Message-ID: <loom.20070302T160517-105@post.gmane.org>
Date: 2007-03-02T15:07:20Z
From: Gregor Gorjanc
Subject: large dataset - lmer2 vector size specified is too large
Douglas Bates <bates at ...> writes:
...
> but even that formula means that you are estimating 448 fixed effects
> for which the model matrix is of size 448 * 43680 * 8 bytes (about 150
> MB). In addition you are attempting to estimate
Not directly related. Is it really necessary to build model matrix i.e. X.
Building X'X directly would be more efficient i.e. only 448 * 448 * 8 bytes.
Gregor