R help - Jul 2013 - image versus levelplot

Dear R users,

I'm currently using the Graphics package to display several hundred of 
matrix objects, using a layout and the image() function.
It works well except for large matrices (> 1000*1000) or for a large 
number of matrices (there is a limitation around 400 if I remember well)
To solve these issues, I move to the Matrix package which is much more 
efficient for large sparse matrix, and to the grid package for the plot 
(using image from Matrix)
However, the levelplot() is much more slow than the graphics image() 
function !!
Basically, with the image() from graphics, I'm able to plot 360 matrices 
in 30 sec, against more than 10 minutes with levelplot().
I would appreciate if anyone has some experience on that, and could give 
me some advice to efficiently use the grid package !?
Regards
Nicolas

 > i=1000
 > M1 <- Matrix(rnorm(i*i), ncol=i)
 > system.time(print(Matrix::image(M1)))
    user  system elapsed
  11.320   0.064  11.406
 > M2 <- matrix(rnorm(i*i), ncol=i)
 > system.time(image(as.matrix(M2)))
    user  system elapsed
   0.837   0.004   0.844





-- 
Nicolas Servant
Plateforme de Bioinformatique
Unit? 900 : Institut Curie - Inserm - Mines ParisTech
26, rue d'Ulm - 75248 Paris Cedex 05 - FRANCE

Email: Nicolas.Servant at curie.fr
Tel: 01 56 24 69 85
http://bioinfo.curie.fr/

Hi

You could try the 'useRaster' argument to levelplot()

Alternatively, if you are not doing much annotation (such as axes and 
labels) around the matrix-representations, you could possibly speed 
things up by calling grid.rect() directly.  Feel free to contact me 
directly if that sounds like an option.

Paul

On 07/16/13 09:55, Nicolas Servant wrote:
> Dear R users,
>
> I'm currently using the Graphics package to display several hundred of
> matrix objects, using a layout and the image() function.
> It works well except for large matrices (> 1000*1000) or for a large
> number of matrices (there is a limitation around 400 if I remember well)
> To solve these issues, I move to the Matrix package which is much more
> efficient for large sparse matrix, and to the grid package for the plot
> (using image from Matrix)
> However, the levelplot() is much more slow than the graphics image()
> function !!
> Basically, with the image() from graphics, I'm able to plot 360
matrices
> in 30 sec, against more than 10 minutes with levelplot().
> I would appreciate if anyone has some experience on that, and could give
> me some advice to efficiently use the grid package !?
> Regards
> Nicolas
>
>  > i=1000
>  > M1 <- Matrix(rnorm(i*i), ncol=i)
>  > system.time(print(Matrix::image(M1)))
>     user  system elapsed
>   11.320   0.064  11.406
>  > M2 <- matrix(rnorm(i*i), ncol=i)
>  > system.time(image(as.matrix(M2)))
>     user  system elapsed
>    0.837   0.004   0.844
>
>
>
>
>
-- 
Dr Paul Murrell
Department of Statistics
The University of Auckland
Private Bag 92019
Auckland
New Zealand
64 9 3737599 x85392
paul at stat.auckland.ac.nz
http://www.stat.auckland.ac.nz/~paul/