Student: Victor Pugliese

R#: 11492336
Course: MATH 5345 Numerical Analysis
Project 03

The problem considered is:

Δ𝑢 = 𝑓, 𝑖𝑛 Ω = (−1,1)2
{
𝑢 = sin 𝑥 cos 𝑦 , 𝑜𝑛 𝜕Ω

If 𝑓 = −2 sin 𝑥 cos 𝑦 ; thus, 𝑢 = sin 𝑥 cos 𝑦 is the analytical solution of equation (1). It is
important to note that the model has inhomogeneous boundary condition. Thus the problem we
have:

Find 𝑢(𝑥, 𝑦) such that,

−Δ𝑢 = 2 sin 𝑥 cos 𝑦 , 𝑖𝑛 Ω = (−1,1)2 (1)
{
𝑢 = sin 𝑥 cos 𝑦 , 𝑜𝑛 𝜕Ω

I. Weak Formulation.
This problem has different trial and test space due to the boundary condition 𝑢 =
𝑔𝐷 . The trial space is given by
𝑉𝑔𝐷 = {𝑣: ‖𝑣‖𝐿2 + ‖∇𝑣‖𝐿2 < ∞, 𝑣|𝜕Ω = 𝑔𝐷 }
The test space is given by
𝑉0 = {𝑣: ‖𝑣‖𝐿2 + ‖∇𝑣‖𝐿2 < ∞, 𝑣|𝜕Ω = 0}

Then,
We multiply −Δ𝑢 = 𝑓 by a test function 𝑣, which is in 𝑉0.
∫ Δ𝑢𝑣𝑑𝑥 = ∫ 𝑓𝑣𝑑𝑥
Ω Ω
Using Green’s formula:
∫ ∇𝑢 ∙ ∇𝑣𝑑𝑥 − ∫ 𝑛 ∙ ∇𝑢𝑣𝑑𝑠 = ∫ 𝑓𝑣𝑑𝑥
Ω 𝜕Ω Ω
𝑣 is zero on the boundary, thus we have the following variational formulation of
(1): find 𝑢 ∈ 𝑉𝑔𝐷 such that
∫ ∇𝑢 ∙ ∇𝑣𝑑𝑥 = ∫ 𝑓𝑣𝑑𝑥 , ∀𝑣 ∈ 𝑉0
Ω Ω

(∇𝑢, ∇𝑣) = (𝑓, 𝑣), ∀𝑣 ∈ 𝑉0 (2)

II. Weak Galerking Finite Element Approximation

 𝑇ℎ is a triangulation of Ω.
 𝜀ℎ is the set of all edges in 𝑇ℎ .
  𝜀ℎ0 = 𝜀ℎ /𝜕Ω is the set of all interior edges.
 ℎ𝑇 is the diameter of element 𝑇 ∈ 𝑇ℎ .
 ℎ = max ℎ𝑇 is the mesh size of 𝑇ℎ .
T∈Th
 ℎ𝑒 is the diameter of each edge 𝑒 ∈ 𝜀ℎ .
 𝑘 ≥ 1; 𝑗 ≥ 0; 𝑙 ≥ 0 are integers.
 𝑚 = max{𝑗, 𝑙}.
 𝑃𝑘 (𝑇) is the space of polynomials of total degree 𝑘 or less on the element
𝑇 ∈ 𝑇ℎ .
 𝑃𝑗 (𝑒) is the space of polynomials of total degree 𝑗 or less on the edge 𝑒 ∈
𝜀ℎ .
 𝑃𝑚 (𝑒) is the space of polynomials of total degree 𝑚 or less on the edge
𝑒 ∈ 𝜀ℎ .
 𝑄𝑘0 is the 𝐿2 projection onto 𝑃𝑘 (𝑇) on each element 𝑇 ∈ 𝑇ℎ .
 𝑄𝑗𝑏 is the 𝐿2 projection onto 𝑃𝑗 (𝑇) on each edge 𝑒 ∈ 𝜀ℎ .
 𝑄𝑙 is the 𝐿2 projection onto [𝑃𝑙 (𝑇)]2 on each element 𝑇 ∈ 𝑇ℎ .
 𝑄ℎ is the 𝐿2 projection onto the finite element space 𝑉ℎ such that 𝑄ℎ | 𝑇 =
{𝑄𝑘0 , 𝑄𝑗𝑏 }.
 𝑄𝑚 is the 𝐿2 projection onto 𝑃𝑚 (𝑒) on each edge 𝑒 ∈ 𝜀ℎ .

