Citizendium - a community developing a quality comprehensive compendium of knowledge, online and free. Click here to join and contribute—free |

CZ thanks . AUGUST 2014 donors; special to Darren Duncan. SEPTEMBER 2014 donations open; need minimum total $100. Let's exceed that. Donate hereTreasurer's Financial Report -- Thanks to August content contributors. -- |

# Green's function

### From Citizendium, the Citizens' Compendium

In physics and mathematics, **Green's functions** are auxiliary functions in the solution of linear partial differential equations. Green's function is named for the self-taught English mathematician George Green (1793 – 1841), who investigated electricity and magnetism in a thoroughly mathematical fashion. In 1828 Green published a privately printed booklet, introducing what is now called the Green function. This was ignored until William Thomson (Lord Kelvin) discovered it, recognized its great value and had it published nine years after Green's death. Bernhard Riemann gave it the name "Green function".^{[1]}

## Contents |

## Definition

Let *L*_{x} be a given linear differential operator in *n* variables * x* = (

*x*

_{1},

*x*

_{2}, ...,

*x*

_{n}), then the

*Green function of*

*L*

_{x}is the function

*G*(

*,*

**x***) defined by*

**y**where δ(* x*-

*) is the Dirac delta function. Once*

**y***G*(

*,*

**x***) is known, any differential equation involving*

**y***L*

_{x}is formally solved. Suppose we want to solve,

for a known right hand side ρ(* x*).
The formal solution is

The proof is by verification,

where in the last step the defining property of the Dirac delta function is used.

The integral operator that has the Green function as kernel may be seen as the inverse of a linear operator,

It is illuminating to make the analogy with matrix equations. Let and be *n*×*n* matrices connected by

then the solution of a matrix-vector equation is

Make the correspondence *i* ↔ * x*,

*j*↔

*, and compare the sum over*

**y***j*with the integral over

*, and the correspondence is evident.*

**y**## Example

We consider a case of three variables, *n* = 3 with * x* = (

*x*,

*y*,

*z*).

The Green function of

is

As an important example of this Green function we mention that the formal solution of the Poisson equation of electrostatics, reading

where ε_{0} is the electric constant and ρ is a charge distribution,
is given by

Indeed,

The integral form of the electrostatic field may be seen as a consequence of Coulomb's law. The field at the point * x* due to the charge d

*Q*= ρ(

*)d*

**y***is equal to*

**y**The field is additive in the charges, so integration gives the total field at * x*.

### Proof of Green function of ∇^{2}

Without loss of generality we take * x* as the origin (0, 0, 0) and replace

*by*

**y***= (*

**r***x*,

*y*,

*z*) in the above formulation. The length of

*is indicated by*

**r***r*.

The proof uses Green's theorem:

where *V*_{a} is a sphere with radius *a* and *S*_{a} is the surface of this sphere. The smooth test function φ and its gradient vanish for large *r*,

Further we notice that

because *r* ≠ 0 in that region (see the figure, where the region is indicated in yellow).
This result is most easily proved if we recall that in spherical polar coordinates

First apply Green's theorem to the large sphere of radius *R*

because by assumption φ and its gradient vanish on *S*_{R}.

We consider the integral on the right hand side and we will show that

from which the result to be proved follows directly. The main trick is to write

and to consider first the integral over *U*_{ε} (the yellow domain in the figure) for non-zero, but small, ε. After the integral has been evaluated, the limit for zero ε is taken.

Since *U*_{ε} has two surfaces, Green's theorem cannot be applied directly, and therefore we write (see the figure),

and apply Green's theorem to the two terms. Recalling that we saw already the first term, we get

The integral between square brackets is zero because ∇^{2}(1/*r*) is zero on *U*_{ε}. The last integral can be shown to vanish for small ε. Because φ and its gradient are smooth and finite, and *r* is constant (equal to ε) on the surface, we may write for small ε

where we assumed that

and that the value of the gradient averaged over the surface may be taken out of the integral. The remaining surface integral is equal to 4πε^{2}.

In order to evaluate the final integral we use

so that

Here we assumed that the test function φ(* r*) is constant over the surface (isotropic) for small ε. The limit ε → 0 gives the desired result.

Summarizing,

and

## References

- ↑ M. Kline,
*Mathematical Thought from Ancient to Modern Times*, Oxford University Press, New York (1972) p. 683

- P. Roman,
*Advanced Quantum Theory*, Addison-Wesley, Reading, Mass. (1965) Appendix 4. - I. M. Gel'fand and G. E. Shilov,
*Generalized Functions*, Vol. 1, Academic Press, New York (1964)