Can we just use a part of the boundary?

  • MHB
  • Thread starter evinda
  • Start date
  • Tags
    Boundary
In summary: Crying)In summary, we discussed the solution of the problem $Lu=0$ in $\Omega$ where $\partial{\Omega}=S_1 \cup S_2$, $u|_{S_1}=0$, and $\frac{\partial{u}}{\partial{\mathcal{v}}}|_{S_2}=0$. We considered the interior sphere condition of $\Omega$ and the theorem stating that if $u$ does not achieve its maximum or minimum in $\Omega$, then $u$ must be constant. However, we realized that this theorem cannot be directly applied in our case and further analysis is needed. We also discussed two lemmas that involve boundary conditions with derivatives, but it is unclear how they can
  • #1
evinda
Gold Member
MHB
3,836
0
Hello! (Wave)

We consider an elliptic operator $L$ in the space $\Omega$ with $c(x) \leq 0$. We suppose that $\partial{\Omega}=S_1 \cup S_2$. What can we say about the solution of the following problem?

$$Lu=0 \text{ in } \Omega \\ u|_{S_1}=0, \frac{\partial{u}}{\partial{\mathcal{v}}}|_{S_2}=0$$

The space $\Omega$ satisfies the interior sphere condition and $\mathcal{v}$ is unit normal to $S_2$.

( In general, if $L$ is an elliptic operator, then $Lu=\sum_{i,j=1}^n a_{ij}(x) u_{x_i x_j}+ \sum_{i=1}^n \beta_i(x) u_{x_i}+cu$)

I thought to use the following theorem:

Theorem: Suppose that $u \in C^2(\Omega)$ satisfies in $\Omega$ the relation $Lu \geq 0$ ( $Lu \leq 0$). We suppose that $\Omega$ satisfies the interior sphere condition.
If $c \leq 0$ then $u$ does not achieve its positive maximum in $\Omega$, i.e. in $\overline{\Omega} \setminus{\partial{\Omega}}$ (negative minimum) if it is not constant.

Applying the above theorem, we have that $u$ does not achieve its positive maximum in $\Omega$ if it is not constant.
That means that either $u \leq 0$ or $u=0$ because of the fact that $u|_{S_1}=0$.

We also have that $u$ does not achive its negative maximum in $\Omega$ if it is not constant.
That means that either $u \geq 0$ or $u=0$ because of the fact that $u|_{S_1}=0$.

So we deduce that $u=0$ in $\overline{\Omega}$.

But is this right, given that $S_1$ is just a part of the boundary? (Thinking)
 
Physics news on Phys.org
  • #2
evinda said:
Hello! (Wave)

We consider an elliptic operator $L$ in the space $\Omega$ with $c(x) \leq 0$. We suppose that $\partial{\Omega}=S_1 \cup S_2$. What can we say about the solution of the following problem?

$$Lu=0 \text{ in } \Omega \\ u|_{S_1}=0, \frac{\partial{u}}{\partial{\mathcal{v}}}|_{S_2}=0$$

The space $\Omega$ satisfies the interior sphere condition and $\mathcal{v}$ is unit normal to $S_2$.

( In general, if $L$ is an elliptic operator, then $Lu=\sum_{i,j=1}^n a_{ij}(x) u_{x_i x_j}+ \sum_{i=1}^n \beta_i(x) u_{x_i}+cu$)

I thought to use the following theorem:

Theorem: Suppose that $u \in C^2(\Omega)$ satisfies in $\Omega$ the relation $Lu \geq 0$ ( $Lu \leq 0$). We suppose that $\Omega$ satisfies the interior sphere condition.
If $c \leq 0$ then $u$ does not achieve its positive maximum in $\Omega$, i.e. in $\overline{\Omega} \setminus{\partial{\Omega}}$ (negative minimum) if it is not constant.

Applying the above theorem, we have that $u$ does not achieve its positive maximum in $\Omega$ if it is not constant.
That means that either $u \leq 0$ or $u=0$ because of the fact that $u|_{S_1}=0$.

