Inconsistencies in Abaqus material

Hello,

i wonder, how the inconsistencies in anisotropic hyperelastic material of Holzapfel-Gasser-Ogden form caused by Macauley-brackets and Heaviside step-function are beeing solved inside Abaqus (cf. Analysis User's Manual 22.5.3).

The Heaviside H(x) is beeing used to allow fiber reinforcement only on extension (in fiber direction). But analytical derivatives of the Macauley-bracket yield terms with H(x), and the Dirac-delta (infinite impulse at x_0).

How would it be possible to include simliar behavior in a material subroutine, without encountering serious convergence problems??

Greetings