Known second periods improve later Gaia G predictions for 114 of 148 selected LMC Cepheids with single-mode labels

We test fixed external periods on public Gaia DR3 G measurements of Large Magellanic Cloud Cepheids. We match the OGLE first/second-overtone catalogue to Gaia Cepheids and select 148 classical Cepheids with a Gaia FIRST_OVERTONE label. We fit earlier measurements and score later measurements. Each time group has equal total weight. The baseline has a constant, a linear time term, and three harmonics of the first period. The second model adds a sinusoid at the known second period. It lowers test error for 114 of 148 stars. The median reduction in mean squared error is 34.8%, with a 95% whole-star bootstrap interval of 23.6% to 40.4%. A reference sample of 38 Gaia MULTI stars has a median reduction of 45.9%. All 100 sets of comparison frequencies have a lower median than the known second periods in each sample. The main-sample median stays positive under the tested changes to harmonics, weights, time split, and periods. Of 112 main-sample stars with fewer than 40 clean measurements, 88 improve. The study is exploratory. A three-star pilot led to lower minimum data counts. The external periods can use observations outside the early Gaia interval, so this is a conditional prediction test. It is not a prospective period search or a completeness estimate. Missing double-mode labels were already known. We provide a new benchmark, fitted coefficients, source-file hashes, selection records, comparison scores, figures, and code. We report no new star or pulsation mode.

Cepheids · Gaia DR3 · OGLE · Large Magellanic Cloud · stellar pulsation · double-mode Cepheids · time series · prediction · reproducible research