How to solve partial differential equation of the form $\partial_x f\ (dx/dt)+\partial_t f =0$

62 Views Asked by At

In some of my clases, they teach me some function which describe infiltration in the soil. The equation is known as Kostiakov's equation. We denote the Kostiakov's equation as $I(t)$, which maps $I:\Bbb R_{\geq 0}\mapsto \Bbb R_{\leq 0}$ and is related by the formula:

$$I(t)=C(t)^m\quad \forall C<0 \quad \forall m\in (0,1)$$

I was trying to define a new equation with the same "form" as the previous equation but also for each value of $x$, which phisically means the position in the soil surface; and impose as condition a bijective function of $t$, such that $f:\Bbb R_{\geq 0}\mapsto \Bbb R_{\geq 0}$, where $f(t)$ phisically means the position of the "front" of the water moving in the soil surface. I intended to make an equation that would describe the total infiltration for each position $x$.

Then I define a new function $\hat I(t,x)$, $\hat I:\Bbb R_{\geq 0}^2\mapsto \Bbb R_{\leq 0}$ and the first thing I notice is that it should have the condition that $\hat I(t,f(t))=0 \quad \forall x> f(t) $, meaning that the function is $0$ when the "front" has not passed some position $x$. Second, it should have the condition that for a pair $(a,b)\in \Bbb R_{\geq 0}^2,\ \hat I(a,b)=I(a-f^{-1}(b))\quad \forall f^{-1}(b)\leq a$. And a third condition, which is $\hat I(t,0)=I(t)\quad \forall t\in\Bbb R_{\geq 0}$.

Just by observing the second condition, I guess that:

$$\hat I(t,x)=\hat I\left(t-f^{-1}(x)\right)=C\left(t-f^{-1}(x)\right)^m $$

and checked that fulfill the third condition. The first condition is fulfill if $\hat I$ is define as a piecewise function:

$$\hat I(t,x)= \begin{cases} C\left(t-f^{-1}(x)\right)^m & ,x\leq f(t)\\[2ex] 0 &,x>f(t) \end{cases} $$

Then I realize that if $f(t)=k\cdot t$ for a constant $k\in\Bbb R_{\geq 0}$ and taking the double partial derivatives with respect to $t$ and $x$ of $\hat I$, it satisfied the hyperbolic PDE:

$${\partial^2\hat I\over\partial x^2} - {1\over k^2}{\partial^2\hat I\over\partial t^2}=0$$

and so the solutions of that PDE are any function of the form $H_1\left(t-{x\over k}\right)+H_2\left(t+{x\over k}\right)$

Now if we use the general case for any $f(t)$ and taking partial derivatives it gives the follow equations:

1. $${\partial \hat I\over \partial x}={\partial\hat I\over\partial t}\left(-{dt\over dx}\right)$$

$$\Rightarrow {\partial \hat I\over \partial x}\dot{x}+{\partial\hat I\over\partial t}=0$$

2. $${\partial^2 \hat I\over \partial x^2}= {\partial^2\hat I\over\partial t^2}\left(dt\over dx\right)^2-{\partial\hat I\over\partial t}\left(d^2 t\over dx^2 \right)$$

$$={\partial^2\hat I\over\partial t^2} \left(1\over \dot{x}\right)^2+{\partial\hat I\over\partial t}{\ddot x\over{\left(\dot x\right)^3}} $$

$$={\partial^2\hat I\over\partial t^2} \left(1\over \dot{x}\right)^2+{\partial\hat I\over\partial x}{\ddot x\over{\left(\dot x\right)^2}} $$

$$\Rightarrow {\partial^2 \hat I\over \partial x^2}\left(\dot x\right)^2 -{\partial\hat I\over\partial x}{\ddot x} - {\partial^2\hat I\over\partial t^2} = 0$$

So my question is how one can resolve the first equation, and maybe the second. The last one looks very hard so I dont think there's a known method for that. Also, for what I read, maybe the function is the solution of a Delay PDE, but I don't understand that type of PDE so I don't know how to write the Delay PDE which has for a particular solution of the function I made. For last, if $\ddot x=0$, then $\dot x=c, c\in\Bbb R$, so the equation become the normal wave equation. And as can be interpreted, the solution could be related to traveling waves with non-constant velocity.