Kantorovitchโ€™s theorem


Let ๐š0 be a point in โ„n,U an open neighborhood of ๐š0 in โ„n and ๐Ÿ:Uโ†’โ„n a differentiable mapping, with its derivativePlanetmathPlanetmath [๐ƒ๐Ÿโข(๐š0)] invertible. Define

๐ก0=-[๐ƒ๐Ÿโข(๐š0)]-1โข๐Ÿโข(๐š0),๐š1=๐š0+๐ก0,U0={๐ฑ||๐ฑ-๐š1|โ‰ค|๐ก0|}.

If U0โŠ‚U and the derivative [๐ƒ๐Ÿโข(๐ฑ)] satisfies the http://planetmath.org/node/765Lipschitz conditionMathworldPlanetmath

|[๐ƒ๐Ÿโข(๐ฎ1)]-[๐ƒ๐Ÿโข(๐ฎ2)]|โ‰คMโข|๐ฎ1-๐ฎ2|

for all points ๐ฎ1,๐ฎ2โˆˆU0, and if the inequalityMathworldPlanetmath

|๐Ÿโข(๐š๐ŸŽ)|โข|[๐ƒ๐Ÿโข(๐š๐ŸŽ)]-1|2โขMโ‰ค12

is satisfied, the equation ๐Ÿโข(๐ฑ)=๐ŸŽ has a unique solution in U0, and Newtonโ€™s method with initial guess ๐š0 converges to it. If we replace โ‰ค with <, then it can be shown that Newtonโ€™s method http://planetmath.org/node/793superconverges! If you want an even stronger version, one can replace |โ€ฆ| with the norm ||โ€ฆ||.

Logic behind the theorem:

Letโ€™s look at the useful part of the theorem:

|๐Ÿโข(๐š๐ŸŽ)|โข|[๐ƒ๐Ÿโข(๐š๐ŸŽ)]-1|2โขMโ‰ค12.

It is a productPlanetmathPlanetmath of three distinct properties of your function such that the product is less than or equal to a certain number, or bound. If we call the product R, then it says that ๐š0 must be within a ball of radius R. It also says that the solution ๐ฑ is within this same ball. How was this ball defined?

The first term, |๐Ÿโข(๐š๐ŸŽ)|, is a measure of how far the function is from the domain; in the Cartesian plane, it would be how far the function is from the x-axis. Of course, if weโ€™re solving for ๐Ÿโข(๐ฑ)=๐ŸŽ, we want this value to be small, because it means weโ€™re closer to the axis. However a function can be annoyingly close to the axis, and yet just happily curve away from the axis. Thus we need more.

The second term, |[๐ƒ๐Ÿโข(๐š๐ŸŽ)]-1|2 is a little more difficult. This is obviously a measure of how fast the function is changing with respect to the domain (x-axis in the plane). The larger the derivative, the faster itโ€™s approaching wherever itโ€™s going (hopefully the axis). Thus, we take the inversePlanetmathPlanetmathPlanetmath of it, since we want this product to be less than a number. Why itโ€™s squared though, is because it is the denominator where a product of two terms of like units is the numerator. Thus to conserve units with the numerator, it is multiplied by itself. Combined with the first term, this also seems to be enough, but what if the derivative changes sharply, but it changes the wrong way?

The third term is the Lipschitz ratio M. This measures sharp changes in the first derivativeMathworldPlanetmath, so we can be sure that if this is small, that the function wonโ€™t try to curve away from our goal on us too sharply.

By the way, the number 12 is unitless, so all the units on the left side cancel. Checking units is essential in applications, such as physics and engineering, where Newtonโ€™s method is used.

Title Kantorovitchโ€™s theorem
Canonical name KantorovitchsTheorem
Date of creation 2013-03-22 11:58:09
Last modified on 2013-03-22 11:58:09
Owner stevecheng (10074)
Last modified by stevecheng (10074)
Numerical id 25
Author stevecheng (10074)
Entry type Theorem
Classification msc 49K10
Synonym Kantorovitch inequality
Related topic LipschitzCondition
Related topic NewtonsMethod
Related topic Superconvergence