The test runs with 2.4.0 revealed a computational inconsistency for objects with broad lines that this PR fixes. Briefly, I was originally doing three rounds of emission-line fitting:
First round with narrow lines only with forbidden and Balmer lines tied together;
Second round with narrow plus broad lines forbidden and narrow Balmer lines tied together and broad Balmer lines (separately) tied together;
Final round (after choosing between narrow or narrow+broad) where the forbidden and narrow Halpha lines were optimized separately.
The problem with this algorithm is that the final round comes after the broad vs narrow+broad decision has been made so the final numbers don't alway match the numbers used to make the decision. For example, the broad Balmer lines have to be at least 250 km/s, but there are objects that have line-widths less than this threshold because the in the final round the algorithm decides that it likes a narrower line!
So in this PR we only do two rounds of fitting and keep the forbidden and narrow Balmer lines separate in both rounds. Also note that the speed-up is 10%-20% because of the fewer rounds of line-fitting!
I also fixed a few more small bugs and more provenance information is written to the headers such as the name of the emission line and photometry parameter files.
Once tests pass I'm going to tag and regenerate the Fuji VAC.
The test runs with
2.4.0revealed a computational inconsistency for objects with broad lines that this PR fixes. Briefly, I was originally doing three rounds of emission-line fitting:The problem with this algorithm is that the final round comes after the broad vs narrow+broad decision has been made so the final numbers don't alway match the numbers used to make the decision. For example, the broad Balmer lines have to be at least 250 km/s, but there are objects that have line-widths less than this threshold because the in the final round the algorithm decides that it likes a narrower line!
So in this PR we only do two rounds of fitting and keep the forbidden and narrow Balmer lines separate in both rounds. Also note that the speed-up is 10%-20% because of the fewer rounds of line-fitting!
I also fixed a few more small bugs and more provenance information is written to the headers such as the name of the emission line and photometry parameter files.
Once tests pass I'm going to tag and regenerate the Fuji VAC.