Ground-State Transition Frequencies for Rb 85,87 are Incorrect

[pbanner]

The code line Rubidium87().getTransitionFrequency(5,0,1/2,5,1,3/2) produces a transition frequency number that is about 130 MHz off from the number in Dan Steck's data (which my lab experimentally agrees with).

The library also gives the same results for Rb 87 vs 85, when these are different by tens of MHz according to Dan Steck's data for 85Rb. With 85Rb, the 5P3/2 hyperfine splittings generated by getHFSCoefficients() also appear to be incorrect.

I got down this after tracking a discrepancy between experimental observation in my lab and 6p-ns transition frequencies, so the discrepancies may go beyond the n=5 manifold.

[nikolasibalic]

Hi @pbanner , thank you for opening this question.

Low levels are extracted from NIST data as indicated in the paper, and this limits expected accuracy there. Your observation is consistent with expected relative accuracy stated in TableB2 in the paper (so absolute accuracy is indeed ~100 MHz for these transitions, but better for Rydberg transitions). Depending on what you do, you might require higher relative accuracy, in which case its best to check latest experimental data.

Regarding getHFSCoefficients when you call

from arc import *
Rubidium85().getHFSCoefficients(5,1,1.5)

you get (25002000.0, 25790000.0) in Hz for A and B which is exactly what is in Steck reference, table 5. Could you please verify this, and let me know if you spot some error?

[pbanner]

Thank you for this response!

I was unable to reproduce the HFS splitting error I reported earlier—checking it again today, all hyperfine splittings look correct for 5s and 5p in both Rb 85 and Rb 87. Apologies for the confusion there.

I see the accuracies given by NIST, so this makes sense. Is there a "built-in" way to ask ARC to use particular values for given transition frequencies, or modify the database from which it searches?