Skip to content

More floating point theorems #1082

Open
@michael-roe

Description

@michael-roe

(Feature request)

Add proofs of the following floating point theorems:

!x . not float_equal x x <=> float_is_nan x

!x . float_unordered x x <=> float_is_nan x

!x . (float_equal x (float_plus_infinity (:'t # 'w))) \/ (float_equal x (float_minus_infinity (:'t # 'w))) <=> float_is_infinite x;

!x . float_equal x (float_plus_zero (:'w # 't)) <=> float_is_zero x

These theorems prove that the typical implementations of fp_is_nan etc. (if you don't have fp_classify) are correct.

Metadata

Metadata

Assignees

No one assigned

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions