CHANGES OF VERSION 5.1.11

1.  Added reference information and cleaning up

2.  BUGS

3. INPUT RELATED CHANGES

4. OUTPUT related changes

5. ALGORITHMS: additions or changes  to existing functional ability

 

1. Added reference information and cleaning up

   ·    added Tutorial directory

   ·   moved doc_src to doc_purpose, and rewrote doc_src to a give a short description of all files. also added to description to a few files

   ·   got rid of the unused pwp.f90. not used after 5.1.1

 

2. BUGS

     ·    fixed problem with randomization of the startvector on Seaborg.

      If the array was too big then the call to random_number hung up

 

  ·    changes check for max difference in potential in etotal_dir to ">=".

      the call from cgemin_gcg uses vin for both potentials so idvmax never

      was set and stayed -1 causing a memory problem.

 

·    changes check for max difference in potential in etotal_dir to ">=".

      the call from cgemin_gcg uses vin for both potentials so idvmax never

      was set and stayed -1 causing a memory problem.

 

·   Corrected  the MFLOPS OUTPUT in the matrix-matrix multpiply in cgemin_gcg

 

·    zeroed out energs%esmear before direct direct minimization routines

      as it is zero, and never is accessed. On 1 processor there were errors

      due to not being zeroed.

 

3. INPUT related changes

 

·     added keyword min_iter_diag as equal to max_iter_diag

 

4. OUTPUT related changes

  ·  added ability to plot potential of spin and down

 

  ·    added output of vxc0(2) to flevel

 

  ·    added options to output potential from scf job

 

·   added option to not output or do forces and stresses

      output_flags nofrcstr

 

·   added option to not output CD, potential, or wavefunctions ever

      output_flags noioscf

 

·   changed the print out of the error check in read_pseudo for the real

      space NLPP. Also made assignment of temp var. mrb2_node to corresponding

      variable in pspot.

5. ALGORITHMS: additions or changes  to existing functional ability

  ·    Fixed pathological problem in Grassmann_metal algorithm.

         If one k-point has much worse convergence than other k-points it will

         have zero occupations. This prevented further WF improvement since occupations

         are zero for that k-point. Fix: occupy half of the bands by default as the weighting

         in Grassmann_metal