Chapter 1 · demo series

The Diagnostic

What makes a light-curve embedding good? Separate template-matching from genuine per-source period reading.

Probe each frozen encoder to regress log-period, then split the result two ways. Overall R² looks impressive — but it conflates two very different skills. A model can score high just by recognizing the class and predicting that class's mean period (template-matching), without ever reading the period off the individual star. So we also measure ρ_within: the within-class rank correlation between predicted and true period, with the class effect removed. That is the part that requires actually reading absolute time.

overall period R² ρ_within (per-source reading) · dashed = Lomb–Scargle ρ_within ceiling
Probe:

Eight encoders + a classical Lomb–Scargle extractor on 25,565 ZTF variables. Long blue bar with a short purple bar = high R² that is mostly template-matching. Only Lomb–Scargle has both bars long. Switch to the MLP probe: R² rises (sharper templates) but ρ_within stays ≤ 0.31.

How to read it

High R², low ρ_within = template matching

Every self-supervised encoder lands here: R² of 0.5–0.7, but ρ_within ≤ 0.26. They identify the class and predict its mean period; they don't read the individual source.

The MLP probe sharpens templates, not reading

Switching to a nonlinear probe raises R² (e.g. BiGRU 0.66 → 0.77) while ρ_within barely moves (≤ 0.31). The period information isn't hiding nonlinearly in the embedding — it largely isn't there.

The signal is recoverable

A direct Lomb–Scargle extractor reaches ρ_within = 0.78 on the same light curves. The within-source period is there to be read; the encoders are under-reading it.

The operational definition. For classification, template-matching is enough — and that's why the foundation models look strong. For physics — period, and anything that depends on the ephemeris — "good" means high ρ_within. That single axis is what the rest of the series builds on: why time-in-the-input doesn't lift it, and where the shortcut finally breaks.

See also

Chapter 2 · Externalization Ladder ↗

Why pouring time into the input doesn't raise ρ_within — and what does.

Embedding Explorer ↗

The map: browse the embedding these numbers are computed on.