A new set of gratings with medium resolution (R ∼ 7500) has been mounted on the LAMOST spectrographs, and the wavelength windows range in 490 ∼ 540nm and 640 ∼ 690 nm respectively for blue and red spectrograph arm. Commissioning observation has been conducted to test the survey based on 16 spectrographs and 4000 fibers. Meanwhile, a spectral analysis pipeline has been developing to get more precise stellar parameters, radial velocities and abundance of chemical elements. Instrument profiles are calculated for each fiber at each exposure according to emission lines both from arc lamp. A template grid spectra with R ∼ 7500 for fundamental parameter (Teff, logg, and [Fe/H] ) are selected from Elodie. During the commissioning observation, each star have been visited for several times, and a fraction targets include APOGEE, Kepler and PASTEL objects which have high precisely measured parameters. With the commissioning spectra, we can understand instrument performance, intrinsic precision of repeat observations, and the accuracy of the pipeline.
Referring to SDSS/SEGUE pipeline for stellar parameters SSPP and other pipelines, two methods,
ULySS<sup>9,10</sup> and CFI (correlation function interpolation) are investigated to estimate stellar parameters
(effective temperature, surface gravity, and metallicity) for AFGK stars based on medium-resolution
spectroscopy. Both of the two methods carry with an interpolator, ULySS provides with an
interpolator of the template library consisting of polynomial expansions of each wavelength element
in powers of stellar parameters while CFI interpolates the maximal correlation coefficent as
functions of stellar parameters. Comparing with known objects observed by the Sloan Digital Sky
Survey (SDSS), their performances are tested, random and systematic errors are examined. By
comparing CFI with ULySS, performances of different interpolations are tested. These two methods
will be integrated into LAMOST stellar parameter pipeline (LASP) and used for the data release of
the LAMOST pilot survey.