Monte Carlo sampling to calculate metallicity uncertainty from strong line flux data.
All the results will be saved in this direcoty by the name of the input file (e.g. exampledata for input exampledata_meas.txt and exampledata_err.txt), which will be created upon running the code in the directory provided by the --path argument.
For a given file and nsample, the following output files are generated:
<fname>_n<nsample>.pkl": (e.g. exampledata_n2000.pkl) the metallicities and their uncertainties calculated and stored in binary (pickled) format in a python dictionary
<fname>_boxplot_n<nsample>_<measurement>.pdf": (e.g. exampledata_boxplot_n2000_1.pdf, exampledata_boxplot2000_m2.pdf) box and whiskers plot showing all metallicities and their uncertainties in a single plot, one plot (and one PDF file) for each measurement in input
"hist" folder, containing all the histograms generated. For each scale calculated, scale <scale_name>, and each input set of lines, or measurement, the distribution histogram is saved in a PDF file <fname>_<nsample>_<scale>_<measurement>.pdf (e.g. exampledata_n2000_KD02_N2O2_1.pdf for scale KD02_N2O2, measurement 1)
if the keyword --asciiout is set, a text file <fname>_n<nsample>_<measurement>.txt (e.g. exampledata_n2000_1.txt) stores the distribution's 50th, 16th and 84th percentiles in the format: E(B-V) 0.121000 0.014000 0.019000 logR23 0.428000 0.003000 0.004000 ...
if the keyword --asciidistrib is set, a csv file <fname>_n<nsample>_<scale>_<measurement>.txt (e.g. exampledata_n2000_KD02_N2O2_1.csv) is created for each scale and each measurement, containing the full distribution of metallicities.
if the keyword --binmode is set to 'kd' (kernel density), the kernel density distribution is saved in a binary (pickle) file <fname>_n<nsample>_<scale>_KDE_<measurement>.pkl (e.g. exampledata_n2000_KD02_N2O2_1_KDE.pkl), having been created for each scale and each measurement. The kernel density is saved as a python sklearn KernelDensity object.
'-1' means: the emissions lines don't satisfy some sort of minimum criteria to derive metallicity
0.000 (no distribution) means no distribution was generated for some reason: you may have selected nsample=0, or have errors = 0.0, and that is the appropriate behavior in this case. Otherwise the distribution may be really screwed up and take a single value, or a shape such that the 16th and 84th percentiles cannot be calculated (e.g. two values), and you should definitely not trust the output.