On each element 𝑇 ∈ 𝑇ℎ , the following local weak finite element space is defined:
𝑉(𝑘, 𝑗, 𝑇) ≔ {𝑣 = {𝑣0 , 𝑣𝑏 }: 𝑣0 ∈ 𝑃𝑘 (𝑇), 𝑣𝑏 |𝑒 ∈ 𝑃𝑗 (𝑒), 𝑒 ⊂ 𝜕𝑇}

𝑉ℎ is a global weak finite element space, constructed by patching local spaces

𝑉(𝑘, 𝑗, 𝑇) through a common value of 𝑣𝑏 on all interior edges:
𝑉ℎ = {𝑣 = {𝑣0 , 𝑣𝑏 }: 𝑣| 𝑇 ∈ 𝑉(𝑘, 𝑗, 𝑇), 𝑣𝑏 |𝑒∩𝜕𝑇1 = 𝑣𝑏 |𝑒∩𝜕𝑇2 ,
𝑒 = 𝜕𝑇1 ∩ 𝜕𝑇2 }

𝑉ℎ0 is the subspace of 𝑉ℎ consisting of all finite element functions with vanishing
boundary value.
𝑉ℎ0 = {𝑣 ∈ 𝑉ℎ : 𝑣𝑏 |𝜕Ω = 0}

𝑉ℎ,𝑔𝐷 is the subspace of 𝑉ℎ consisting of all finite element functions with 𝑔

boundary value
𝑉ℎ,𝑔𝐷 = {𝑣 ∈ 𝑉ℎ : 𝑣|𝜕Ω = 𝑔𝐷 }

∇𝑤 𝑣 is a discrete weak gradient of 𝑣 ∈ 𝑉ℎ , defined as a vector-valued polynomial,

(∇𝑤 𝑣, 𝜙⃗) 𝑇 = −(𝑣0 , ∇ ∙ 𝜙⃗) 𝑇 + 〈𝑣𝑏 , 𝜙⃗ ∙ 𝑛⃗〉𝜕𝑇 , ∀𝜙⃗ ∈ [𝑃𝑙 (𝑇)]2

(∇𝑤 𝑣, 𝜙⃗) 𝑇 = (∇𝑣0 , 𝜙⃗) 𝑇 + 〈𝑣𝑏 − 𝑣0 , 𝜙⃗ ∙ 𝑛⃗〉𝜕𝑇 , ∀𝜙⃗ ∈ [𝑃𝑙 (𝑇)]2

Now, 𝑎(𝑢, 𝑣) ≔ (∇𝑢, ∇𝑣), 𝑢, 𝑣 ∈ 𝐻1 (Ω) is the bilinear form defined by:

2
 𝑎ℎ (𝑢, 𝑣) = ∑ (∇𝑤 𝑢, ∇𝑤 𝑣) 𝑇 , 𝑢, 𝑣 ∈ 𝑉ℎ
𝑇∈𝑇ℎ

𝑠(∙,∙) (stabilizer) is a semi-positive definite bilinear form on 𝑉ℎ × 𝑉ℎ so that 𝑠(𝑣, 𝑣)

provides a metric for the continuity of the finite element function 𝑣 ∈ 𝑉ℎ .
𝑠(𝑢, 𝑣) = ∑ 〈𝑗𝜕 (𝑢), 𝑗𝜕 (𝑣)〉𝜕𝑇
𝑇∈𝑇ℎ

𝑗𝜕 is a locally defined linear operator from 𝑉(𝑘, 𝑗, 𝑇) to a polynomial subspace

𝑃𝑚 (𝜕𝑇) of 𝐿2 (𝜕𝑇). 〈∙,∙〉𝜕𝑇 is the 𝐿2 standard inner product in 𝑃𝑚 (𝜕𝑇)