We also have that $u$ does not achive its negative maximum in $\Omega$ if it is not constant.
That means that either $u \geq 0$ or $u=0$ because of the fact that $u|_{S_1}=0$.

So we deduce that $u=0$ in $\overline{\Omega}$.

But is this right, given that $S_1$ is just a part of the boundary? (Thinking)

Hey evinda! (Smile)

Let's pick an example.
Suppose we pick $\Omega=(0,1)$, $S_1=\{0\}$, $S_2=\{1\}$, and $u$ as in:
[TIKZ][scale=5,font=\large]
\draw (0,0) -- node[below] {$\Omega$} (1,0);
\draw (0,0.05) -- (0,-0.05) node[below] {$S_1$} (1,0.05) -- (1,-0.05) node[below] {$S_2$};
\draw[ultra thick,blue] (0,0) parabola bend (1,0.5) (1,0.5) node
{$u$};
[/TIKZ]
Could $u$ be a solution? (Wondering)​
 
  • #3
I like Serena said:
Hey evinda! (Smile)

Let's pick an example.
Suppose we pick $\Omega=(0,1)$, $S_1=\{0\}$, $S_2=\{1\}$, and $u$ as in:
[TIKZ][scale=5,font=\large]
\draw (0,0) -- node[below] {$\Omega$} (1,0);
\draw (0,0.05) -- (0,-0.05) node[below] {$S_1$} (1,0.05) -- (1,-0.05) node[below] {$S_2$};
\draw[ultra thick,blue] (0,0) parabola bend (1,0.5) (1,0.5) node
{$u$};
[/TIKZ]
Could $u$ be a solution? (Wondering)​


How can we find the value of $\frac{\partial{u}}{\partial{v}}|_{S_2}$ ? (Thinking)​
 
  • #4
evinda said:
How can we find the value of $\frac{\partial{u}}{\partial{v}}|_{S_2}$ ? (Thinking)

Don't we have $v=(1)$ and $\frac{\partial{u}}{\partial{v}}=u'$? (Wondering)
 
  • #5
I like Serena said:
Don't we have $v=(1)$ and $\frac{\partial{u}}{\partial{v}}=u'$? (Wondering)

So we have that $u'=0$. So if it holds $Lu=0$ in $\Omega$ then in the way that I used the proposition, we would get that $u=0$ or $u \leq 0$, which is wrong. (Worried)
 
  • #6
evinda said:
So we have that $u'=0$. So if it holds $Lu=0$ in $\Omega$ then in the way that I used the proposition, we would get that $u=0$ or $u \leq 0$, which is wrong. (Worried)

What went wrong? (Crying)
 
  • #7
I like Serena said:
What went wrong? (Crying)

Having $\frac{\partial{u}}{\partial{v}}|_{S_2}=0$ doesn't imply that $u=0$ in $S_2$.
And we cannot just use the fact that $u|_{S_1}=0$, right? But how can we use the fact that $\frac{\partial{u}}{\partial{v}}|_{S_2}=0$ at the theorem? (Thinking) Or can't we use it?
 
  • #8
evinda said:
Having $\frac{\partial{u}}{\partial{v}}|_{S_2}=0$ doesn't imply that $u=0$ in $S_2$.
And we cannot just use the fact that $u|_{S_1}=0$, right? But how can we use the fact that $\frac{\partial{u}}{\partial{v}}|_{S_2}=0$ at the theorem? (Thinking) Or can't we use it?

Yes. (Worried)

Which theorems do you have that include a boundary condition with the derivative? (Wondering)
 
  • #9
There are the following lemmas:

Lemma 1: Let $L$ be an elliptic operator , $c \equiv 0, Lu \geq 0$ in $\Omega \subset \mathbb{R}^n$. We also suppose that $\Omega$ satisfies the interior sphere condition in $x_0 \in \partial{\Omega}$ and
  1. at $x_0$ u is continuous
  2. $u(x_0)> u(x) \forall x \in \Omega$

then if at $x_0$ $\exists \frac{\partial{u}}{\partial{v}}$ then $\frac{\partial{u}}{\partial{v}}(x_0)>0$.

