A NumericalPoint is a collection of floating point value with the better precision the platform can support. In many i386 platforms the typical floating point type is double. But it can be set to long double on ia64 architectures.

The NumericalPoint can be seen as a vector of this type, and support the common accessors and operators. It can also be mixed with Matrix or Tensor objects in various operations.

A NumericalPoint can be saved or reloaded through the Study.