This option can be used for the following purposes:

* adapting the OF matrix to the new setting of the requested forgetting index (the Lapses parameter),

* smoothing the OF matrix in case it has become excessively irregular.

WARNING! **Unless you understand how Approximate works, you should not use it**. Misguided use of **Approximate** may slow down the learning process. What **Approximate** does is to find the closest compromise between the RF matrix (displayed by the **Process **: **Retention **option) and the theoretically predicted value of the OF matrix. In other words, it uses the retention factors as input data and tries to find parameters of the ideal function of optimal factors that provide the closes fit to the RF matrix.

As soon as the parameters of the function of optimal intervals are found, the function is translated into the OF matrix. For the repetition number equal 1, the function of optimal intervals has a close-to-linear nature and is approximated by a procedure akin to linear regression with the number of repetition instances considered in the computation. For the repetition number greater than 1, the approximation procedure is iterative in nature. In this case, the four parameters of the function of optimal intervals are initially set to average, expected values, and the algorithm proceeds toward minimizing the objective function called the deviation (sum of square differences between OFs and RFs). During iterations, all the parameters are displayed on-screen together with the measure of steps used in the hill-climbing algorithm (akin to the Rosenbrock method). The counter variable indicates the number of steps and iterations of the algorithm as it progresses. The progress parameters indicate the decrease of the objective function Deviation in particular steps and iterations, as well as the trailing average of the iteration progress. Iterations proceed until the trailing progress drops below 0.01 or Esc is pressed.