>>             ylab="Columns", cex.lab = 0.5, zlim=c(min(D,
na.rm=TRUE),max(D,
>> na.rm=TRUE)),
>>             las=1, axes=FALSE)
>> p1 <- par(no.readonly=TRUE)
>>
>> # if you remove this points() function, axis will show nothing.
>>
>> points(1.5, 1.5, type="p")
>> p2 <- par(no.readonly=TRUE)
>>
>> ##########
>> axis(1, at=seq(from=0, to=1, length=nrow(D)), labels=0:2, cex.axis=0.5)
>> axis(2, at=seq(from=0, to=1, length=ncol(D)), labels=0:3, las=1,
>> cex.axis=0.5)
>>
>> identical(p1, p2)
>>
>> ______________________________________________
>> R-help at r-project.org mailing list -- To UNSUBSCRIBE and more, see
>> https://stat.ethz.ch/mailman/listinfo/r-help
>> PLEASE do read the posting guide
http://www.R-project.org/posting-guide.html
>> and provide commented, minimal, self-contained, reproducible code.

>>>>> Marc Girondot via R-help <r-help at r-project.org>
>>>>>     on Mon, 24 Jul 2017 09:35:06 +0200 writes:
    > Thanks for the proposition. As you see bellow, par("usr") is
the same
    > before and after the points() (the full code is bellow):
    > ....
    >> par("usr")
    > [1] -0.2500000  1.2500000 -0.1666667  1.1666667
    >> # if you remove this points() function, axis will show nothing.
    >> 
    >> points(1.5, 1.5, type="p")
    >> p2 <- par(no.readonly=TRUE)
    >> par("usr")
    > [1] -0.2500000  1.2500000 -0.1666667  1.1666667
    > ...

    > I can reproduce it in Ubuntu and MacosX R Gui and Rstudio (R 3.4.1).

    > Marc

    > Here is the code:
    > library(fields)
    > par(mar=c(5,4.5,4,7))
    > D <- matrix(c(10, 20, 25, 30, 12, 22, 32, 35, 13, 25, 38, 40),
nrow=3)

    > p0 <- par(no.readonly=TRUE)
    > image.plot(D, col=rev(heat.colors(128)),bty="n",
xlab="Lines",
    >           ylab="Columns", cex.lab = 0.5, 
    >           zlim=c(min(D, na.rm=TRUE),max(D, na.rm=TRUE)),
    >           las=1, axes=FALSE)
    > p1 <- par(no.readonly=TRUE)

    > par("usr")
    > par("xpd")

    > # if you remove this points() function, axis will show nothing.

    > points(1.5, 1.5, type="p")
    > p2 <- par(no.readonly=TRUE)
    > par("usr")
    > par("xpd")

    > ##########
    > axis(1, at=seq(from=0, to=1, length=nrow(D)), labels=0:2, cex.axis=0.5)
    > axis(2, at=seq(from=0, to=1, length=ncol(D)), labels=0:3, las=1,
    > cex.axis=0.5)

    > identical(p1, p2)

Have you ever carefully read the detailed help page about image.plot()?
   
I haven't, but a cursory reading already shows me that the
author of the function did this partly on purpose:

  > Side Effects:
  > 
  >      After exiting, the plotting region may be changed to make it
  >      possible to add more features to the plot. To be explicit,
  >      ?par()\$plt? may be changed to reflect a smaller plotting region
  >      that has accommodated room for the legend subplot.

Unfortunately, there it does _not_ mention the following :
>From looking at its code, and then re-reading parts of the help page,
I see that there is a 'graphics.reset' argument which you can
set to TRUE in such a case:

 image.plot(D, col=rev(heat.colors(128)),bty="n",
xlab="Lines",
           ylab="Columns", cex.lab = 0.5, 
           zlim= range(D, na.rm=TRUE),
	   graphics.reset = TRUE, # <<<<< the solution
           las=1, axes=FALSE)

Also note that
     zlim = range(D ...)
is infinitely more elegant than
     zlim = c(min((D, ...), max(D, ...)))


Martin Maechler
ETH Zurich (and R core)

Thanks... I agree that the problem was explained in the documentation 
but I can't find a way to have axis() working even manipulating 
par("plt") or with graphics.reset = TRUE:
- adding graphics.reset=TRUE does not allow axis() to be shown;
- I see that par()$plt is involved but it is the not sufficient to 
explain why axis() works because if it is changed by hand, axes are not 
shown.

Thanks for the trick about range(). I didn't notice that it has also a 
na.rm option. It is more elegant.

Here is the code showing the various propositions and checks:


library(fields)
D <- matrix(c(10, 20, 25, 30, 12, 22, 32, 35, 13, 25, 38, 40), nrow=3)
(pplt <- par()$plt)
#### [1] 0.08844944 0.95469663 0.14253275 0.88541485

# original problem. Axis() not shown

image.plot(D, col=rev(heat.colors(128)),bty="n",
xlab="Lines",
            ylab="Columns", cex.lab = 0.5, zlim=range(D, na.rm=TRUE),
            las=1, axes=FALSE)
# Check the value of par()$plt; it is indeed modified
par()$plt
#### [1] 0.08844944 0.86408989 0.14253275 0.88541485
# axis() does not work
axis(1, at=seq(from=0, to=1, length=nrow(D)), labels=0:2, cex.axis=0.5)
axis(2, at=seq(from=0, to=1, length=ncol(D)), labels=0:3, las=1, 
cex.axis=0.5)

