[R] Very slow assignments

[Rolf Turner]

On 08/09/11 12:05, William Dunlap wrote:
> You did not show the code you used to populate your
> object, but consider the following ways to make as.list(1:50000)
> via repeated assignments:
>
>   >  system.time( { z0<- list() ; for(i in 1:50000)z0[i]<- list(i) } )
>      user  system elapsed
>     13.34    0.00   13.42
>   >  system.time( { z1<- list() ; for(i in 1:50000)z1[[i]]<- i } )
>      user  system elapsed
>     12.98    0.00   13.36
>   >  system.time( { z2<- vector("list",50000) ; for(i in 1:50000)z2[i]<- list(i) } )
>      user  system elapsed
>      0.28    0.00    0.26
>   >  system.time( { z3<- vector("list",50000) ; for(i in 1:50000)z3[[i]]<- i } )
>      user  system elapsed
>      0.22    0.00    0.20
>   >  identical(z0,z1)&&  identical(z0,z2)&&  identical(z0,z3)
>   [1] TRUE
>
> Preallocating a vector to its ultimate size can be much faster than
> repeatedly expanding it.

This is *very* interesting to me.  Years ago I learned to use the ``z1'' 
syntax,
and thought I was being cool.  It's stunning how much more efficient the
``z2'' and ``z3'' syntaxes are.  I'll bet that there are many bunnies 
like me out
there who are unaware of the advantages of preallocation via 
vector("list",...).

Thanks for the enlightenment.

     cheers,

         Rolf Turner