.. _model-global_surrogate-function_train-arithmetic_tolerance: """""""""""""""""""" arithmetic_tolerance """""""""""""""""""" A secondary rounding tolerance used for post-processing **Specification** - *Alias:* None - *Arguments:* None - *JSON argument key:* None - *Default:* 1e-10 **Description** After forming one or more function train approximations, this tolerance is used for rolling up final results, in particular, combining multilevel/multifidelity expansions and forming products for higher-order moment estimation. It is generally intended as a looser tolerance than ``rounding_tolerance``. *Default Behavior* If this tolerance is not specified, then we adopt the internal C3 default.