Replacing 𝑉0 with 𝑉ℎ,𝑔𝐷 in equation (2), we obtain the finite element method: find
𝑢ℎ ∈ 𝑉ℎ,𝑔𝐷 such that
∫ ∇𝑢ℎ ∙ ∇𝑣𝑑𝑥 = ∫ 𝑓𝑣𝑑𝑥 , ∀𝑣 ∈ 𝑉ℎ,0
Ω Ω

𝑎ℎ (𝑢ℎ , 𝑣) + 𝑠(𝑢ℎ , 𝑣) = (𝑓, 𝑣0 ), ∀𝑣 ∈ 𝑉ℎ0

(3)
𝑎𝑠 (𝑢ℎ , 𝑣) = (𝑓, 𝑣0 ), ∀𝑣 ∈ 𝑉ℎ0

III. Existence and Uniqueness

In the WG methods, the polynomial degree and the stabilizer must be chosen so
that the bilinear form 𝑎𝑠 (∙,∙) is coercive with respect to the semi-norm ‖∙‖1,ℎ in
𝑉ℎ . More precisely, there exist constants 𝐶1 , 𝐶2 > 0 such that for any 𝑣 ∈ 𝑉ℎ , the
following hold true:
2 2
𝐶1 ‖𝑣‖1,ℎ ≤ 𝑎𝑠 (𝑣, 𝑣) ≤ 𝐶2 ‖𝑣‖1,ℎ (4)

Theorem 1. Assume that the coercivity (4) holds true. Then, the weak Galerkin
finite element discretization scheme (3) has one and only one solution.
1
−
Theorem 2. Assume that 𝑙 ≥ 𝑘 − 1, 𝑗𝜕 (𝑣) = ℎ𝑇 2 𝑄𝑚 (𝑣𝑏 − 𝑣0 |𝜕𝑇 ) where
𝑄𝑚 : 𝐿2 (𝑒) → 𝑃𝑚 (𝑒) is the 𝐿2 projection operator and 𝑚 = max{𝑗, 𝑙}. Then the
estimate (4) holds true.
1
−
Theorem 3. Assume that 𝑙 ≥ 𝑘 − 1, 𝑗𝜕 (𝑣) = ℎ𝑇 2 (𝑣𝑏 − 𝑣0 |𝜕𝑇 ). Then the estimate
(4) holds true.

IV. Derivation of a Linear System of Equation

To derive an equation for 𝑢ℎ , we write it in the form

3
 𝑢ℎ = 𝑢ℎ,0 + 𝑢ℎ,𝑔𝐷 , 𝑢ℎ,𝑜 ∈ 𝑉ℎ,0 & 𝑢ℎ,𝑔𝐷 ∈ 𝑉ℎ,𝑔𝐷
Because 𝑢ℎ,𝑔𝐷 is known it only remains to determine 𝑢ℎ,0 .

For a given element 𝑇 ∈ 𝑇ℎ , let 𝜙0,𝑖 , 𝑖 = 1, … , 𝑁0 , be a set of basis functions for

𝑃1 (𝑇)m and 𝜙𝑏,𝑖 , 𝑖 = 1, … , 𝑁𝑏 , be a set of basis functions for ∑𝑒∈𝜕𝑇 𝑃1 (𝑒). Then
every 𝑣ℎ = {𝑣0 , 𝑣𝑏 } ∈ 𝑉ℎ has the following representation in 𝑇.
𝑁0 𝑁𝑏