If $c \leq 0$ the same holds if $u(x_0) \geq 0$.Lemma 2: Let $L$ be an elliptic operator , $c \equiv 0, Lu \leq 0$ in $\Omega \subset \mathbb{R}^n$. We also suppose that $\Omega$ satisfies the interior sphere condition in $x_0 \in \partial{\Omega}$ and
  1. at $x_0$ u is continuous
  2. $u(x_0)< u(x) \forall x \in \Omega$

then if at $x_0$ $\exists \frac{\partial{u}}{\partial{v}}$ then $\frac{\partial{u}}{\partial{v}}(x_0)<0$.

If $c \leq 0$ the same holds if $u(x_0) \leq 0$.But... in our case we have $\partial{\Omega}=S_1 \cup S_2$. What can we do? (Worried)
 
  • #10
Suppose we try to apply Lemma 1 to the example I gave... (Thinking)

[TIKZ][scale=5,font=\Large]
\draw[very thick] (0,0) -- node[below] {$\Omega$} (1,0);
\draw[ultra thick,green!70!black] (0,0) parabola bend (1,0.5) (1,0.5) node
{$u$};
\foreach \x/\label in {0/S_1,1/S_2} \filldraw (\x,0) circle (0.5pt) [yshift=-0.3pt] node[below] {$\x$} [yshift=0.6pt] node[above] {$\label$};[/TIKZ]

That is, $x_0=1$ and we suppose that $u(x_0)>0$.
Since $\frac{\partial{u}}{\partial{v}}(x_0)=0$ it follows from the lemma that $u(x_0)\le u(x)$.
But $u(0)=0$, so we get a contradiction!​
 
  • #11
I like Serena said:
Suppose we try to apply Lemma 1 to the example I gave... (Thinking)

[TIKZ][scale=5,font=\Large]
\draw[very thick] (0,0) -- node[below] {$\Omega$} (1,0);
\draw[ultra thick,green!70!black] (0,0) parabola bend (1,0.5) (1,0.5) node
{$u$};
\foreach \x/\label in {0/S_1,1/S_2} \filldraw (\x,0) circle (0.5pt) [yshift=-0.3pt] node[below] {$\x$} [yshift=0.6pt] node[above] {$\label$};[/TIKZ]

That is, $x_0=1$ and we suppose that $u(x_0)>0$.
Since $\frac{\partial{u}}{\partial{v}}(x_0)=0$ it follows from the lemma that $u(x_0)\le u(x)$.
But $u(0)=0$, so we get a contradiction!​


Isn't $u(x_0)>u(x)$ a condition so that we can apply the lemma?

- - - Updated - - -

From the lemma, we get that it has to hold that $\frac{\partial{u}}{\partial{v}}|_{S_2}>0$, that is a contradiction. Right? (Thinking)​
 
  • #12
evinda said:
Isn't $u(x_0)>u(x)$ a condition so that we can apply the lemma?

From the lemma, we get that it has to hold that $\frac{\partial{u}}{\partial{v}}|_{\partial{\Omega}}=0$, that is a contradiction. Right? (Thinking)

Let's suppose that $u(x_0)>u(x)$, as is the case in the example, then we can apply the lemma and get that contradiction.
So one of the conditions of the lemma must be false.
The only one that qualifies is: $u(x_0)> u(x) \forall x \in \Omega$.
So we get that $\exists x \in\Omega: u(x_0)\le u(x)$.

To be fair, that's only at least one $x$ and not all of them, so my argument that $u(x_0) > u(0)$ is not sufficient. (Blush)

Still, we now have a supposed maximum $u(x_0) > 0$ on the boundary, and a point $x\in\Omega$ that is at least as high.
Can't we apply the lemma that says that if we have an internal maximum, that then the only solution is a constant function? (Wondering)
 
  • #13
So for the general case:

We suppose that $u(x_0)>u(x)$ for some $x_0 \in S_2$ and for all $x \in \Omega$.

Then applying the lemma we get that if $\exists \frac{\partial{u}}{\partial{v}}$ at $x_0$ then $ \frac{\partial{u}}{\partial{v}}(x_0)>0$.

