Solution to second order partial differential equation

495 Views Asked by At

Suppose $Q$ is the solution to $Q_{tt}-Q_{xx} =0$.

Let $\bar{Q}$ be the solution to $Q_{tt}-Q_{xx}+k Q_t=0$

Is there any known relation between these two solution, given that in the limit of $k$ going to zero, both solution should approach each other?

1

There are 1 best solutions below

0
On BEST ANSWER

In order to avoid confusion two different notations $P(x,t)$ and $Q(x,t)$ will be used :

$P$ is the solution to $\quad P_{tt}-P_{xx} =0$.

$Q$ is the solution to $\quad Q_{tt}-Q_{xx}+k Q_t=0$

The general solution expressed on the form of integral is : $$Q(x,t)=\int_{\forall \lambda} F(\lambda)e^{\lambda t +\frac12\left(k+\sqrt{k^2+4\lambda^2} \right)x }d\lambda +\int_{\forall \lambda} G(\lambda)e^{\lambda t +\frac12\left(k-\sqrt{k^2+4\lambda^2} \right)x }d\lambda $$ $F(\lambda)$ and $G(\lambda)$ are arbitrary functions to be determined according to some boundary conditions (not mentioned in the wording of the question).

In case of $k=0$ the general solution is the preceding equation which is reduced to : $$P(x,t)=\int_{\forall \lambda} f(\lambda)e^{\lambda (t+x) }d\lambda + \int_{\forall \lambda} g(\lambda)e^{\lambda (t-x) }d\lambda$$ $f(\lambda)$ and $g(\lambda)$ are arbitrary functions to be determined according to some boundary conditions. There is no reason for $f=F$ and $g=G$.

Note that one can write equivalently $$P(x,t)=\Phi(t+x)+\Gamma(t-x)$$ where $\Phi$ and $\Gamma$ are arbitrary functions.

The question raised by the OP is if whether or not a general relationship exist between $P(x,t)$ and $Q(x,t)$.

Since both $P(x,t)$ and $Q(x,t)$ depends from boundary conditions it is doubtful that such a relationship exists because they are an infinity of possible different boundary conditions.

Supposing that the question be less general, but raised for only a particular boundary condition common to the two PDEs, it is difficult to give a definitive answer without the particular boundary condition be specified.

APPROXIMATE SOLUTION for $k$ close to $0.$ :

$\lambda t +\frac12\left(k\pm\sqrt{k^2+4\lambda^2} \right)x \simeq\lambda t +\frac12\left(k\pm 2\lambda \right)x $

$$Q(x,t)\simeq\int_{\forall \lambda} F(\lambda)e^{\lambda t +\frac12\left(k+2\lambda \right)x }d\lambda +\int_{\forall \lambda} G(\lambda)e^{\lambda t +\frac12\left(k-2\lambda \right)x }d\lambda $$

$$Q(x,t)\simeq\int_{\forall \lambda} F(\lambda)e^{\lambda (t+x) +\frac{k}{2} x }d\lambda +\int_{\forall \lambda} G(\lambda)e^{\lambda (t-x) +\frac{k}{2}x }d\lambda $$

$$Q(x,t)\simeq e^{\frac{k}{2} x}\int_{\forall \lambda} F(\lambda)e^{\lambda (t+x) }d\lambda +e^{\frac{k}{2} x}\int_{\forall \lambda} G(\lambda)e^{\lambda (t-x)}d\lambda $$

$$Q(x,t)\simeq e^{\frac{k}{2} x}\big( \Phi(t+x)+\Gamma(t-x)\big)+\epsilon(x,t)\qquad k\:\:\text{small}$$ $$Q(x,t)\simeq e^{\frac{k}{2} x}P(x,t)+\epsilon(x,t)$$ The magnitude of the deviation $\epsilon(x,t)$ is proportional to $k^2$.