Determining a Reasonable Range of Relative Numerical Tolerance Values for Simulating Deterministic Models of Biochemical Reactions

摘要

Adeterministic model of a biological process in the form of differential equations presents mathematical relationships among variables of precise values, but in simulation of such a model, only approximate values of the variables can be employed. The approximation is governed by either an absolute or a relative numerical tolerance value. An absolute numerical tolerance value is the absolute difference between the actual and expected values. Dividing such an absolute difference by the expected value gives rise to a relative numerical tolerance value. Because a model of a biological system may produce different simulation outcomes with different ranges of numerical tolerance values, it is important to determine a reasonable range of numerical tolerance values. Currently a reasonable range of numerical tolerance values for modeling a biological process such as that involving biochemical reactions has not been explored.