Since $x_0 \in S_2$ is arbitrary, we get that it has to hold $\frac{\partial{u}}{\partial{v}}|_{S_2}>0$.

But we are given that $\frac{\partial{u}}{\partial{v}}|_{S_2}=0$.

So one of the conditions of the lemma must be false.

The only one that qualifies is: $u(x_0)>u(x), \forall x \in \Omega $.

So we get that $\exists x \in \Omega$: $u(x_0) \leq u(x)$.

Since we have found an internal maximum, we deduce that $u$ is constant.

Since $u|_{S_1}=0$ and $u$ is continuous, we deduce that $u=0$ in $\overline{\Omega}$.

Is it right? Could I improve something? (Thinking)
 
  • #14
evinda said:
So for the general case:

We suppose that $u(x_0)>u(x)$ for some $x_0 \in S_2$ and for all $x \in \Omega$.

Then applying the lemma we get that if $\exists \frac{\partial{u}}{\partial{v}}$ at $x_0$ then $ \frac{\partial{u}}{\partial{v}}(x_0)>0$.

Since $x_0 \in S_2$ is arbitrary, we get that it has to hold $\frac{\partial{u}}{\partial{v}}|_{S_2}>0$.

But we are given that $\frac{\partial{u}}{\partial{v}}|_{S_2}=0$.

So one of the conditions of the lemma must be false.

The only one that qualifies is: $u(x_0)>u(x), \forall x \in \Omega $.

So we get that $\exists x \in \Omega$: $u(x_0) \leq u(x)$.

Since we have found an internal maximum, we deduce that $u$ is constant.

Since $u|_{S_1}=0$ and $u$ is continuous, we deduce that $u=0$ in $\overline{\Omega}$.

Is it right? Could I improve something? (Thinking)

We assume that $S_1 \ne \varnothing$, which I believe is not given is it? (Wondering)
And we also assume that $S_2 \ne \varnothing$.

Oh, and don't we also assume that $u(x_0)>0$?
Because otherwise it $u(x_0)$ might still actually be a minimum.

Btw, which lemma are we applying exactly to deduce that $u$ is constant?
I've gotten lost between the harmonic, subharmonic, and elliptical operators. (Sweating)

Can we assume that $\Omega$ is open and connected? Because I think that matters as well.
Suppose $S_1$ and $S_2$ are boundaries of unconnected parts of $\Omega$! (Worried)

I think we need to distinguish a couple more cases. (Thinking)
 
  • #15
I like Serena said:
We assume that $S_1 \ne \varnothing$, which I believe is not given is it? (Wondering)
And we also assume that $S_2 \ne \varnothing$.

No, it is not given.

I like Serena said:
Oh, and don't we also assume that $u(x_0)>0$?
Because otherwise it $u(x_0)$ might still actually be a minimum.

And do we apply afterwards the lemma 2 and suppose that $u(x_0)<0$ ? Or can't we do so? (Worried)

I like Serena said:
Btw, which lemma are we applying exactly to deduce that $u$ is constant?
I've gotten lost between the harmonic, subharmonic, and elliptical operators. (Sweating)

We have the following theorem:

Theorem: We suppose that $u \in C^2(\Omega)$ satisfies in the space $\Omega$ the relation $Lu \geq 0$ ($Lu \leq 0$). We suppose that $\Omega$ satisfies the interior sphere condition.
If $c \leq 0$ then $u(x)$ does not achieve its positive maximum in $\Omega$, i.e. in $\overline{\Omega} \setminus{\partial{\Omega}}$ (negative minimum) if it is not constant.
If $c \equiv 0$ then $u$ does not achieve its maximum in $\Omega$ (minimum) if it is not constant.

I like Serena said:
Can we assume that $\Omega$ is open and connected? Because I think that matters as well.
Suppose $S_1$ and $S_2$ are boundaries of unconnected parts of $\Omega$! (Worried)

I think that we assume that $\Omega$ is connected, as also at the proof of the maximum principle value.