𝑣ℎ | 𝑇 = {∑ 𝑣0,𝑖 𝜙0,𝑖 , ∑ 𝑣𝑏,𝑖 𝜙𝑏,𝑖 }

𝑖=1 𝑖=1

On each 𝑇, the local stiffness matrix 𝑀𝑇 can be written as a block matrix

𝑀0,0 𝑀0,𝑏
𝑀𝑇 = [ ]
𝑀𝑏,0 𝑀𝑏,𝑏

Where 𝑀0,0 is a 𝑁0 × 𝑁0 matrix, 𝑀0,𝑏 is a 𝑁0 × 𝑁𝑏 matrix, 𝑀𝑏,0 is a 𝑁𝑏 × 𝑁0

matrix, 𝑀𝑏,𝑏 is a 𝑁𝑏 × 𝑁𝑏 matrix. These matrices are defined, respectively, by

𝑀0,0 = [𝑎𝑠 (𝜙0,𝑖 , 𝜙0,𝑗 ) 𝑇 ]

𝑖,𝑗

𝑀0,𝑏 = [𝑎𝑠 (𝜙0,𝑖 , 𝜙𝑏,𝑗 ) 𝑇 ]

𝑖,𝑗

𝑀𝑏,0 = [𝑎𝑠 (𝜙𝑏,𝑖 , 𝜙0,𝑗 ) 𝑇 ]

𝑖,𝑗

𝑀𝑏,𝑏 = [𝑎𝑠 (𝜙𝑏,𝑖 , 𝜙𝑏,𝑗 ) 𝑇 ]

𝑖,𝑗

Where the bilinear form is defined as in (3).

|𝑒2 | |𝑒3 | |𝑒3 | |𝑒2 |
( + )
3 3 6 6
|𝑒3 | |𝑒1 | |𝑒3 | |𝑒1 |
𝑀0,0 = ℎ−1𝑇 ( + )
6 3 3 6
|𝑒2 | |𝑒1 | |𝑒1 | |𝑒2 |
( + )
[ 6 6 3 3 ]

|𝑒2 | |𝑒2 | |𝑒3 | |𝑒3 |

0 0
6 3 3 6
|𝑒1 | |𝑒1 | |𝑒3 | |𝑒3|
𝑀0,𝑏 = −ℎ−1
𝑇 0 0
3 6 6 3
|𝑒1 | |𝑒1 | |𝑒2 | |𝑒2 |
( 6 0 0 )
3 3 6
4
 𝑇
𝑀𝑏,0 = 𝑀0,𝐵

