Mon Oct 5 13:14:46 CEST 2009
Starting from . What I have now is a specific propagation engine
(the idea is to use real numbers / floats, so sets of integers are
currently not yet supported). The normalizing should work too. So
what I need is a specification and propagate the problem description
to a solution.
Let's start with something trivial first:
(= (+ (* 3 x) (* 2 y)) 10) ;; define intermediate nodes
(< (+ x y) 3) ;; safety constraints
This should lead to the following behaviour: generate code for the
This produces directed equations + relevant check of constraints.
So, I have a routine now that produces a list of equalities and
inequalities. In the first iteration, the equalities need to be
arranged into a matrix. This means the unknowns need to be
discovered. If the equation is in normal form, we can simply gather
them from the '*' forms.
Next was sorting terms according to order.