Error occurs with constant responses

[mynanshan]

I have the following example:

xx <- seq(0.01,1,0.01)
zz <- cbind(sin(xx*2*pi), cos(xx*2*pi))
y <- rep(1, 100)

grpreg::grpreg(zz, y, group = c(1,2), lambda = 10^seq(-3,20,1))
#> Error: Algorithm failed to converge for any values of lambda.  This indicates a combination of (a) an ill-conditioned feature matrix X and (b) insufficient penalization.  You must fix one or the other for your model to be identifiable.

^{Created on 2021-08-11 by the reprex package (v2.0.0)}

The grpreg throws an error rather than fit every coefficient to zero. What might cause such problem?

[pbreheny]

For linear regression, grpreg checks convergence by comparing coefficient changes relative to the standard deviation of y. This was coded as:

maxChange < eps*sdy

for a constant response, however, the RHS is zero, so convergence could never be reached. Easy fix, replacing the < with <=. Should be working now. Thanks for bringing this to my attention!