multiple line matches in H-like iso-sequence
n - n' lines in the H-like iso-sequence have nearly degenerate wavelengths and appear as a single line in most astronomical spectroscopy. But, when treated as nl resolved levels, there are a large number of these lines and we want the total intensity, since that is what is measured.
The decision was made to add the intensities of all of the n'l' - nl lines together and place that sum in the first of the lines, then set the remaining lines to zero. This works well since it is unambiguous, we normally do not print lines of zero intensity, and the line-matching algorithm, at the time, would settle upon the first match.
The line-match logic was changed to find the "best" match in the entire lines array. That then detects that there are many n'l' - nl lines with the same wavelength and emits this warning:
WARNING: multiple matching lines found in search for "H 1" 1.73621m
WARNING: match 1 is "H 1 1.73621m [ 10^2P-> 4^2S]" (dwl=-0.00195312A)
WARNING: match 2 is "H 1 1.73621m [ 10^2S-> 4^2P]" (dwl=-0.00195312A)
WARNING: match 3 is "H 1 1.73621m [ 10^2D-> 4^2P]" (dwl=-0.00195312A)
WARNING: match 4 is "H 1 1.73621m [ 10^2P-> 4^2D]" (dwl=-0.00195312A)
WARNING: match 5 is "H 1 1.73621m [ 10^2F-> 4^2D]" (dwl=-0.00195312A)
WARNING: match 6 is "H 1 1.73621m [ 10^2D-> 4^2F]" (dwl=-0.00195312A)
WARNING: match 7 is "H 1 1.73621m [ 10^2G-> 4^2F]" (dwl=-0.00195312A)
The auto test suite has ~15k of such lines:
cloud9 auto:grep "WARNING: match" *.out | wc
15284 167487 1450071
The warnings like dire and were a point of concern for the participants in the recent Warsaw workshop. They are about a design decision that was made in how we treat these nearly energy-degenerate lines. We should either figure out a way to silence the warning (do not do it for the H-like iso sequence) or disable this check in release versions.