Blurb::
Use BFGS method to compute quasi-hessians
Description::
Broyden-Fletcher-Goldfarb-Shanno (BFGS) update will be used to compute
quasi-Hessians.

\f[
B_{k+1} = B_{k} - \frac{B_k s_k s_k^T B_k}{s_k^T B_k s_k} + 
\frac{y_k y_k^T}{y_k^T s_k}
\f]

where \f$B_k\f$ is the \f$k^{th}\f$ approximation to the Hessian, 
\f$s_k = x_{k+1} - x_k\f$ is the step and 
\f$y_k = \nabla f_{k+1} - \nabla f_k\f$ is the corresponding yield 
in the gradients. 

<b> Notes </b>

\li Initial scaling of 
\f$\frac{y_k^T y_k}{y_k^T s_k} I\f$ is used for \f$B_0\f$ prior to the first 
update. 
\li Numerical safeguarding is used
to protect against numerically small denominators within the updates. 
\li This safeguarding skips the update if 
\f$|y_k^T s_k| < 10^{-6} s_k^T B_k s_k\f$ 
\li Additional safeguarding can be 
added using the \c damped option, which utilizes an alternative 
damped BFGS update when the curvature condition \f$y_k^T s_k > 0\f$ 
is nearly violated.



Topics::	
Examples::
Theory::
Faq::
See_Also::	
