[seqfan] Re: constants of nature in the OEIS

[Brad Klee]

Hi Andrey,

Outdated estimates do have important historical value. There is always an
interesting story associated, but history also provides a data set for
understanding the role of uncertainty in metrology. We have the following
from Randolf Pohl:

> https://www.quantamagazine.org/proton-radius-puzzle-
deepens-with-new-measurement-20160811/
>> http://pdg.lbl.gov/2012/reviews/rpp2012-rev-history-plots.pdf

Looking over the Berkely charts, there are a number of data points above 3
standard deviations relative to the most recent measurement. It looks like
there are also a few above 4 standard deviations, possibly one around 5
standard deviations. Given enough data, eventually we could see a 6
standard deviation error.

This logic affects your question about muon-electron mass ratio. What do
you mean by "known for sure"? Compare the following:

http://physics.nist.gov/cgi-bin/cuu/Value?mmusme
http://physics.nist.gov/cgi-bin/cuu/Info/Constants/definitions.html

Considering the historical data, it's possible that a deviation over 5
standard uncertainties will change the trailing 2 into a 3.

One possibility is to set a limit at the usual three standard deviations,
and mark a sequence dead when three sigma ranges do not intersect. If OEIS
put some time into this and expanded coverage, in the future it may be
possible to generate a plot such as above just from the version control
data.

Thanks,

Brad



On Fri, Jun 23, 2017 at 1:06 PM, Андрей Заболотский via SeqFan <
seqfan at list.seqfan.eu> wrote:

> Dear OEIS contributors,
> this is a question about some principles of the OEIS.
>
> OEIS has quite many entries for constants of nature, both dimensionless
> (such as the fine-structure constant A003673) and dimensioned (such as the
> electron mass A081801). Some of them can be found in the index
> https://oeis.org/index/Na . They are usually determined from experiments
> and thus are known only up to some error.
>
> As far as I understand the current consensus, if a constant is known only
> up to some error, only the digits that are known for sure should be
> included. Example: according to CODATA 2014, muon-electron mass ratio
> A057720 equals 206.7682826(46), so its true value is between 206.7682780
> and 206.7682872, so only digits 2067682 should be included in the
> sequence entry.
>
> Also, only the most recent estimate (confirmed by an authoritative source
> such as CODATA) should present: there should not be many different sequence
> entries for the same constant (such as A070064 and A081822, or A271369 and
> A248503), even though older estimate has been published. New estimates are
> published on a regular basis, so there's no reason to include all published
> estimates to the OEIS. (If an outdated estimate has some special historical
> significance, then it may be kept, of course.) Moreover, if the
> recommendation from the previous paragraph is followed, then there mostly
> will be no need for separate entries as the digits which were declared
> established usually don't change in updated estimates. If we remove the
> digits which weren't known for sure from A070064 and A081822, they will
> become the same.
>
> Am I right that these two recommendations are consensusal?
> If yes, one in every of two above-given pairs of sequence entries should
> be declared "dead", and probably these recommendations should be added to
> https://oeis.org/wiki/OEIS_format_for_decimal_representation_of_constants
>
>
>
> Best regards,
> Andrey
>
> A081801
>
> --
> Seqfan Mailing list - http://list.seqfan.eu/
>