[R-sig-phylo] Why does ace set a random number generator seed?

[David Bapst]

Emmanuel, all-

I noticed today that a workspace I was working with had a random
number seed set in it, but didn't remember setting one. Finally, I
discovered the culprit was ace. Here's a reproducible example,
demonstrating that a seed exists after running ace:

library(ape)
tree<-rtree(10)
ace(1:10,tree)
.Random.seed

I am using ape 4.1, and it doesn't seem to be addressed in the
forthcoming version, given my reading of the changes log (but I might
have missed it). What's going? Why is a random number seed being set?

Cheers,
-Dave
--
David W. Bapst, PhD
Postdoc, Ecology & Evolutionary Biology, Univ of Tenn Knoxville
Lecturer, Geology & Geophysics, Texas A & M University
https://github.com/dwbapst/paleotree
Google Calender: https://goo.gl/EpiM4J

_______________________________________________
R-sig-phylo mailing list - R-sig-***@r-project.org
https://stat.ethz.ch/mailman/listinfo/r-sig-phylo
Searchable archive at http://www.mail-archive.com/r-sig-***@r-project.org/

[Emmanuel Paradis]

Hi Dave,

The seed is created every time you generate random data (see details in
?.Random.seed):

R> exists(".Random.seed")
[1] FALSE
R> sample(1)
[1] 1
R> exists(".Random.seed")
[1] TRUE

So in your code below the seed is created by calling rtree(). Indeed,
with ape 4.1:

R> rm(".Random.seed")
R> data(bird.orders)
R> x <- 1:23
R> exists(".Random.seed")
[1] FALSE
R> o <- ace(x, bird.orders)
R> exists(".Random.seed")
[1] FALSE

However, with the new "soon-to-be-on-CRAN" version, a seed is created by
the same code. The "culprit" is actually Rcpp: the new ape has C++ code
linked to Rcpp. For a reason I ignore, it seems that a random seed is
initialized when calling a function linked to Rcpp. Here is an example:

R> rm(".Random.seed")
R> library(RcppEigen)
R> exists(".Random.seed")
[1] FALSE
R> o <- fastLm(1, 1)
R> exists(".Random.seed")
[1] TRUE

The new C++ code in ape is called by reorder(phy, "postorder") which is
used by many functions in ape (ace, pic, plot.phylo, vcv, ...).

Best,

Emmanuel

Post by David Bapst
Emmanuel, all-
I noticed today that a workspace I was working with had a random
number seed set in it, but didn't remember setting one. Finally, I
discovered the culprit was ace. Here's a reproducible example,
library(ape)
tree<-rtree(10)
ace(1:10,tree)
.Random.seed
I am using ape 4.1, and it doesn't seem to be addressed in the
forthcoming version, given my reading of the changes log (but I might
have missed it). What's going? Why is a random number seed being set?
Cheers,
-Dave

_______________________________________________
R-sig-phylo mailing list - R-sig-***@r-project.org
https://stat.ethz.ch/mailman/listinfo/r-sig-phylo
Searchable archive at http://

[David Bapst]

Thanks, Emmanuel. I was not familiar with .Random.seed objects until R
CHECK started complaining about such elements in saved workspaces
within .rdata files within packages (requiring rm to remove them), and
then was mainly familiar with them regarding set.seed() (and so I
thought ace must be setting a seed, not just initializing one). It
makes sense that any random number generation would incur a seed, and
apparently fresh R sessions do not immediately initialize a seed.

Thanks to everyone who helped me see my confusion! Just me missing a
basic comp sci concept.
-Dave

On Fri, Oct 20, 2017 at 4:26 AM, Emmanuel Paradis

Post by Emmanuel Paradis
Hi Dave,
The seed is created every time you generate random data (see details in
R> exists(".Random.seed")
[1] FALSE
R> sample(1)
[1] 1
R> exists(".Random.seed")
[1] TRUE
So in your code below the seed is created by calling rtree(). Indeed, with
R> rm(".Random.seed")
R> data(bird.orders)
R> x <- 1:23
R> exists(".Random.seed")
[1] FALSE
R> o <- ace(x, bird.orders)
R> exists(".Random.seed")
[1] FALSE
However, with the new "soon-to-be-on-CRAN" version, a seed is created by the
same code. The "culprit" is actually Rcpp: the new ape has C++ code linked
to Rcpp. For a reason I ignore, it seems that a random seed is initialized
R> rm(".Random.seed")
R> library(RcppEigen)
R> exists(".Random.seed")
[1] FALSE
R> o <- fastLm(1, 1)
R> exists(".Random.seed")
[1] TRUE
The new C++ code in ape is called by reorder(phy, "postorder") which is used
by many functions in ape (ace, pic, plot.phylo, vcv, ...).
Best,
Emmanuel

Post by David Bapst
Emmanuel, all-
I noticed today that a workspace I was working with had a random
number seed set in it, but didn't remember setting one. Finally, I
discovered the culprit was ace. Here's a reproducible example,
library(ape)
tree<-rtree(10)
ace(1:10,tree)
.Random.seed
I am using ape 4.1, and it doesn't seem to be addressed in the
forthcoming version, given my reading of the changes log (but I might
have missed it). What's going? Why is a random number seed being set?
Cheers,
-Dave

--
David W. Bapst, PhD
Postdoc, Ecology & Evolutionary Biology, Univ of Tenn Knoxville
Lecturer, Geology & Geophysics, Texas A & M University
https://github.com/dwbapst/paleotree
Google Calender: https://goo.gl/EpiM4J

_______________________________________________
R-sig-phylo mailing list - R-sig-***@r-project.org
https://stat.ethz.ch/mailman/listinfo/r-sig-phylo
Searchable archive a