[go: up one dir, main page]

Skip to content

2nd and higher reduction-orders for the heuristic

Problem

Consider simply A <-> B in a steady state, and a steady-state data point for one of these interacting species. Removing both reactions at once give a viable (empty) model, whereas removing only one gives non-viable model. Hence, the OAT-heuristic will not be able to identify the viable reduced model.

(In genetics e.g.: double non-lethal KOs, with single KOs lethal; cf. work of Thomas Moran Hunt.)

Solution

Parameter of a method defining order of reductions to consider before summing them up. Increases heuristic cost per point from n to n^k, where n is number of params and k is the reduction order.