I have a question. As you know when you specify that a step should consider nonlinear geometry (enabling NLGEOM) the strain rates would be then true strain or logarithmic strains rather than simple or engineering strains. I have a model with which I test a vumat routine that Ive developed. When I use nlgeom I encounter some problems regarding convergency which do not exist when NLGEOM is disabled. My question is that should my vumat code be something different than what I use as small strain without nlgeom. I thought maybe I should use another definition for updating the stress as the strain is now different from when nlgeom is not used. Any suggestion is appreciated.