# I restore par("plt") to it original value
par(plt=pplt)

# graphics.reset = TRUE. Axis() not shown

par()$plt
### [1] 0.08844944 0.95469663 0.14253275 0.88541485
image.plot(D, col=rev(heat.colors(128)),bty="n",
xlab="Lines",
            ylab="Columns", cex.lab = 0.5, zlim=range(D, na.rm=TRUE),
            las=1, axes=FALSE, graphics.reset = TRUE)
# Check an effect on par()$plt. Indeed with graphics.reset = TRUE 
par()$plt is restored
par()$plt
### [1] 0.08844944 0.95469663 0.14253275 0.88541485
# But the axes at not shown
axis(1, at=seq(from=0, to=1, length=nrow(D)), labels=0:2, cex.axis=0.5)
axis(2, at=seq(from=0, to=1, length=ncol(D)), labels=0:3, las=1, 
cex.axis=0.5)

# Check an effect of points() on par()$plt
# There is an effect
# axes are shown

par()$plt
### [1] 0.08844944 0.95469663 0.14253275 0.88541485
image.plot(D, col=rev(heat.colors(128)),bty="n",
xlab="Lines",
            ylab="Columns", cex.lab = 0.5, zlim=range(D, na.rm=TRUE),
            las=1, axes=FALSE)
points(1.5, 1.5, type="p")
par()$plt
### [1] 0.08844944 0.86408989 0.14253275 0.88541485
axis(1, at=seq(from=0, to=1, length=nrow(D)), labels=0:2, cex.axis=0.5)
axis(2, at=seq(from=0, to=1, length=ncol(D)), labels=0:3, las=1, 
cex.axis=0.5)

# Try to reproduce the effect of points() on par()$plt
# axes are not shown. Then points() is doing something else !

image.plot(D, col=rev(heat.colors(128)),bty="n",
xlab="Lines",
            ylab="Columns", cex.lab = 0.5, zlim=range(D, na.rm=TRUE),
            las=1, axes=FALSE)
par(plt=c(0.08844944, 0.86408989, 0.14253275, 0.88541485))
axis(1, at=seq(from=0, to=1, length=nrow(D)), labels=0:2, cex.axis=0.5)
axis(2, at=seq(from=0, to=1, length=ncol(D)), labels=0:3, las=1, 
cex.axis=0.5)
# check that par("plt") is correctly setup
par()$plt
### [1] 0.08844944 0.86408989 0.14253275 0.88541485

I think that it will remain a mystery !
At least the trick with points() is working.

Thanks
Marc


Le 25/07/2017 ? 13:03, Martin Maechler a ?crit :
>>>>>> Marc Girondot via R-help <r-help at
r-project.org>
>>>>>>      on Mon, 24 Jul 2017 09:35:06 +0200 writes:
>      > Thanks for the proposition. As you see bellow,
par("usr") is the same
>      > before and after the points() (the full code is bellow):
>      > ....
>      >> par("usr")
>      > [1] -0.2500000  1.2500000 -0.1666667  1.1666667
>      >> # if you remove this points() function, axis will show
nothing.
>      >>
>      >> points(1.5, 1.5, type="p")
>      >> p2 <- par(no.readonly=TRUE)
>      >> par("usr")
>      > [1] -0.2500000  1.2500000 -0.1666667  1.1666667
>      > ...
>
>      > I can reproduce it in Ubuntu and MacosX R Gui and Rstudio (R
3.4.1).
>
>      > Marc
>
>      > Here is the code:
>      > library(fields)
>      > par(mar=c(5,4.5,4,7))
>      > D <- matrix(c(10, 20, 25, 30, 12, 22, 32, 35, 13, 25, 38, 40),
nrow=3)
>
>      > p0 <- par(no.readonly=TRUE)
>      > image.plot(D, col=rev(heat.colors(128)),bty="n",
xlab="Lines",
>      >           ylab="Columns", cex.lab = 0.5,
>      >           zlim=c(min(D, na.rm=TRUE),max(D, na.rm=TRUE)),
>      >           las=1, axes=FALSE)
>      > p1 <- par(no.readonly=TRUE)
>
>      > par("usr")
>      > par("xpd")
>
>      > # if you remove this points() function, axis will show nothing.
>
>      > points(1.5, 1.5, type="p")
>      > p2 <- par(no.readonly=TRUE)
>      > par("usr")
>      > par("xpd")
>
>      > ##########
>      > axis(1, at=seq(from=0, to=1, length=nrow(D)), labels=0:2,
cex.axis=0.5)
>      > axis(2, at=seq(from=0, to=1, length=ncol(D)), labels=0:3, las=1,
>      > cex.axis=0.5)
>
>      > identical(p1, p2)
>
> Have you ever carefully read the detailed help page about image.plot()?
>     
> I haven't, but a cursory reading already shows me that the
> author of the function did this partly on purpose:
>
>    > Side Effects:
>    >
>    >      After exiting, the plotting region may be changed to make it
>    >      possible to add more features to the plot. To be explicit,
>    >      ?par()\$plt? may be changed to reflect a smaller plotting
region
>    >      that has accommodated room for the legend subplot.
>
> Unfortunately, there it does _not_ mention the following :
>
> >From looking at its code, and then re-reading parts of the help page,
> I see that there is a 'graphics.reset' argument which you can
> set to TRUE in such a case:
>
>   image.plot(D, col=rev(heat.colors(128)),bty="n",
xlab="Lines",
>             ylab="Columns", cex.lab = 0.5,
>             zlim= range(D, na.rm=TRUE),
> 	   graphics.reset = TRUE, # <<<<< the solution
>             las=1, axes=FALSE)
>
> Also note that
>       zlim = range(D ...)
> is infinitely more elegant than
>       zlim = c(min((D, ...), max(D, ...)))
>
>
> Martin Maechler
> ETH Zurich (and R core)
>