Repeating the 3B method

An improvement may be obtained for the variable $x_i$ with $i>1$ while no improvement was obtained for the variable $x_1,\ldots,x_{i-1}$. But during the calculation for these variables the range for the variable $x_i$ was the initial one and not the improved value. Hence it may be interesting to repeat the procedure as soon as one the variable is improved.

The 3B procedure will be repeated if:

• the variable `ALIAS/Full3B` is set to 1 and the change in at least one variable is larger than
  `ALIAS/Full3B_Change`
• the variable `ALIAS/Full3B` is set to 1 or 2 (default value: 0) and if there are two changes on the variable (a change is counted when a variable is changed either on the left or right side) or the change in at least one variable is larger than `ALIAS/Full3B_Change`

The 3B method usually exhibit an asymptotically slow convergence speed and it is hence bad policy of repeating the process for very small improvements in the range.