Dear Tomasz, We are aware of the short-comings of file-outputs throughout the code. These are legacy fixed fields which does require revisiting.
We highly welcome contributions to the public Siesta repository: https://code.launchpad.net/siesta Since there are *many* places around the code which has these inadequacies it would be nice with external contributions. Feel free to help us by creating a merge-request! :) 2018-04-18 22:48 GMT+02:00 tkos <[email protected]>: > Dear colleagues, > > siesta-4.1-b* seem to produce erronous *.EIG output (*** instead of kp > integers) if we use too many k-points (100000+, eg.60x60x60 in my case). > In result the *.EIG file cannot be used to obtain DOS with Eig2DOS. > It can be corrected by the following modifications in ioeig.f (Src > subdir), e.g.: > line: write(iu,"(3i6)") no, nspinS, nk > change to: write(iu,"(3i7)") no, nspinS, nk ! or e.g. "(i6,i1,x,i7)" > line: write(iu,"(i5,10f12.5,/,(5x,10f12.5))") > change to: write(iu,"(i6,10f12.5,/,(5x,10f12.5))") > > I am also not sure if the format for weights in *KP lines is correct (is > this accuracy sufficient for large nk? Maybe better use exponential format? > > Best wishes, > Tomasz > > > > -- Kind regards Nick
