Geometry optimizations with FHI-aims against ASE:
----------------------------------------------------------------------
Note, that the results and accuracy of the geometry optimization routines have 
been tested against the ASE using FHI-aims and hartree fock with a strict 
convergence criteria. The tests have been conducted by Eszter Pos (2020) 

The following routines are tested and give perfect agreement:
	- BFGS
	- L-BFGS
	- BFGS-TRM
	- CG: there is an issue about the line-search of the CG routine, the
	      function needs to be evaluated ~100 times, while with ASE , the routine
	      converges after 7 steps. Needs to be checked further. Final geometry agrees
	      well enough with the ASE geometry, but the linesearch needs to be checked and
	      optimized. 

Partially tested, but problematic routines with FHI-aims against ASE:
	- SD: could not be tested against ASE, however, it gives the same final
optimized geometry as the BFGS routines. (takes a very long time becasue of
the line search.)
        - NEB: the Nudged Elastic Band routine is disabled until further
               debugging is conducted on the routine and its spring constants. 
	


