I'm working on the differential equation: $$u_t(t,x,y) = u_y(t,x,y) + u_{xy}(t,1,1) \cdot u_x(t,x,y) $$
I haven't seen anything like this before in my courses, any ideas of what we could do? Would characteristics be appropriate here?
Thanks!
I'm working on the differential equation: $$u_t(t,x,y) = u_y(t,x,y) + u_{xy}(t,1,1) \cdot u_x(t,x,y) $$
I haven't seen anything like this before in my courses, any ideas of what we could do? Would characteristics be appropriate here?
Thanks!
In the first step, the pde can be easily "solved" by the method of characteristics to achieve
$$ u(t,x,y) = C(x+\phi(t),y+t)\\ \phi'(t)=u_{xy}(t,1,1) $$ Assuming you have inital conditions at $t=0$, namely $u(0,x,y)=C(x+\phi(0),y)=u_0(x,y)$ we can eliminate $C$ and end up with $$ u(t,x,y)=u_0(x-\phi(0)+\phi(t),t+y)\\ \phi'(t)=(u_0)_{xy}(1-\phi(0)+\phi(t),1+t) $$ So, if you have a specific sufficiently smooth initial condition $u_0$ you can solve the second equation, get $\phi(t)$ and subsequently get an (explicit) solution $u$.
Example for $u_0=x^2y$: $$ \Rightarrow \phi'(t)=2(1-\phi(0)+\phi(t))\\ \Rightarrow\phi(t)=\phi(0)+e^{2t}-1\\ \Rightarrow u(t,x,y)=(x+e^{2t}-1)^2(t+y) $$
Note: this relatively easy approach works only for $\mathbb{R}^+\times\mathbb{R}^2$. Otherwise the first part will remain the same but the function $C$ will look different.