𝑀𝑏,𝑏 = (∇𝑤 𝜙𝑏,𝑖 ∙ ∇𝑤 𝜙𝑏,𝑗 )|𝑇| + ℎ−1

𝑇 〈𝜙𝑏,𝑖 , 𝜙𝑏,𝑗 〉𝜕𝑇

(∇𝑤 𝜙𝑏,𝑖 ∙ ∇𝑤 𝜙𝑏,𝑗 )|𝑇| =

2
|𝑒1 |2 (𝑛1,𝑥 2 2
|𝑒1 |2 (𝑛1,𝑥 2 |𝑒1 ||𝑒2 |(𝑛1,𝑥 𝑛2,𝑥 + 𝑛1,𝑦 𝑛2,𝑦 ) |𝑒1 ||𝑒2 |(𝑛1,𝑥 𝑛2,𝑥 + 𝑛1,𝑦 𝑛2,𝑦 ) |𝑒1 ||𝑒3 |(𝑛1,𝑥 𝑛3,𝑥 + 𝑛1,𝑦 𝑛3,𝑦 ) |𝑒1 ||𝑒3 |(𝑛1,𝑥 𝑛3,𝑥 + 𝑛1,𝑦 𝑛3,𝑦 )
+ 𝑛1,𝑦 ) + 𝑛1,𝑦 )
2
|𝑒1 |2 (𝑛1,𝑥 2 2
|𝑒1 |2 (𝑛1,𝑥 2 |𝑒1 ||𝑒2 |(𝑛1,𝑥 𝑛2,𝑥 + 𝑛1,𝑦 𝑛2,𝑦 ) |𝑒1 ||𝑒2 |(𝑛1,𝑥 𝑛2,𝑥 + 𝑛1,𝑦 𝑛2,𝑦 ) |𝑒1 ||𝑒3 |(𝑛1,𝑥 𝑛3,𝑥 + 𝑛1,𝑦 𝑛3,𝑦 ) |𝑒1 ||𝑒3 |(𝑛1,𝑥 𝑛3,𝑥 + 𝑛1,𝑦 𝑛3,𝑦 )
+ 𝑛1,𝑦 ) + 𝑛1,𝑦 )
|𝑒1 ||𝑒2 |(𝑛1,𝑥 𝑛2,𝑥 + 𝑛1,𝑦 𝑛2,𝑦 ) |𝑒1 ||𝑒2 |(𝑛1,𝑥 𝑛2,𝑥 + 𝑛1,𝑦 𝑛2,𝑦 ) 2
|𝑒2 |2 (𝑛2,𝑥 2 2
|𝑒2 |2 (𝑛2,𝑥 2 |𝑒2 ||𝑒3 |(𝑛2,𝑥 𝑛3,𝑥 + 𝑛2,𝑦 𝑛3,𝑦 ) |𝑒2 ||𝑒3 |(𝑛2,𝑥 𝑛3,𝑥 + 𝑛2,𝑦 𝑛3,𝑦 )
1 + 𝑛2,𝑦 ) + 𝑛2,𝑦 )
= 2 2 2 2
4|𝑇| |𝑒1 ||𝑒2 |(𝑛1,𝑥 𝑛2,𝑥 + 𝑛1,𝑦 𝑛2,𝑦 ) |𝑒1 ||𝑒2 |(𝑛1,𝑥 𝑛2,𝑥 + 𝑛1,𝑦 𝑛2,𝑦 ) |𝑒2 |2 (𝑛2,𝑥 + 𝑛2,𝑦 ) |𝑒2 |2 (𝑛2,𝑥 + 𝑛2,𝑦 ) |𝑒2 ||𝑒3 |(𝑛2,𝑥 𝑛3,𝑥 + 𝑛2,𝑦 𝑛3,𝑦 ) |𝑒2 ||𝑒3 |(𝑛2,𝑥 𝑛3,𝑥 + 𝑛2,𝑦 𝑛3,𝑦 )
|𝑒1 ||𝑒3 |(𝑛1,𝑥 𝑛3,𝑥 + 𝑛1,𝑦 𝑛3,𝑦 ) |𝑒1 ||𝑒3 |(𝑛1,𝑥 𝑛3,𝑥 + 𝑛1,𝑦 𝑛3,𝑦 ) |𝑒2 ||𝑒3 |(𝑛2,𝑥 𝑛3,𝑥 + 𝑛2,𝑦 𝑛3,𝑦 ) |𝑒2 ||𝑒3 |(𝑛2,𝑥 𝑛3,𝑥 + 𝑛2,𝑦 𝑛3,𝑦 ) 2
|𝑒3 |2 (𝑛3,𝑥 2 2
|𝑒3 |2 (𝑛3,𝑥 2
+ 𝑛3,𝑦 ) + 𝑛3,𝑦 )
2
|𝑒3 |2 (𝑛3,𝑥 2 2
|𝑒3 |2 (𝑛3,𝑥 2
(|𝑒1 ||𝑒3 |(𝑛1,𝑥 𝑛3,𝑥 + 𝑛1,𝑦 𝑛3,𝑦 ) |𝑒1 ||𝑒3 |(𝑛1,𝑥 𝑛3,𝑥 + 𝑛1,𝑦 𝑛3,𝑦 ) |𝑒2 ||𝑒3 |(𝑛2,𝑥 𝑛3,𝑥 + 𝑛2,𝑦 𝑛3,𝑦 ) |𝑒2 ||𝑒3 |(𝑛2,𝑥 𝑛3,𝑥 + 𝑛2,𝑦 𝑛3,𝑦 ) + 𝑛3,𝑦 ) + 𝑛3,𝑦 ) )

