Show that the BFGS formula (B.23) is the solution to the constrained optimization problem:

\[ \mathbf{C}_{\mathrm{BFGS}}=\underset{\mathbf{A} \text { subject to } \mathbf{A} \boldsymbol{g}=\mathbf{\delta}, \mathbf{A}=\mathbf{A}^{\top}}{\operatorname{argmin}} \mathscr{D}(\mathbf{0}, \mathbf{C} \mid \mathbf{0}, \boldsymbol{A}) \]

where \(\mathscr{D}\) is the Kullback-Leibler discrepancy defined in (B.22). On the other hand, show that the DFP formula (B.24) is the solution to the constrained optimization problem:
\[ \mathbf{C}_{\mathrm{DFP}}=\underset{\mathbf{A} \text { subject to } \mathbf{A} \boldsymbol{g}=\boldsymbol{\delta}, \mathbf{A}=\mathbf{A}^{\top}}{\operatorname{argmin}} \mathscr{D}(\mathbf{0}, \mathbf{A} \mid \mathbf{0}, \mathbf{C}) . \]