I like Serena said:
I think we need to distinguish a couple more cases. (Thinking)

What cases? (Thinking)
 
  • #16
evinda said:
I think that we assume that $\Omega$ is connected, as also at the proof of the maximum principle value.
And bounded? So that we'll have a maximum and a minimum? (Wondering)

What cases? (Thinking)
I think:
  1. $S_1=\varnothing$
  2. $S_2=\varnothing$
  3. $S_1 \cap S_2 \ne \varnothing \wedge \exists x_0\in S_2:u(x_0)=\displaystyle\max_{\xi\in S_2}(u(\xi)) \ge 0$
  4. $S_1 \cap S_2 \ne \varnothing \wedge \exists x_0\in S_2:u(x_0)=\displaystyle\min_{\xi\in S_2}(u(\xi)) < 0$
What do you think? (Thinking)
 
  • #17
I like Serena said:
And bounded? So that we'll have a maximum and a minimum? (Wondering)

At some propositions, it is stated that $\Omega$ is bounded , at some others only that it satisfies the interior sphere condition.

I like Serena said:
I think:
  1. $S_1=\varnothing$
  2. $S_2=\varnothing$
  3. $S_1 \cap S_2 \ne \varnothing \wedge \exists x_0\in S_2:u(x_0)=\displaystyle\max_{\xi\in S_2}(u(\xi)) \ge 0$
  4. $S_1 \cap S_2 \ne \varnothing \wedge \exists x_0\in S_2:u(x_0)=\displaystyle\min_{\xi\in S_2}(u(\xi)) < 0$
What do you think? (Thinking)

Couldn't it also hold that $S_1 \cap S_2= \varnothing$ ? (Thinking)

Also couldn't we have at the boundary both a maximum and a minimum? Or do we always need just one such condition?

Do this four cases cover all the ones that we could have? (Thinking)
 
  • #18
evinda said:
At some propositions, it is stated that $\Omega$ is bounded , at some others only that it satisfies the interior sphere condition.

Couldn't it also hold that $S_1 \cap S_2= \varnothing$ ? (Thinking)

Also couldn't we have at the boundary both a maximum and a minimum? Or do we always need just one such condition?

Do this four cases cover all the ones that we could have? (Thinking)

Oh yes. It should be:
  1. $S_1=\varnothing$
  2. $S_2=\varnothing$
  3. $S_1\ne \varnothing \wedge S_2 \ne \varnothing \wedge \exists x_0\in S_2:u(x_0)=\displaystyle\max_{\xi\in S_2}(u(\xi)) \ge 0$
  4. $S_1\ne \varnothing \wedge S_2 \ne \varnothing \wedge \exists x_0\in S_2:u(x_0)=\displaystyle\min_{\xi\in S_2}(u(\xi)) < 0$
  5. $S_2 \ne \varnothing \wedge (S_2$ does not contain its supremum or $S_2$ does not contain its infimum$)$
Did we miss anything?
Or would this cover all possible situations?
Well... if $S_2$ is not bounded it could be that there is no maximum or minimum...
Oh! And if $S_2$ is open, it can be that its supremum or infimum is not part in the set... (Thinking)

And yes, we can and will always have a maximum and a minimum, so 3 and 4 are not mutually exclusive.
It's just a matter if we have at least one point above 0 or not.

Anway, we'll still have to see if we can solve each of these cases. (Sweating)

They look like this (no interior minimum or maximum):
[TIKZ][scale=5,font=\Large]
%preamble \usetikzlibrary{arrows}
\draw[gray,thin,-triangle 60] (0,-0.3) -- (0,0.5) node
{$u$};
\draw[very thick] (0,0) -- node[below] {$\Omega$} (1,0) node
{2};
\draw[ultra thick,red!70!black] (0,0.2) cos (0.5,0.3) sin (1,0.4) node
{1};
\draw[ultra thick,green!70!black] (0,0) parabola bend (1,0.25) (1,0.25) node
{3};
\draw[ultra thick,blue] (0,0) parabola bend (1,-0.25) (1,-0.25) node
{4};
\foreach \x/\label in {0/S_1,1/S_2} \filldraw (\x,0) circle (0.5pt);
[/TIKZ]​
 
Last edited:
  • #19
For the case 3, we have the following:

We suppose that $u(x_0)>u(x)$ , $\forall x \in \Omega$.

