This section has some important "gotchas", in particular the impossibility of getting extra precision from a float:

https://mpmath.org/doc/current/basics.html#providing-correct-input

Current docs: https://mpmath.readthedocs.io/en/latest/basics.html#providing-correct-input
// edit by @skirpichev

@asmeurer suggested:

Perhaps mpmath should have a short "gotchas" page in the docs about this, similar to the one in the SymPy docs).

See #632 for a discussion and #670 for some recent changes.