In solving the method of least squares, we have the option of using either the normal equation or the representer theorem, as discussed in Section 7.6. However, in solving the semisupervised extension of this method, the representer theorem is the right choice. Explain the rationale for this statement.