Radius/SMA Bin Plots
These plots put the same 5x3 planet classes on the radius vs. luminosity-scaled-SMA plane, drawn to scale. Where the radius/luminosity bar charts show how many, these show where -- each bin is a rectangle in its true position, carrying its number inside it.
Two different programs contribute to this section, which is worth knowing because they behave differently:
util/rad-sma-rectangle-bin-plot.py, driven byutil/rad-sma-rectangle-plot-driver.sh, draws the binned tables fromreduce-radlum.csvandreduce-earth.csv. These are written as both.pngand.pdf.plot_drm_rad_sma_chars.py, one of the gallery modules, draws the scatter and density of individual characterizations fromreduce-earth-char-list.csv.gz. PNG only.
All are named det-rad-sma-*.png and appear in the Radius/SMA Bin Plots
section of the ensemble page.
The plane
The x coordinate is the luminosity-scaled semi-major axis, a / sqrt(L), not
the raw orbit. That is the coordinate the bins are defined in, and it is what
makes planets around different stars comparable. Both axes are logarithmic.
Drawn behind the data are the 5x3 bin rectangles and the outline of the
earthlike class. That class is a region rather than a bin, so it straddles bin
boundaries. The earthlike bin placement, and its name in the titles, can be adjusted by
creating a configuration file, config-reduce.json. See
Tailoring the Earthlike Planet Class.
Counts and throughputs
The binned tables come in two kinds, and the distinction is the same one the radius/luminosity plots make between full and strict:
- Counts (
char_full,char_strict) -- the mean number of unique characterizations per simulation falling in each bin. Full means the planet scored a full characterization in at least one spectral band; strict means it scored one in every band. Strict is always a subset, so its numbers never exceed the full ones. - Throughputs (
char_tput_full,char_tput_strict) -- the proportion of the planets present in that bin that were characterized. The reduction forms this as
h_RpL_char_tput_full = h_RpL_char_full / (Nstar * h_RpL_population)
where h_RpL_population is the per-star occurrence rate, so the denominator
is the expected number of planets of that type across the target list.
A throughput is a rate on an absolute scale, which makes it the right plot for comparing scenarios; a count depends on how many planets the universe happened to generate. When a bin holds no planets, the throughput is 0/0 and shows as NaN rather than zero -- an empty cell means "that planet type was not present", which is a different statement from "the mission reached none of them".
To see throughput as a smooth function of Radius and SMA, rather than per-bin, see the Radius/SMA Density plots.
The plots
Seven, always produced; this family has no extras. The first five are the binned tables, and are written as PNG and PDF both.
| Plot | Shows |
|---|---|
rad-sma-population |
occurrence rate per bin: planets of that type per star |
rad-sma-char-full |
mean unique characterizations per bin, full in any band |
rad-sma-char-strict |
the same, requiring full in every band |
rad-sma-char-tput-full |
characterized / present, per bin |
rad-sma-char-tput-strict |
the same, strict |
rad-sma-scatter-point |
one point per successful characterization, pooled over the ensemble |
rad-sma-scatter-density |
the same points as a kernel density estimate |
The two scatter plots are drawn from the characterization list rather than the binned histograms, so they show individual events and are not averaged per simulation. They pool every simulation in the ensemble, which is why their point counts are much larger than the per-bin means beside them.
Clarifications/Caveats
- The binned tables are per-simulation means; the scatter plots are cumulations across the ensemble.
- NaN is not zero. See above -- it marks an absent planet population, and the distinction matters most in the corner bins.
- Strict characterizations carry no SNR, since SNR differs across bands. Nothing in this section is SNR-weighted.
- The bin grid is fixed at 5x3. It can be moved via the
RpL_binsgroup ofconfig-reduce.json, but not reshaped; the reduction and both plotting programs assume that shape.