Weak 

solutions of the Poisson equation 

Now we are ready to demonstrate the usefulness of Sobolev spaces in the simplest situation,
namely, we prove the existence of weak solutions of the Poisson equation.

Let is a bounded open subset of and let us consider the Dirichlet problem for the Poisson
equation:

∆ ,

0, .

Since we are interested in the zero boundary condition, it is reasonable to consider as the
class of test functions. Let be the classical solution of the above Dirichlet problem. Then
multiplying the Poisson equation by an arbitrary test function and integrating
the obtained equality by parts yields

· 1

The following remarks are appropriate to mention here:

,
(a) The latter equality may also be interpreted for any function of a larger class .
(b) A bilinear form

, : ·

in the left hand side looks like a scalar product.

(c) The identity (1) can be interpreted now as equation on a linear functional

: 2

and the scalar product , that resembles the Riesz representation theorem
(d) If , the linear functional defined above is bounded in -norm for
subspace and moreover, we have
/ /
| | ,

,
that is extends (by Hahn-Banach theorem) to a bounded functional in .

We return to the scalar product defined above. In fact, the bilinear form

, : ·

,
is well-defined in the whole , symmetric and positive there (notice that | | 0 yields
, which implies, by the zero boundary condition, 0). Denote by | | ,
the associated norm.
Hence we obtain a weak formulation of (1): given a function , any classical solution
to the Poisson equation with zero boundary condition is a solution of the following problem

, , , 3

where is defined by (2). The following definition is then natural.

,
Definition. A function which solves (3) is called a weak solution to the Poisson
equation with zero boundary conditions.

In order to threat the existence of a weak solution we show that the new scalar product
generates the same metric (and topological) structure in , .

Poincare’s inequality 

Theorem  (Poincare’s inequality).  If    is  a  bounded  open  subset  of    then  there  exists  a 
constant  0 such that  

| |  

,
for all  . Moreover, the inequality holds for any function in   

Denote by the same letter the function obtained by extending by zero outside of . Let
be large enough such that the support of is contained in the cube :| | .
Notice that 1, hence integrating by parts we find

2 2 | |

Applying Cauchy inequality we obtain

1 1
| |
4 4
and the desired property follows.

Remark 1. The question on the optimal constant in the Poincare inequality has many
relations to other problems in mathematics and mathematical physics. For instance, if one
think of as a membrane then the best constant √ in the Poincare inequality is exactly
the fundamental frequency sometimes is called also the fundamental tone of . If
0, is the one‐dimensional interval then / . Besides the fundamental tone,
one distinguish also a series of higher “tones” see the first picture below . The second
picture shows the oscillating two‐dimensional membrane corresponding to the 7th tone
sin 2 sin 4 in the rectangle 0,1 0,1 .
Remark 2. The above Poincare inequality is a partial case of a more general relation the
so‐called Poincare‐Friedrichs inequality inequality

| | ,  

which can be proved by the same method.

Corollary 1. If    is  a  bounded  open  subset  of    then  there  exists  a  constant 
,
0such that for any    

,  
,
Indeed, for any Poincare’s inequality yields

, | | | | 1 | | | |

Corollary 2 Existence and uniqueness of the weak solution . For any there is a
,
unique function   which solves (3). 

By Corollary 1 we have , and, in the other direction,

| | | | | | | | , .

It follows that | | and , are equivalent norms.

On the other hand, defined by (2) is a bounded linear functional in , . Hence, applying
the Riesz representation theorem to the new scalar product , we immediately conclude that
,
there is a unique function such that
,
, ,

That is is a weak solution to the Dirichlet problem formulated above.

Existence of weak solutions for general elliptic equations

The method described above can be applied also to the Dirichlet problem for more general
elliptic equations. Consider a differential operator in divergence form

.
,

where , , and satisfying the (uniform)

ellipticity condition: there is a constant 0 such that for any

.
,

Starting from a smooth solution and integrating by parts, we see that the natural definition of a
weak solution is the following.
,
Definition. A function which solves (3) is called a weak solution to 0 with
zero boundary conditions if

, . 4
,

It is then convenient to restate the condition for a weak solution as follows:

, , , 5

where · s defined as before by formula (2) and the new bilinear form is given by

, .
,

The existence of a weak solution will follow immediately if we show that , is a scalar
product whose associated norm is equivalent to the standard norm in , . So we need to
investigate under which conditions , meets these criteria.

First we notice that , is symmetric by our assumption on the coefficients . Next, by

using of the ellipticity condition we obtain the following lower estimate:

, ,

where max : . Let us assume that

max :
where is the constant in the Poincare inequality. If 0 then , , and if
0 then

, .

Applying then Corollary 1 we obtain

, ,

where and are some positive constants. The latter inequality is called coercive condition.

Now, we notice that functions | | and | | are continuous in , hence they bounded
there. Denoting by the common upper bound we find

, ,

,
for all , and, consequently, for all .

Summarizing, we conclude that , is a symmetric bilinear form satisfying the bilateral

inequality

, , , , 0 .

Let | | , . Then the new norm is equivalent to , and we can again apply the

Riesz theorem to show that for any bounded linear functional there is a function
,
such that equality (5) above holds. Thus we have obtained

Theorem. In the made assumptions, the weak solution of  0 exists and is uniquely defined.