Skip to content

Model failed to converge in bootMer

2 messages · Francisco Fabuel, Ben Bolker

#
I'm trying to estimate some quantities derived from a lmer object using
bootstrap. When I use the function bootMer, a large proportion of
replications seems to fail to converge, but if I program the bootstrap
using package boot, everything seems go well.

Please, find attached a simple example (R script and data) of the problem:
with bootMer in 10 simulations, I've got 7 warnings:

Warning messages:1: In checkConv(attr(opt, "derivs"), opt$par, ctrl =
control$checkConv,  :
  Model failed to converge with max|grad| = 0.00132463 (tol = 0.001,
component 1)


Thank you in advance.
Francisco Fabuel
2 days later
#
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
On 15-07-21 07:48 AM, Francisco Fabuel wrote:
Scripts and data generally get stripped from messages sent to the
mailing list; you may need to post them somewhere else and send along
a URL.

  This is a little surprising since bootMer internally calls boot, so
the devil must be in the details somewhere -- which goes back to the
fact that we do probably need to see the code (and preferably data)...

  Ben Bolker

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.11 (GNU/Linux)

iQEcBAEBAgAGBQJVsSW3AAoJEOCV5YRblxUHOnUIAJAiMOVPVdzo7j8ZBwSZ7fvf
rcNOrUQQyEk2gN6ym4i8uplSaM4nh5GHlxlWdsD4IGdJQ3HHBkGVEgpSQl8Q5Jrk
7sbx10WhEYKvuJfnbNyBYFzJvw+Gsm6hQqJsKUZGZrlUTgP/cuH0xLeZX/ljV+Qf
8Cm5KvK5oRUClr8GD3MVd0zC+ZryOpZyZaRM02i+IwfM6B1FGIe4vBrCp8XTK7ht
gixDTyx9UFjXHRKsU33XHbg9zA9D+eW7sX45XeyqyHYiwK0Yeb85ujx4Tdkv7PCB
yDhUXEyVDC8Z/5HFwh57EalJI6yizZ31KaX5z3QVapH+pVH7hjYlhtq9u3EVG+M=
=VEhi
-----END PGP SIGNATURE-----