Then applying the lemma $1$ we get that if $\exists \frac{\partial{u}}{\partial{v}}$ at $x_0$ then $ \frac{\partial{u}}{\partial{v}}(x_0)>0$.

But we are given that $\frac{\partial{u}}{\partial{v}}|_{S_2}=0$, so $ \frac{\partial{u}}{\partial{v}}(x_0)=0$.

So one of the conditions of the lemma must be false.

The only one that qualifies is: $u(x_0)>u(x), \forall x \in \Omega $.

So we get that $\exists x \in \Omega$: $u(x_0) \leq u(x)$.

Since we have found an internal maximum, we deduce that $u$ is constant.

Since $u|_{S_1}=0$ and $u$ is continuous, we deduce that $u=0$ in $\overline{\Omega}$.And for the case 4, we have the following:

We suppose that $u(x_0)<u(x)$ , $\forall x \in \Omega$.

Then applying the lemma $2$ we get that if $\exists \frac{\partial{u}}{\partial{v}}$ at $x_0$ then $ \frac{\partial{u}}{\partial{v}}(x_0)<0$.

But we are given that $\frac{\partial{u}}{\partial{v}}|_{S_2}=0$, so $ \frac{\partial{u}}{\partial{v}}(x_0)=0$.

So one of the conditions of the lemma must be false.

The only one that qualifies is: $u(x_0)<u(x), \forall x \in \Omega $.

So we get that $\exists x \in \Omega$: $u(x_0) \geq u(x)$.

Since we have found an internal minimum, we deduce that $u$ is constant.

Since $u|_{S_1}=0$ and $u$ is continuous, we deduce that $u=0$ in $\overline{\Omega}$.Right so far? (Thinking)
 
  • #20
I think we also need to assume, without loss of generality, that $\Omega$ is an open set.
Otherwise our $x$ is not necessarily an internal point.

Other than that it looks good to me! (Mmm)
 
  • #21
Ok. At the first case, when $S_1=\varnothing$ we have the following problem

$$Lu=0 \text{ in } \Omega \\ \frac{\partial{u}}{\partial{v}}|_{S_2}=0$$

We suppose that for some $x_0 \in S_2$ we have that $u(x_0)>u(x) , \forall x \in \Omega$.

Then by lemma 1, if at $x_0$ $\exists \frac{\partial{u}}{\partial{v}}$ then $\frac{\partial{u}}{\partial{v}}(x_0)>0$.

But we are given that $ \frac{\partial{u}}{\partial{v}}|_{S_2}=0$ and so $\frac{\partial{u}}{\partial{v}}(x_0)=0$.

So we deduce that $\exists x \in \Omega$ such that $u(x_0) \leq u(x)$.

So we have found an internal maximum and so we deduce that $u$ is constant.

Applying the lemma 2, we would get the same, right?

In this case, we cannot find the value of the constant function $u$, right?
 
  • #22
Agreed. The solution is the set of constant functions. (Mmm)
 
  • #23
I like Serena said:
Agreed. The solution is the set of constant functions. (Mmm)

Nice (Smirk)

At the second case when $S_2=\varnothing$, we have the following problem

$$Lu=0 \text{ in } \Omega \\ u|_{S_1}=0$$

the only solution of which is $u(x)=0$. Right?

- - - Updated - - -

What can we say about the case 5? (Thinking)
 
  • #24
evinda said:
Nice (Smirk)

At the second case when $S_2=\varnothing$, we have the following problem

$$Lu=0 \text{ in } \Omega \\ u|_{S_1}=0$$

the only solution of which is $u(x)=0$. Right?