|𝑒1 | |𝑒1 |
0 0 0 0
3 6
|𝑒1 | |𝑒1 |
0 0 0 0
6 3
|𝑒2 | |𝑒2 |
0 0 0 0
ℎ−1 −1 3 6
𝑇 〈𝜙𝑏,𝑖 , 𝜙𝑏,𝑗 〉𝜕𝑇 = ℎ 𝑇 |𝑒2 | |𝑒2 |
0 0 0 0
6 3
|𝑒2 | |𝑒3 |
0 0 0 0
3 6
|𝑒3 | |𝑒2 |
( 0 0 0 0
6 3 )
 The local element load vector 𝑏 𝐾 , formed by the nodes 𝑖, 𝑗, 𝑘, is calculated as
follow:
𝑓𝑖
𝑏𝑖𝐾 = ∫ 𝑓𝜙𝑖 𝑑𝑥 = |𝐾|
K 3
𝑓1
𝑓2
𝑓3
1 0
𝑏 𝐾 = |𝐾| 0
3
0
0
0
[0]
Finally, 𝑏 𝐾 is added to the appropriate places in 𝑏 using the information in the
connectivity matrix T.

Then, we have a 𝑛𝑑𝑜𝑓 × 𝑛𝑑𝑜𝑓 linear system for the unknowns 𝜉𝑖 .

𝐴𝜉 = 𝑏
𝐴00 𝐴0𝑔 𝜉𝑖 𝑏0
[ ][ ] = [ ]
0 𝐼 𝜉𝑔 𝑏𝑔

Where 𝐴00 is the upper left 𝑛𝑖 × 𝑛𝑖 block of A, 𝐴0𝑔 in the 𝑛𝑖 × 𝑛𝑔 upper right
block of A, 𝐼 is the 𝑛𝑔 × 𝑛𝑔 identity matrix, 𝑏0 is the first 𝑛𝑖 × 1 block of 𝑏, 𝑏𝑔 is
the 𝑛𝑔 × 1 vector with nodal values of 𝑢ℎ,𝑔𝐷 , 𝜉𝑖 is the 𝑛𝑖 × 1 vector with nodal
values of 𝑢ℎ,0 , and 𝜉𝑔 is the 𝑛𝑔 × 1 vector with nodal values of 𝑢ℎ,𝑔𝐷 . Rearranging
the first 𝑛𝑖 equations we obtain the 𝑛𝑖 × 𝑛𝑖 linear system
𝐴00 𝜉0 = 𝑏0 − 𝐴0𝑔 𝜉𝑔

V. A Priori Error Estimation

For the stabilizer with projected element-boundary-discrepancy we have:
Theorem 4. Let 𝑘, 𝑗, and 𝑙 ≥ 𝑘 − 1 be the non-negative integers that define the
−1/2
weal finite element space 𝑉ℎ . Assume that 𝑗𝜕 (𝑣) = ℎ𝑇 𝑄𝑚 (𝑣𝑏 − 𝑣0 |𝜕𝑇 ) where
𝑄𝑚 : 𝐿2 (𝑒) → 𝑃𝑚 (𝑒) is the 𝐿2 projection operator and 𝑚 = max{𝑗, 𝑙}. Then the
estimate (4) holds true. Thus, the numerical scheme (3) has one and only one
solution 𝑢ℎ ∈ 𝑉ℎ0 . Furthermore, the following error estimates hold true:
1. For 𝑗 < 𝑙, set 𝜏 = min{𝑘, 𝑗} and assume 𝜏 ≥ 1. Then,
‖𝑄ℎ 𝑢 − 𝑢ℎ ‖1,ℎ < 𝐶ℎ𝜏 ‖𝑢‖𝜏+1
2. For 𝑗 ≥ 𝑙. Then,
‖𝑄ℎ 𝑢 − 𝑢ℎ ‖1,ℎ < 𝐶ℎ𝑘 ‖𝑢‖𝑘+1
 For the stabilizer with element-boundary-discrepancy we have:
Theorem 5. Let 𝑘, 𝑗, and 𝑙 ≥ 𝑘 − 1 be the non-negative integers that define the
−1/2
weal finite element space 𝑉ℎ . Assume that 𝑗𝜕 (𝑣) = ℎ𝑇 (𝑣𝑏 − 𝑣0 |𝜕𝑇 ). Then the
estimate (4) holds true. Thus, the numerical scheme (3) has one and only one
solution 𝑢ℎ ∈ 𝑉ℎ0 . Furthermore, the following error estimates hold true:

