#
# BULK SODIUM
#
# Enter PARATEC or ESPRESSO directory.
#
# ESPRESSO: generate k-points (directory 00); you must manually add the k-points to espresso input files
ESPRESSO: sh ./script_0
#
# create symbolic links
#
sh ./script_1
#
# generate charge density (directory 01)
# run nscf calculations (directories 02 to 04)
# generate dielectric matrix (directory 07)
# calculate self-energy corrections (directory 08)
#
Submit script_2. Suggested ncpu = 64, walltime = 0:15:00
#
# A couple of notes about the differences between this and a semicondutor 
# calculation:
#
# 1). Note that instead of using a k-grid shifted by a small q-vector 
# we use a uniform, unshifted fine grid to represent the small q 
# behavior in 03-wfn_fi. This  changes way that we calculate epsilon. 
# Note that in 07-epsilon we calculate epsilon on a uniform grid 
# but exclude the first small q point that is given in a calculation 
# on a semiconductor (see silicon example). The first point in the 
# qpoint list in 07-epsilon/epsilon.inp does NOT have the '1' flag at the end. 
# In 07-epsilon_fine we calculate a single q point of size 1/gridsize where 
# the gridsize is the linear dimension of the grid size specified in the 
# 03-wfn_fi calculation.
#
# 2). For the calculation of Sigma, we link the epsmat file from 07-epsilon 
# to epsmat in the 08-sigma directory. But now we link the epsmat file 
# from 07-epsilon_fine to eps0mat in the 08-sigma directory.
#
# 3). The fine grid should be dense enough to adequately represent the 
# intraband transitions as well as the density of states. If you plot 
# the head of inverse epsilon as a function of the small q-vector, 
# you should see a parabolic behavior which agrees very well with 
# a Thomas-Fermi model of the system calculated from the density 
# of states at the Fermi level.
#
# 4). We cannot run the BSE codes on sodium because the unusual case 
# of a system in which some k-points have no occupied bands is not 
# implemented. We could overcome this limitation by using a semicore 
# pseudopotential for sodium at the price of a higher computational 
# cost.
