Numerical Methods Course Notes Version 0.1 (UCSD Math 174, Fall ...
Numerical Methods Course Notes Version 0.1 (UCSD Math 174, Fall ...
Numerical Methods Course Notes Version 0.1 (UCSD Math 174, Fall ...
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
4.3. SECANT METHOD 57<br />
relation between e k+1 and the previous two errors, e k , e k−1 .<br />
Indeed, this is the case. Omitting all the nasty details (see Cheney & Kincaid [7]), we arrive at<br />
the imprecise equation:<br />
e k+1 ≈ − f ′′ (r)<br />
2f ′ (r) e ke k−1 = Ce k e k−1 .<br />
Again, the proof relies on finding some “attractor” region and using continuity.<br />
We now postulate that the error terms for the secant method follow some power law of the<br />
following type:<br />
|e k+1 | ∼ A |e k | α .<br />
Recall that this held true for Newton’s method, with α = 2. We try to find the α for the secant<br />
method. Note that<br />
|e k | = A |e k−1 | α ,<br />
so<br />
Then we have<br />
|e k−1 | = A − 1 α |ek | 1 α .<br />
A |e k | α = |e k+1 | = C |e k | |e k−1 | = CA − 1 α |e k | 1+ 1 α ,<br />
Since this equation is to hold for all e k , we must have<br />
α = 1 + 1 α .<br />
This is solved by α = 1 ( √ )<br />
2 1 + 5 ≈ 1.62. Thus we say that the secant method enjoys superlinear<br />
convergence; This is somewhere between the convergence rates for bisection and Newton’s method.