Develop "reality" concept

[grumpyinca]

Recent discussion as to how to best model and cope with various "corner cases" (for example, the solid condition of a compression spring) has developed the notion of "reality" and a separate "reality space" in addition to the original feasible space and relatively new valid space. The idea is that reality is a proper subset of validity and feasibility is a proper subset of reality.

The motivation for introducing this extra level of complexity is that situations like the solid condition of a compression spring are numerically valid, but not physically real. Experience has shown that adding a conditional to the math model that limits deflection at the solid condition incurs problems with the numerical search.

There is at least one possible approach that might eliminate the need to introduce the complexity of a reality concept. In the general case, it may be possible to introduce an additional dependent variable that measures the deviation from reality and then apply the existing validity and feasibility limits.