this is for holding javascript data
Richard Otis edited Numerical_Approach.md
over 8 years ago
Commit id: 1918a595728ff0af3bd5586e5297bdf5f15f8fae
deletions | additions
diff --git a/Numerical_Approach.md b/Numerical_Approach.md
index 22a1c27..a8f4c67 100644
--- a/Numerical_Approach.md
+++ b/Numerical_Approach.md
...
## Alternative Formulation
The refinement sub-problem can also be solved by constructing an augmented Hessian matrix including all phases and
constraints. constraints \cite{Nocedal2006}.
This is similar to the approach of Lukas \cite{Lukas1982}.
The advantage of this approach is you can solve for the chemical potentials and site fractions of all phases simultaneously.
The disadvantages of this are most apparent in large multi-component systems and systems with miscibility gaps;
the Hessian matrix will
either be
sparse large and
sparse or it will have to be rebuilt every time the stable set of phases changes.
\[
\chi(\lambda) \textrm{min }L =
\left| \begin{array}
\lambda \sum_i f^{i}G^{i}_{m}(T, P, y^i_{s, j}) -
a & -b & -c \\
-d & \lambda - e & -f \\
-g \sum_k \lambda_k c_k(T, P, f^{i}, y^i_{s, j})\]
\[f(x) = \left\{
\begin{array}{lr}
x^2 &
-h : x < 0\\
x^3 &
\lambda - i : x \ge 0
\end{array}
\right|.\] \right.
\]