Loss of Hölder continuity by solutions of the Euler equations

Click For Summary
The discussion focuses on the loss of Hölder continuity in solutions to the Euler equations, specifically examining the relationship between the squared norms of the velocity components and their spatial variables. The thesis proposes a bound involving constants Cα, Cβ, and Cγ, which relate to the Hölder continuity conditions. However, it is concluded that the final inequality intended to support the thesis is incorrect, particularly when Z is set to zero. This indicates a fundamental flaw in the proposed bounds for the continuity of the solutions. The findings highlight the complexities involved in establishing continuity properties for the Euler equations.
Coelum
Messages
97
Reaction score
32
Homework Statement
Show that if the initial conditions [itex]u(0,x), x\in\mathbb R^3[/itex] for a three-dimensional shear flow are such that [itex]u_x(0,y,z)[/itex] is [itex]\gamma[/itex]-Holder continuous and [itex]u_y(z)[/itex] is [itex]\beta[/itex]-Holder continuous, then the solution [itex]u(t,x)[/itex] for [itex]t\neq 0[/itex] need not be any smoother than [itex]\alpha=\beta\gamma<\min\{\beta,\gamma\}[/itex].
[This is part of project 8.5 from Craig's "A Course on Partial Differential Equations", AMS 2018]
Relevant Equations
A function [itex]v(y):\mathbb R^d\rightarrow\mathbb R^d [/itex] is a Holder continuous function of [itex]y[/itex] with Holder exponent [itex]0<\beta<1[/itex] if [itex]\exists C_\beta\in\mathbb R: \forall y,y'\in\mathbb R^d[/itex]
[tex]\begin{equation*}
\lVert v(y)-v(y')\rVert\leq C_\beta\lVert y-y'\rVert^\beta.
\end{equation*}[/tex]
A shear flow in 3D takes the form
[tex]\begin{align*}
u_x(t,x,y,z)&=u_x(y - t u_y(z)) \\
u_y(t,x,y,z)&=u_y(z) \\
u_z(t,x,y,z)&=c
\end{align*}[/tex]
where c is a constant.
Our thesis can be restated as follows: \exists C_\alpha\in\mathbb R_+ s.t. \forall w\in\mathbb R^2_+
\begin{align*}<br /> \lVert u(t,w)-u(t,w&#039;)\rVert^2<br /> \leq C_\alpha^2\lVert w-w&#039;\rVert^{2\alpha}<br /> \end{align*}
where w=(y,z) and \alpha=\beta\gamma.
We get an upper bound for each (squared) component of u(w) by applying the Holder continuity conditions:
\begin{align*}<br /> |u_x(t,w)-u_x(t,w&#039;)|^2&amp;=|u_x(y-tu_y(z))-u_x(y&#039;-tu_y(z&#039;))|^2 \\<br /> &amp;\leq C_\gamma^2|(y-tu_y(z))-(y&#039;-tu_y(z&#039;))|^{2\gamma} \\<br /> &amp;\leq C_\gamma^2|(y-y&#039;)-t(u_y(z)-u_y(z&#039;))|^{2\gamma} \\<br /> &amp;\leq C_\gamma^2|(y-y&#039;)-tC_\beta|z-z&#039;|^\beta|^{2\gamma} \\<br /> |u_y(t,w)-u_y(t,w&#039;)|^2&amp;=|u_y(z)-u_y(z&#039;)|^2 \\<br /> &amp;\leq C_\beta^2|z-z&#039;|^{2\beta} \\<br /> |u_z(t,w)-u_z(t,w&#039;)|^2&amp;=|c-c|^2 \\<br /> &amp;=0.<br /> \end{align*}
The RHS of the inequality we want to prove is bound by the sum of the bounds on each component:
\begin{equation*}<br /> \lVert u(t,w)-u(t,w&#039;)\rVert^2 <br /> \leq C_\gamma^2[(y-y&#039;)-tC_\beta|z-z&#039;|^\beta]^{2\gamma} + C_\beta^2|z-z&#039;|^{2\beta} .<br /> \end{equation*}
Now, in order to prove our thesis, we need to show that \exists C_\alpha\in\mathbb R_+ such that
\begin{equation*}<br /> C_\gamma^2[Y-tC_\beta|Z|^\beta]^{2\gamma} + C_\beta^2|Z|^{2\beta} <br /> \leq C_\alpha^2[Y^2+Z^2]^{\beta\gamma}<br /> \end{equation*}
where we let Y=y-y&#039;, \; Z=|z-z&#039;|.
Unfortunately, the last inequality is wrong - as it is easy to spot by letting Z=0.
 
At first, I derived that: $$\nabla \frac 1{\mu}=-\frac 1{{\mu}^3}\left((1-\beta^2)+\frac{\dot{\vec\beta}\cdot\vec R}c\right)\vec R$$ (dot means differentiation with respect to ##t'##). I assume this result is true because it gives valid result for magnetic field. To find electric field one should also derive partial derivative of ##\vec A## with respect to ##t##. I've used chain rule, substituted ##\vec A## and used derivative of product formula. $$\frac {\partial \vec A}{\partial t}=\frac...