Make sure that the maximization procedure worked properly. For
example, make sure that eiread's retcode is zero. Is the
variance matrix from the normal approximation reasonable? Most
importantly, check to make sure the simulations are coming from the
mean posterior contours by comparing eigraph's tomogp to
estsims (these are both in tomogS); verify that
eigraph's post gives unimodal distributions. Was the
importance sampling successful? Check to make sure that eiread's
resamp is not much larger than 20 or so (it is not fatal if
it is larger, but its worth checking).
