Type = Real(+)

Default = 1.0

Inital value for the penalty term β, i.e. β(1) as described in Equation 8. The default value should typically be sufficient, but for problems which have a low number of OD entries to be estimated and few screenlines, the penalty value should be set substantially higher. The reason for this is that the penalty multiplier (DELTA for static estimation, DYNDELTA for dynamic estimation) increases the value at every iteration which enforces the boundary constraints as the iteration number increases. However, for certain problems such as those mentioned here convergence may occur in very few iterations before this term properly takes effect on the problem and so a large initial value may be necessary to ensure that the OD entries remain within the desired bounds so as to not result in undesired results due to the solution converging quickly.