Skip to content

Radius/Luminosity Plots

These are bar charts of yield, broken out by planet type. Each bar is one of the 15 radius/insolation bins -- five radius classes, each split into hot, warm, and cold -- averaged over the ensemble, with an error bar giving the spread across simulations. They answer "what kind of planets did this mission get, and how many."

The data comes from reduce-radlum.csv (the binned histograms) and reduce-earth.csv (the focused tally for exo-Earths), both written by the reduction code reduce_drms.py. Plots are named det-radlum-*.png and appear in the Radius/Luminosity Plots section of the ensemble webpage.

Reading the axis

The x-axis runs through the 15 bins in a fixed order: five groups of three, one group per radius class, each labeled Hot / Warm / Cold. The pale gray stripes behind the bars mark the radius groups, since only the insolation label is written out.

The leftmost bar is set off in green, labeled with the short name of the earthlike class, and counts that class separately. The class is a region of the plane rather than a bin, so it cuts across the grid and its count is not the sum of anything to its right -- see Tailoring the Earthlike Planet Class for where its bounds come from and how to move them.

Bars are stacked wherever a count has two components, and the inset text block on each plot says what the segments mean.

Unique, All, and Strict

Three of these plots count characterizations, and they differ in two separate ways: Repeats and Bands.

Repeats. A planet characterized on three separate visits can count once or three times.

  • Unique counts each planet once, on its first successful characterization. The reduction keeps a running set (set_chars_uniq) and only tallies planets not already there.
  • All counts every successful characterization, repeats included. These are the xchar quantities in the reduction, parallel to xdet for detections.

Bands. An observation may characterize in several spectral bands at once, and EXOSIMS reports band-by-band status: 1 for a full characterization, -1 for a partial char within that band. Here's how we manage that detail:

  • Full means status 1 in any band. A planet fully characterized in blue but only partially in red still counts as full.
  • Partial means all status was -1. This is mutually exclusive with Full.
  • Strict means status 1 in every band. The reduction accumulates charizations_strict across bands of that observation and tallies only planets that scored in all of them.

So the three plots are:

Plot Repeats Bands
radlum-char unique full (lower segment) + partial (upper segment)
radlum-char-all with repeats full + partial, stacked the same way
radlum-char-strict unique all bands full; no partials at all

Note that -all and -strict differ in both ways, so they are not a nested pair. The nesting that does hold (see the reduction code, within the unique counts for the earthlike class):


    exoE_char_strict  <=  exoE_char_full  <=  chars_earth_unique
                                           =  exoE_char_full + exoE_char_part

Note, the rightmost term (exoE_char_full + exoE_char_part) is the total that the plain radlum-char plot shows.

Two consequences worth noting:

  • "Strict" offers nothing special for a single-band scenario, so radlum-char-strict equals the "full" segments of of radlum-char, and the plot carries no extra information.
  • Strict characterizations have no SNR attached -- the SNR differs between bands, so the reduction records none. Thus, the SNR plot has no strict variant.

Plot Summary

Seven, always produced; this family has no extras.

Plot Shows
radlum-population occurrence rate: planets of each type per star, over the whole target list
radlum-det unique detections, split by detection mode (blue / other)
radlum-det-all detections with repeats, same split
radlum-char unique characterizations, full + partial stacked
radlum-char-all characterizations with repeats, full + partial stacked
radlum-char-strict unique characterizations, full in every band
radlum-char-snr mean characterization SNR per bin, full chars only

radlum-population is the sanity check of the set: it is a property of the Simulated Universe and the target list, not of the schedule, so it is where you confirm planets are being generated at the intended rate. Its leftmost green bar is the observed eta for the earthlike class.

Clarifications

  • Error bars are +/- 1 sigma across the ensemble, not standard errors of the mean. They describe how much one simulation differs from another. Divide by the square root of the ensemble size to get the standard error of the mean.
  • The population plot is normalized per star; the yield plots are counts per simulation. Do not read a throughput by dividing one plot by another by eye -- the binned throughputs live on the radius/SMA bin plots instead.
  • Missing reduction data skips a plot rather than failing. Summary CSVs from years-old reductions may lack some quantities. The driver says which plot it skipped and why.