0
$\begingroup$

For class we were given three regions defined as either $\Omega_{S}$, $\Omega_{C}$, or $\Omega_{P}$ all in ${R^2}$.

The regions are defined as:

$Ω_{S}$ = {x ∈ $R^2$ : |x| ≤ 1} a circular region

$Ω_C$ = {x ∈ $R^2$ : $x_2$ $\geq$ 0, −$x_2$ $\leq$ $x_1$ $\leq$ $x_2$ $\leq 1$} a triangular region

$Ω_P$ = {x ∈ $R^2$ : $x_2$ ≥ 0, -$\sqrt{x_2} \leq x_1 \leq \sqrt{x_2} \leq 1$} a bottom half of an ellipse

We then were prompted to calculate the outward pointing unit normal vector n(x) for $\partial \Omega_{S}$, $\partial \Omega_{C}$, and $\partial \Omega_{P}$.

I'm having a hard time determining how to calculate such a unit normal vector even when the answers were given. Some help in calculating n(x) would be helpful.

The answers are, respectively:

n(x) = x,

n(x) = $ \frac{1}{\sqrt{2}}(1,-1) $ for $x_{1}>0$ and n(x) = $\frac{1}{\sqrt{2}}$ (-1,-1) for $x_{1} < 0$

and n(x) = $( \frac{2x_{1}}{\sqrt{1+4x_{1}^{2}}}, \frac{-1}{\sqrt{1+4x_{1}^{2}}})$

  • 0
    Is $\Omega_S$ supposed to be a circle?2017-02-18
  • 0
    Ah, yes. Let me fix that. Thank you.2017-02-18

1 Answers 1

1

For $\Omega_S$, it might be easiest to perform by inspection. For a vector pointing to a point on the circle, draw the normal vector on the circle. You should be able to see that the vectors have the same length and direction, i.e., $n(x) = x$.

For $\Omega_C$, the boundary of the triangle is made up of three lines: $x_2=x_1$, $x_2=-x_1$, and $x_2=1$. We can easily calculate the normal vector by thinking about the perpendicular lines for these. For instance, $x_2=x_1$ has a slope of $1$ so the perpendicular line will have a slope of $-1$ (for $y=mx$ the perpendicular line will have a slope of $\frac {-1}m$). We wish to pick a vector parallel to $x_2 = (-1)\cdot x_1$ and in the correct direction so that if points outward from the boundary, e.g. $(1,-1)$. To make it the unit normal, we need the vector to have length one. Normalizing $(1,-1)$ gives $\frac 1 {\sqrt 2} (1,-1)$. You can do the same thing for $x_2 =-x_1$, and the normal vector for $x_2 = 1$ is easy to find.

For $\Omega_p$, you can notice that the boundary lies on the parabola $x_2 = x_1^2$. Thus $r(x_1) = (x_1, x_1^2)\in\mathbb R^2$ parameterizes the boundary for $x_1 \in(-1,1)$. The taking the derivative of $r$ give the tangent vector. The normalized tangent vector at $(x_1, x_2)$ is $$T(x_1) =\frac{r'(x_1)}{\|r'(x_1)\|} = \frac {(1, 2x_1)} {\sqrt{1+4x_1^2}} = \left(\frac 1 {\sqrt{1+4x_1^2}}, \frac {2x_1}{\sqrt{1+4x_1^2}}\right).$$ Now for any vector in $(a,b)\in\mathbb R^2$, we have that $(b,-a)$ is perpendicular (this is most easily seen by taking the dot product $(a,b)\cdot(b,-a)=ab - ab =0$). So $$\left(\frac {2x_1}{\sqrt{1+4x_1^2}}, \frac {-1} {\sqrt{1+4x_1^2}}\right)$$ is perpendicular to the tangent vector $T(x_1)$ at $(x_1,x_2)$ so it must be parallel to the normal vector at $(x_1, x_2)$. We can see that the above vector has length one and is pointing out from the region, so it must be the unit normal vector.

  • 0
    Thank you. This makes clear sense now.2017-02-19