[Quantlib-dev] Some more explanation about LongstaffSchwartz

[Andrea <mar...@go...>]

In ql/methods/montecarlo/longstaffschwartz.hpp at around line 155 there is the following code

             else {
             // if number of itm paths is smaller then the number of
             // calibration functions -> no early exercise
                 coeff_[i] = Array(v_.size(), 0.0);
             }

The array of coefficients is used at around line 113 & 164.
These are coefficients of a linear combination of basis functions

                     Real continuationValue = 0.0;
                     for (Size l=0; l<v_.size(); ++l) {
                         continuationValue += coeff_[i][l] * v_[l](x[k]);
                     }

so that if the array is full of 0.0, the continuation value is 0.0 as well.
Then, the check is

                     if (continuationValue < exercise[j]) {
                         prices[j] = exercise[j];
                     }

which implies exactly the opposite of what the comment says (exercise[j] is > 0.0)

This is not very important because it only happens for paths where the exercise is positive which 
were anyway very few.

I just find the comment misleading.