Skip to content

Draft: Add Brute-force algorithm to estimate better degree and absorbing lengths for leads

Adel Kara Slimane requested to merge AdelKS/tkwant:absorbing-algo-update into master

The starting point of this changed is the following discovery on automatic_boundary() in leads.py:

  • On a given fixed system.
  • Calling automatic_boundary() with tmax=500: it doesn't find a good absorbing boundary and then says to use the simple one, the lead size is 8000
  • Calling automatic_boundary() with tmax=5000: it finds a good absorbing boundary with a total lead size of 2600

That means that it is still interesting to make automatic_boundary() test bigger tmax and may find a more advantageous AbsorbingBoundary lead size when compared to SimpleBoundary. This draft implements that, and also tests different degrees (that supposes that everything still works with a degree != 6).

Merge request reports