Error Messages/41323
Example error messages
The first constraint specified in the linear optimization problem defined in 'My_opt' is non-linear:
(((3*(X^2)) + 4) <= 8)
The linear optimization defined in 'My_opt' contains the non-linear constraint 'Inventory_requirement'.
The non-linearity was first detected in 'Variable Customer_growth_pot'.
The linear optimization defined in 'Portfolio_QP' contains a non-linear constraint 'Risk_thresh'.
Cause
In the specification of your optimization problem using DefineOptimization, you have specified the type of optimization problem in the «type» parameter. When DefineOptimization analyzed your model, it found that a constraint was not of the specified type. For example, if you declare type: "LP", but a constraint turns out to be a non-linear function of the decision variables, than this message results.
In some cases, Analytica may conclude that a constraint is non-linear or non-quadratic when intermediate computations involve non-quadratic, or potentially non-quadratic, operations. A simple example would be:
(x^3 - 7) - x^3
Although this is a quadratic relationship of the decision variable x, an intermediate involves a non-quadratic (x^3), and hence Analytica will conclude that the relationship is not quadratic.
Remedy
DefineOptimization allows you to specify the «type» explicitly so that you can catch cases where non-linearity or non-quadraticity is accidentally introduced into the model. Since linear and quadratic problems usually solve faster and more reliably, odds are that you'll want to remove the non-linearity or non-quadratic operation.
If you decide to relax your optimization, say to a non-linear optimization, you can either remove the «type» parameter entirely from your call to DefineOptimization, or set it to type: "NLP". If you set it type: "NLP", you can save time by allowing DefineOptimization skip its attempt to determine whether the model is linear or quadratic.
Enable comment auto-refresher