what to do when half an asymmetric error is >1

[edbennett]

Currently, when formatting with abbreviation, when e.g. an upper error is larger than 1 but the lower is smaller than one, then you get a weird situation where the one without a decimal point doesn't visually match the one with one.

For example, format_multiple_errors(1.234, 0.056, (1.078, 0.091), abbreviate=True, latex=True) gives $1.234(56)({}^{+1.078}_{-91})$.

Should both errors have the decimal made explicit in such cases? (I.e. should the above instead look like $1.234(56)({}^{+1.078}_{-0.091})$?)

[chillenzer]

Yes, definitely! The other one looks totally confusing!

[edbennett]

Darn, that means a rethink of the implementation, as I can no longer just map the same thing to every number.

[chillenzer]

I can give that a thought once I actually looked at the code.