Dear Luke,
I'll have a look, if I find obvious things to speed it up. I will also
have a closer look into fitDiscrete, as in phangorn there is quite a
lot C-Code which could speed it up. I'll keep you informed.
Cheers,
Klaus
On 11/15/10, Luke Harmon wrote:
> Just a quick note re: fitContinuou
Hi all,
Thank you everyone for the replies. I was finally able to get an
estimate of lambda for my nearly 4000 species following Emmanuel
Paradis' suggestion with corPagel and gls. The models took 2-4 hours to
run which is pretty good.
Thanks again for the quick and very helpful responses.
Man
Hi all,
gls() with 'corPagel' correlation structure and an intercept-only
model also estimates lambda. The results will not be identical with
fitContinuous but usually close. To compare the timings, with n = 100:
tr <- rcoal(100)
co <- corPagel(1, tr)
x <- rTraitCont(tr)
system.time(fitConti
Just a quick note re: fitContinuous and large sample sizes:
I think trying to speed up these calculations for big data sets is a great
idea. There are two reasons why fitContinuous is slow - and it's very slow!
1. fitContinuous uses loops and other non-optimal algorithms. My code is not
optimi
Enrico,
This looks very interesting - mind if I play with it too?
Peter
On Nov 15, 2010, at 10:36, Enrico Rezende wrote:
> Manuela and David,
> perhaps you want to try the OU-transformation syntax I wrote to translate the
> MATLAB code developed by Blomberg, Ives & Garland. It wor
Manuela and David,
perhaps you want to try the OU-transformation syntax I wrote to
translate the MATLAB code developed by Blomberg, Ives & Garland. It
works well with small phylogenies, I never tried it in large ones and am
actually curious to see what would be the outcome.
You will find the
Hi Manuela,
I just ran the fitContinuous last night (lambda model) on a dataset of
1766 graptoloid species and their geologic durations. It took it more
than 12 hours to resolve, but it did come up with an answer, finally.
So, maybe your analysis was not permanently stuck.
However, I have seen fit
Hello,
I am trying to estimate the phylogenetic "inertia" in a variable
reflecting counts for 3975 mammalian species (with an ultrametric and
fully resolved phylogeny).
I can successfully obtain an estimate of K (as in Blomberg et al 2003)
using the command phylosignal (picante package), but I