Hmm...
“we would like to constrain the NCAP share for each of these 4 groups to
certain given shares”.
I
am not at all sure that I fully understand the question or the problem you
have. The
problem would probably become much clearer if you would explain what you have
tried so far, but you are not giving any clue as to why your attempts have
failed.
But
OK, I will try to give some help anyway. I am assuming that you would like to
specify a constraint consisting solely of a set of NCAP variables for a single
period, or similar constraints for several individual periods.
The
general form for such a constraint is the following (for a single period):
∑i Ai ×NCAPi
≤ 0
The
constraint is thus basically very simple: It consists only of certain NCAPi
variables, each of which have a coefficient Ai, and the right hand
side (RHS) is zero. With constraints of
this general form you could define any share constraints (UP/LO) for any groups
of NCAP variables in the total sum of the NCAP variables over all i, by
defining the appropriate coefficients Ai that will define exactly the
constraint desired.
In
order to define these kind of constraints in VEDA, you just need to define the
coefficients Ai and the zero RHS value. However, I don’t see any benefit of using an
example with 100 processes, it is unnecessarily large. Let's assume for simplicity that an example constraint involves only 6 processes. For six processes, the straightforward
specification of the constraint would look like this (this is just one possible
layout out of numerous different layouts available under VEDA):
In
the example above, I have used 2020 as the period for the input data
specification (see the Year column). The process names are specified in the Pset_PN column. The coefficients Ai
for the NCAP variables are defined by the UC_NCAP attribute. The values are thus specified in the UC_NCAP column in the figure above (shown by the <An> placeholders). The
zero RHS value is specified under the UC_RHSRT
column header, and I have also included an interpolation option for the RHS
value, which is specified under the UC_RHSRT~0
column. The option 5 extrapolates the
zero RHS value to all periods after 2020, and that will cause the same
constraint to be generated for all those later periods as well.
As
you can see, the specification is very simple and straightforward, and
therefore it is difficult to imagine why you did not succeed to define such a
constraint. Specifying different Ai coefficients for different
periods is equally simple: Just add more lines for the coefficients in
different periods. And the generalization to 100 processes would be equally
straightforward.
However,
with a set of 100 processes, the straightforward specification will obviously become
a bit cumbersome, and therefore I would suggest to utilize process name filters as
well. In my next post I will show another example utilizing process name
filters for an easier specification with a large number of processes.