‖𝑄ℎ 𝑢 − 𝑢ℎ ‖1,ℎ < 𝐶ℎ 𝑠 ‖𝑢‖𝑠+1

Where 𝑠 = min{𝑘, 𝑗}

VI. A Posteriori Error Estimation

The WG finite element solution 𝑢ℎ satisfies the estimated

‖𝑢 − 𝑢ℎ ‖2 𝐻 1 (Ω) ≤ 𝐶 ∑ 𝜂𝐾2
𝐾
Where the residual 𝜂𝐾 is defined by

1 1/2
𝜂𝐾 = ℎ𝐾 ‖𝑓 + Δ𝑢ℎ ‖𝐿2 (K) + ℎ𝐾 ‖[𝑛 ∙ ∇𝑢ℎ ]‖𝐿2 (𝜕𝐾/𝜕Ω)
2
Here, [𝑛 ∙ ∇𝑢ℎ ] denotes the jump in the normal derivative of 𝑢ℎ on the interior
edges of the element K. Also, since 𝑢ℎ is linear on K, Δ𝑢ℎ = 0.

Using a L2-norm, the discrete error is estimated as follow:

‖𝑢 − 𝑢ℎ ‖2 𝐿2(Ω) = ∑ ℎ𝐾2 (𝑒𝑖2 + 𝑒𝑗2 + 𝑒𝑘2 )
𝐾
Where 𝑖, 𝑗, 𝑘 are the vertices of triangle K.

7
Now, the results of the implementation of FEM in Matlab is presented:

The bounded two-dimensional domain [-1,1]2 is shown in Figure 1. In this case, the mesh size
ℎ𝐾 = 1. This mesh was generated with the function initmesh.

The triangle mesh information is storage in 3

matrices. P, called the point matrix, is of size 2 ×
𝑛𝑝 and each column contains the coordinates 𝑥1
and 𝑥2 of each node. T, called the connectivity
matrix, is of size 3 × 𝑛𝑡 and each column
contains the numbers of the three nodes in each
triangle. E, called the edge matrix, contain the Figure 2 Triangulation
nodes at the boundary of the domain.

In order to apply the Weak Galerking method, a

new matrix containing the following information
is constructed. First, the total number of degree
Figure 1 Triangulation.
of freedom was calculated,
𝑑𝑜𝑓 = 3 ∗ 𝑛𝑒𝑙𝑒𝑚𝑒𝑛𝑡𝑠 + 2 ∗ 𝑛𝑒𝑑𝑔𝑒𝑠
Row 1: Node
Row 2: X coordinate
Row 3: Y coordinate
Row 4: Element in which the node belongs.
Row 5: Edge in which the node belongs

The total number of columns correspond to the total number of degree of freedom.

In order to evaluate the performance of the Finite Element Method, we will compare the
numerical solution with the analytical solution for a simple solution:
𝑢=𝑥

Mesh Size L2 Error

1 2.45e-15
0.5 1.94e-15
0.25 3.68e-15
0.125 1.23e-14
0.0625 2.95e-14
We could state that the numerical Weak Galerking method was implemented correctly and it is
a good approximation of the Poisson problem.

Now, using problem (1), the error estimation based on 𝐿2 and 𝐻1 norms were used to assess the
FEM. FEM was implemented using different mesh size (dx = [1, 0.5, 0.25, 0.125, 0.0625]) in order
to estimate the convergence order.

8
The theoretical order of convergence of the error, estimated based in L2-norm, is 2. If we use the
H1-norm, the convergence order is 1. presents the plot of log10 of error vs log10 of mesh size.
The slope of the straight line represent the convergence order.

a) b)
Figure 3 Error Estimation. a) L2 error. b) H1 error

The matlab files present the details about FEM implementation.

References
Larson, Mats G., and Fredrik Bengzon. The Finite Element Method: Theory, Implementation, and
Applications. Berlin: Springer-Verlag, 2013.
MathWorks. n.d. https://www.mathworks.com.