01 — Does the exoplanet radius valley move with stellar age?
Supplied working report; original verdicts not validated.
Review status: Needs review. This is an archival source note, not a passed proof.
README.md
# 01 — Does the exoplanet radius valley move with stellar age? **Result: the valley is at 1.69 Earth radii in systems younger than 1 Gyr and 1.86 in systems older than 4 Gyr — a 0.18-radii shift (95% CI 0.05–0.35). The valley does not deepen with age; if anything it is shallower in old systems, though the interval includes zero.** ## The proposal The notebook argues that the 1.8-Earth-radius gap in the planet size distribution is a fossil: photoevaporation strips envelopes only while the star is young and the disk has dispersed, so the gap should record *when* that happened, and should therefore change with stellar age. Under the steady-state reading it is a fixed feature of the physics of mass loss and should sit at the same radius at every age. ## What was tested 6,125 confirmed transiting planets from the NASA Exoplanet Archive with radius between 0.5 and 5 Earth radii, orbital period under 100 days, and a stellar age in the catalogue. 2,019 host systems. 868 planets (577 systems) younger than 1 Gyr; 3,092 (1,416 systems) older than 4 Gyr. The radius distribution of each group was smoothed with a 0.02-dex Gaussian kernel, and two numbers were taken from each curve: the location of the minimum between the two peaks, and the depth of that minimum relative to the peaks on either side. Intervals come from resampling host systems, not planets, so a system with five planets cannot masquerade as five independent measurements. ## What happened | | young (< 1 Gyr) | old (> 4 Gyr) | |---|---|---| | planets / systems | 868 / 577 | 3,092 / 1,416 | | valley radius | **1.69 Re** (1.60–1.78) | **1.86 Re** (1.77–2.00) | | valley depth | 0.40 (0.29–0.56) | 0.30 (0.22–0.41) | | median period | 10.4 d | 10.1 d | The location difference is 0.18 Re with a 95% system-bootstrap interval of 0.05 to 0.35: **P1 is supported**. The depth difference is −0.10 with an interval of −0.26 to +0.05 and a label-permutation p of 0.72: **P2 is not supported**, in the direction the proposal expects. The location result is not an artefact of the smoothing width: at 0.015, 0.02, 0.03 and 0.05 dex the young valley sits at 1.72, 1.69, 1.64 and 1.60 and the old valley at 1.86, 1.86, 1.87 and 1.82. One planet per system gives 1.73 versus 1.95. The shift is the same sign and similar size across all four choices. One sensitivity fails: restricting to periods of 10–100 days pushes both curves against the lower edge of the search window (1.603 Re for both groups). That subset is dominated by planets near the window edge and cannot locate the valley; it is reported rather than hidden. ## What it means The gap is not at a fixed radius. That is what the fossil reading predicts and what a strictly steady-state model does not. But the second half of the proposal — that the gap *deepens* as the transition completes — fails: the old population shows the shallower deficit, and the interval on the difference is wide enough that only the direction is informative. A shift of 0.18 Re is real in this catalogue and small in absolute terms, and the catalogue is not a clean instrument. Ages for young stars come disproportionately from clusters and moving groups; ages for old field stars come from models with systematic errors of tens of per cent. The two age groups also differ in how they were found: young hosts are active, photometrically noisy, and biased against the detection of small planets, which pushes the measured distribution to larger radii — the opposite of the observed shift. That makes the sign of the result harder to dismiss than the size of it. ## What this test cannot see It cannot see the high-energy output history that P3 requires, so it cannot say whether the shift tracks X-ray dose or something else. It cannot separate a genuine time evolution from a difference in the planet populations that happen to orbit stars of different ages. It cannot see non-transiting planets, so the radius distribution is conditional on transit geometry. And the archive's radii are a heterogeneous mixture of measurements and mass-radius inferences, which is why the curves are as lumpy as they are. Preprint draft · not peer reviewed · designed and executed with AI assistance · numbers regenerate with `python3 run.py`.
hypothesis.md
# The proposal, quoted Source: Incoprea notebook, *Astronomy & Planetary Science* #12, "The Radius Valley Is a Fossil of Timing". > The radius valley is not a steady-state feature of atmospheric stripping but a fossil of when photoevaporation happened relative to disk dispersal, so its position and depth should change with stellar age. Stated predictions in the notebook: - **P1** — The valley radius should differ measurably between hundred-million-year and billion-year populations. - **P2** — The valley should deepen with age as the transition completes. - **P3** — The valley's slope should track the star's high-energy output history rather than a single efficiency parameter. Stated refutation: > A valley at the same radius and depth across ages, at fixed detection sensitivity, would restore the steady-state picture. P3 needs X-ray/UV histories for each host star, which the archive does not carry. It was not tested.