The INCAR file is the central input file of VASP. It determines "what to do and how to do it". There is a relatively large number of parameters that may be set by means of tags in the INCAR file. Most of these parameters have convenient defaults, and a user unaware of their meaning should not change any of the default values.
Be very careful about the INCAR file, it is the main source of errors and false results!
The INCAR file is a tagged format free-ASCII file: Each line consists of a tag (i.e. a string) the equation sign '=' and a number of values. It is possible to give several parameter-value pairs ( tag = values ) on a single line, if each of these pairs are separated by a semicolon ';'. If a line ends with a backslash the next line is a continuation line (beware of blanks after the backslash, some versions of VASP might not accept those). Comments are normally to be preceded by the number sign '#', but in most cases comments can be append to a parameter-value pair without the '#'. In this case semicolons should be avoided within the comment.
Lines with tabs are not properly read by VASP, please avoid using them in the INCAR.
A typical (relative complex) INCAR can be found below:
SYSTEM = Rhodium surface calculation # start parameters for this Run (automatic defaults are finem, hence not often required) ISTART = 0 # job : 0-new 1- orbitals from WAVECAR ICHARG = 2 # charge: 1-file 2-atom 10-const # electronic optimization ENCUT = 300.00 eV # defaults from POTCAR, but wise to include ALGO = Normal # alorithm for electron optimization, can be also FAST or ALL NELM = 60 # of ELM steps, sometimes default is too small EDIFF = 1E-06 # stopping-criterion for ELM # broadening and DOS related values; this works almost always SIGMA = 0.05; ISMEAR = 0 # broad. in eV, -4-tet -1-fermi 0-gaus # ionic relaxation EDIFFG = -1E-02 # stopping-criterion for IOM (all forces smaller 1E-2) NSW = 20 # number of steps for IOM IBRION = 2 # CG for ions, often 1 (RMM-DISS) is faster POTIM = .5 # step for ionic-motion (for MD in fs) # performance optimization KPAR = 4 # make 4 groups, each group working on one set of k-points NCORE = 4 # one orbital handled by 4 cores LREAL = A # real space projection; slightly less accurate but faster