Abstract Mass ( M ) and luminosity ( L ) are fundamental parameters of stars but can only be measured indirectly. Typically, effective temperature ( T eff ), surface gravity ( <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" overflow="scroll"> <mml:mi>log</mml:mi> <mml:mspace width="0.33em"/> <mml:mi>g</mml:mi> </mml:math> ), and metallicity ([M/H]) are derived from stellar spectra, then M , L , and stellar age ( t ) can be obtained by interpolating in the grid of stellar evolutionary models. In this paper, we use random forest in combination with the evolutionary grid from PARSEC 1.2S to determine M , L , t , and initial mass ( M i ) for early-type main-sequence stars ( T eff ≥ 7000 K) as identified from the LAMOST survey. The convex hull algorithm is employed to select the main-sequence stars. The results demonstrate that the prediction precision is 32% and 9% for L and M , respectively, which is comparable to that achieved by fitting evolutionary tracks. Furthermore, the predicted L also aligns with Gaia’s observations, with a relative difference of 36%. The prediction for t is relatively less accurate, indicating a difference of 0.44 Gyr for two components in wide binaries. This discrepancy is due to the inconsistent metallicity measurements. For the two sets of atmospheric parameters we used, the relative differences in L , M , t , and M i are 29%, 7%, 36%, and 7%, respectively. The influence of metallicity on these parameters is analyzed, with the conclusion that metallicity has the greatest impact on t . Consequently, two catalogs are presented, which could be useful for studying stellar populations such as the luminosity function and initial mass function of early-type stars.

Mass, Luminosity, and Stellar Age of Early-type Stars from the LAMOST Survey