CMS luminosity precision in 2017 proton-proton collisions at LHC at $\sqrt{s}=13$ TeV

The calibration for the luminosity measurement by the CMS experiment for the 2017 proton-proton LHC run at $\sqrt{s}=13$ TeV is described. To guarantee smooth and uninterrupted luminosity measurements the CMS experiment is equipped in Run II with three online luminometers: the Forward Calorimeter (HF), the Pixel Luminosity Telescope (PLT), and the Fast Beam Condition Monitor (BCM1F). For the offline luminosity measurement and a cross check of the online detectors the pixel detector is used (Pixel Cluster Counting, PCC). The principal calibration is derived from the analysis of the Van der Meer scan program in CMS taken during LHC fill 6016. In addition, the performance and stability of the CMS luminometers are also evaluated using emittance scans taken throughout the course of the year. The systematic uncertainty in the absolute calibration from the Van der Meer scan is derived with a precision of 1.5%, with the dominant contributions arising from the x-y correlations of the beam shape and the scan to scan variation. The total systematic uncertainty, including terms from stability and linearity effects, is 2.3%. For a low pile-up portion of data at the end of 2017 the calibration uncertainty was lower (1.7%) because of lower impact from linearity uncertainty.