Example 3x2 table would look like below. What do you think about the idea?
* I thought about something similar, but maybe more flexible (and possibly expensive):
  • user could provide arbitrary precision functions, in the same way as patches are modelled in 3D.
  • he could of course "edit" the gridpoints of the patch in a chosen resolution (ie: table)
  • the tuningsoftware would revise a table representation of the patch, where the linear-interpolation gives the best possible match. This can be implemented relatively simply with either
  • simulated annealing or
  • genetic algorithm
  • the user could provide a weightmap (or choose between default weightmaps) so that in given loadsites any error in the approximation weights more than at other loadsites.
  • while the patch approach might seem more problematic at first, it eliminates some naughty and unprecise operations related to editing axis-bins, inserting lines, cols.