Theorem
(Broyden's Method for n-dimensional Systems). To
solve the non-linear system
, given
one initial approximation
, and
generating a sequence
which
converges to the solution
, i.e.
. Compute
the initial Jacobian
matrix
.
Use it to compute the first approximation using Newton's
method
.
For
. Suppose
that
has
been obtained, use the following steps to
obtain
.
Step
1. Evaluate the
function
.
Step 2. Update
the approximate Jacobian using
, and ![]()
.
Step
3. Compute
using
the Sherman-Morrison formula
Step 4. Compute the next
approximation
.
Remark.
As the iteration proceeds,
, and
then
![]()
or
![]()
where, for
,
,
,
,
then
![]()
and
.
(c) John H. Mathews 2005