If glm in dlogr.init does not converge, the coefficients can be infinite and thus, the covariance very large. That can break the algorithm.
Make a loop in logistic.dma.default: If initialsamp is not given and the glm did not converge, increase the sample size (from 1/10 to 2/10 etc)
If glm in dlogr.init does not converge, the coefficients can be infinite and thus, the covariance very large. That can break the algorithm.
Make a loop in logistic.dma.default: If initialsamp is not given and the glm did not converge, increase the sample size (from 1/10 to 2/10 etc)