- - - Updated - - -

What can we say about the case 5? (Thinking)
Right! (Nod)

- - - Updated - - -

If $\Omega$ is not bounded, we could actually also have yet another case: $S_1=S_2=\varnothing$.
As an example, pick $\Omega=\mathbb R$ and $Lu=u''-u$.
The solution is $u(x)=c_1 e^x + c_2 e^{-x}$, which is unbounded as well.
[TIKZ][scale=3,font=\Large]
%preamble \usetikzlibrary{arrows}
\draw[gray,thin,-triangle 60] (0,-0.3) -- (0,1) node
{$u$};
\draw[very thick,triangle 60-triangle 60] (-1,0) -- node[below] {$\Omega$} (1,0);
\draw[ultra thick,red!70!black,domain=-1:1,variable=\x] plot (\x,{e^(\x)/3});
[/TIKZ]
Does that match the problem statement and is it a proper solution?
Or am I wrong? (Wondering)

Or as another example with unbounded $\Omega$, we could have $\Omega=(0,\infty), S_1=\{0\},Lu=u''-u$ with solution $u(x)=c_1 \sinh x$.
That is:
[TIKZ][scale=2,font=\Large]
%preamble \usetikzlibrary{arrows}
\draw[gray,thin,-triangle 60] (0,-0.3) -- (0,1.2) node
{$u$};
\draw[very thick,-triangle 60] (0,0) -- node[below] {$\Omega$} (3,0);
\draw[ultra thick,red!70!black,domain=0:3,variable=\x] plot (\x,{(e^\x-e^(-\x))/20});
\foreach \x/\label in {0/S_1} \filldraw (\x,0) circle (1pt) node
{$S_1$};
[/TIKZ]

Maybe we should suppose that $\Omega$ is bounded? (Wondering)
Do your notes perchance say anything about $\Omega$ in a prologue, in chapter 1, or at the beginning of the chapter? (Wondering)

- - - Updated - - -

Alternatively, $S_2$ could be an open set with its supremum on its boundary.
Assuming that $\Omega$ is connected, that would mean that the supremum is in $S_1$.
What could we say then? (Wondering)​
 
Last edited:
  • #25
I like Serena said:
If $\Omega$ is not bounded, we could actually also have yet another case: $S_1=S_2=\varnothing$.
As an example, pick $\Omega=\mathbb R$ and $Lu=u''-u$.
The solution is $u(x)=c_1 e^x + c_2 e^{-x}$, which is unbounded as well.
[TIKZ][scale=3,font=\Large]
\draw[gray,thin,-triangle 60] (0,-0.3) -- (0,1) node
{$u$};
\draw[very thick,triangle 60-triangle 60] (-1,0) -- node[below] {$\Omega$} (1,0);
\draw[ultra thick,red!70!black,domain=-1:1,variable=\x] plot (\x,{e^(\x)/3});
[/TIKZ]
Does that match the problem statement and is it a proper solution?
Or am I wrong? (Wondering)

Or as another example with unbounded $\Omega$, we could have $\Omega=(0,\infty), S_1=\{0\},Lu=u''-u$ with solution $u(x)=c_1 \sinh x$.
That is:
[TIKZ][scale=2,font=\Large]
\draw[gray,thin,-triangle 60] (0,-0.3) -- (0,1.2) node
{$u$};
\draw[very thick,-triangle 60] (0,0) -- node[below] {$\Omega$} (3,0);
\draw[ultra thick,red!70!black,domain=0:3,variable=\x] plot (\x,{(e^\x-e^(-\x))/20});
\foreach \x/\label in {0/S_1} \filldraw (\x,0) circle (1pt) node
{$S_1$};
[/TIKZ]

Maybe we should suppose that $\Omega$ is bounded? (Wondering)
Do your notes perchance say anything about $\Omega$ in a prologue, in chapter 1, or at the beginning of the chapter? (Wondering)​


At some theorem, it is required that $\Omega$ is bounded at some others that it satisfies the interior sphere condition.
So we need that $\Omega$ is bounded in order the solution to be bounded? (Thinking)
I like Serena said:
Alternatively, $S_2$ could be an open set with its supremum on its boundary.
Assuming that $\Omega$ is connected, that would mean that the supremum is in $S_1$.
What could we say then? (Wondering)

Why is in this case the supremum in $S_1$?​
 
  • #26
By supposing that $\Omega$ is bounded, we have to check just the first four cases, right? (Thinking)
 
  • #27
I think that if we assume that the maximum and minimum value are achieved, then we get that the solution is constant. So the only solution of the problem is the zero function.
 
