search

ESCOMP / PhysicalConstantsDictionary

YAML dictionary of physical constants and tools to create consistent constant sets for Earth System Models
5 stars 0 forks source link

Uncertain about uncertainty #5

Open gold2718 opened 5 years ago

gold2718 commented 5 years ago

There seem to be different usages of the uncertainty keyword in the current YAML dictionary. First, the entry, uncertainty: exact is used for mathematical constants and physical constants such as speed_of_light_in_vacuum which have precise definitions. The other uncertainty term is relative_uncertainty (e.g., relative_uncertainty: 4.7E-05).

Why are there two different terms? For ease of use (i.e., parsing), why not just use one attribute name? They appear in the same column in the CODATA table.

MIredell commented 5 years ago

Perhaps the uncertainty and relative_uncertainty keywords are not needed? The number of digits given in the value keyword could be good enough to give the uncertainty to an order of magnitude.

For example, pi is listed as exact, but we of course cannot give its exact value. We give 25 digits of pi and so the relative uncertainty of that given value is by implication roughly 1E-25.

Another example is newtonian_gravitation_constant. The given value is 6.6740831E-11 but the given relative_uncertainty is 4.7E-05. We should just give the value as 6.674E-11. There's no need for extra uncertain digits and the uncertainty is implied to be roughly 1E-4.

gold2718 commented 5 years ago

I believe the number of digits used is identical to that used in the cited reference. My understanding is that the uncertainty is around the listed value, not around the truncated or rounded value. See the reference for the full uncertainty analysis.