commit b7845e8128 introduced a new way of automatically handling the changing of the integration timestep on a CP restart.
the sio2-us-lda-2.in input have dt=15.0 and is a electron damping run, while
the sio2-us-lda-3.in input have dt=12.0 and should start with wfc velocities that are almost zero from the prev run
since the almost zero electronic velocities are rescaled (correctly) the initial condition for the lda-3 run are slightly different, producing slightly different results in the stress. The trajectories are with very good approximation almost equal.
The test suite was failing with the following:
cp_sio2 - sio2-us-lda-2.in: Passed.
cp_sio2 - sio2-us-lda-3.in: **FAILED**.
s1
ERROR: absolute error 1.44e-03 greater than 5.00e-04. (Test: -5.25740981. Benchmark: -5.25597336.)
ERROR: relative error 2.73e-04 greater than 1.00e-04. (Test: -5.25740981. Benchmark: -5.25597336.)
cp_sio2 - sio2-us-lda-4.in: **FAILED**.
s1
ERROR: absolute error 1.40e-03 greater than 5.00e-04. (Test: 3.97451018. Benchmark: 3.97590607.)
ERROR: relative error 3.51e-04 greater than 1.00e-04. (Test: 3.97451018. Benchmark: 3.97590607.)
- removed (almost) duplicated xminus1 and xminus1_state routines (it was almost a copy of calphi, that was generalized really a little to cover the case of xminus1)
- added tests for CG us case, and some CG dynamics with cp.x
- fprettify on cg_sub.f90
- added pre_state input variables in electrons namelist to control cp's CG routine preconditioning a little bit. If pre_state is .true. the preconditioning includes the kinetic energu factor for that band. Before it was an hard coded variable
Most of the changes consist in small updates to Makefile and make.depend and
in the replacement of module "parameters" with "upf_params" whenever variables
lmaxx and lqmax are USEd. Routine "read_pseudo" moved back to Modules/;
"simpsn" moved to upflib/; a copy of "matches" temporarily added to upflib/
Note that everything compiles (except EPW) but nothing has been tested yet!
Paolo Giannozzi