  • #28
evinda said:
At some theorem, it is required that $\Omega$ is bounded at some others that it satisfies the interior sphere condition.
So we need that $\Omega$ is bounded in order the solution to be bounded? (Thinking)

I believe so yes.
Why is in this case the supremum in $S_1$?

If we have a supremum, it means we can create a sequence that approaches this supremum in the limit.
Since such a sequence would be entirely inside $S_2$, the limit must be in $\overline {S_2}$, which is a subset of $\partial \Omega$.
So this supremum could be in the boundary of $S_2$ and not in $S_2$ itself, but if so, it must still be in $\partial \Omega$ and therefore in $S_1$.
Note that with continuity of $\frac{\partial u}{\partial v}$, the derivative will still be zero.

So I think we should rephrase cases 3 and 4 a little bit to cover this as well:

3. $S_1\ne \varnothing \wedge S_2 \ne \varnothing \wedge \exists x_0\in \partial\Omega: u(x_0)=\displaystyle\sup_{\xi\in S_2}(u(\xi)) \ge 0$
4. $S_1\ne \varnothing \wedge S_2 \ne \varnothing \wedge \exists x_0\in \partial\Omega:u(x_0)=\displaystyle\inf_{\xi\in S_2}(u(\xi)) < 0$
(Thinking)
evinda said:
By supposing that $\Omega$ is bounded, we have to check just the first four cases, right? (Thinking)

Yes, but with slightly modified cases 3 and 4 as I just mentioned.
evinda said:
I think that if we assume that the maximum and minimum value are achieved, then we get that the solution is constant. So the only solution of the problem is the zero function.

Yes, if minimum and maximum are achieved, then the solution is constant.
For it to be zero we need $S_1\ne\varnothing$ though. (Mmm)
 

FAQ: Can we just use a part of the boundary?

Can we use just a small portion of the boundary for our experiment?

This depends on the specific boundaries and conditions of the experiment. In some cases, using a small portion of the boundary may be sufficient for accurate results. However, in other cases, using only a small portion may not provide enough data or may alter the results. It is important to carefully consider the purpose and limitations of the experiment before deciding on using a part of the boundary.

Will using a part of the boundary affect our results?

Yes, using only a part of the boundary can affect the results of an experiment. Boundaries can play a crucial role in shaping the conditions and variables of an experiment. Altering or using only a small portion of the boundary can lead to different outcomes and may not accurately represent the full picture. It is important to carefully consider the potential impact on results before using a part of the boundary.

Are there any ethical concerns with using only a part of the boundary?

Yes, there may be ethical concerns with using only a part of the boundary. Boundaries can serve as protective and regulatory measures for certain ecosystems or populations. Using only a small portion of the boundary may disrupt or harm the natural balance and well-being of these systems. It is important to consider the potential ethical implications and seek proper permissions before using a part of the boundary.

Can we use a part of the boundary if it is not within our jurisdiction?

It is not recommended to use a part of the boundary that is not within your jurisdiction. Boundaries often have specific regulations and laws in place to protect and manage the area. Using a part of the boundary outside of your jurisdiction may not only be unethical, but it can also lead to legal consequences. It is important to respect boundaries and seek proper permissions before conducting any experiments.

What precautions should we take when using a part of the boundary?

When using a part of the boundary, it is important to take necessary precautions to minimize any potential negative impacts. This may include obtaining proper permissions, following ethical guidelines, and considering the potential consequences on the environment or populations within the boundary. It is also important to thoroughly document and report any research conducted on a part of the boundary for transparency and accountability.

Similar threads

Replies
21
Views
4K
Replies
18
Views
4K
Replies
1
Views
1K
Replies
21
Views
4K
Replies
16
Views
4K
Replies
1
Views
2K
Replies
1
Views
1K
Replies
16
Views
3K
Back
Top