0

Five points in a vertical plane are given

$$ P1(0,0,-p),\, P2(0,0,0),\,I(x,y,0),\,Q1(a,b,q)\,,Q2(a,b,0)\,; $$

EDIT 1&2:

Simplify both ${x,y}$ to express them in terms of $(p,q,a,b,\mu)$ such that the following obtained result [Cross ratio of lengths for $\mu>1$ ] is the given starting point of Snell's Law.

$$\dfrac{\over{P2-I}} {\over{P1-I}} = \dfrac{\over{Q2-I}} {\over{Q1-I}} \mu$$

Narasimham
  • 42,260
  • Do you have any attempt at a solution? – Nitin Oct 18 '19 at 20:56
  • I started to setup with Fermat principle this result. – Narasimham Oct 18 '19 at 21:05
  • Not sure I understant: Fermat principle states that the light goes through the quickest path. Knowing the speed in both medium you can then optimize for time and get directly Snell-Descartes' law. Is it what you are trying to do? – Jean-Claude Arbaut Oct 18 '19 at 22:16
  • Yes, after such a procedure we obtain Snell's law as you stated. Now all I want is a more elegant algebraic simplified solution... as I got into a mess... – Narasimham Oct 18 '19 at 22:31

1 Answers1

1

Let's see about the mess.

I'll change the notation a little bit since I don't know exactly which is which in your question.

Let $A(x_1,y_1)$, $B(x_2,y_2)$ two points in two media, with an interface at $y=0$. Assume $y_1>0$ and $y_2<0$. You want to find a ray (or a guy, or anything), such that in the first medium its speed if $\nu_1$, while in the second medium its speed is $\nu_2$. You want to minimize the time to go from $A$ to $B$.

In a homogeneous medium, the shortest path is a line, so we are really looking for a path made of two lines, which both intersect the interface at some point $I(a,0)$. The unknown is $a$.

Let $d_1=AI$ and $d_2=IB$.

What's the time spent?

$$t(a)=\frac{d_1}{\nu_1}+\frac{d_2}{\nu_2}=\frac{1}{\nu_1} \sqrt{(a-x_1)^2+y_1^2}+\frac{1}{\nu_2} \sqrt{(x_2-a)^2+y_2^2}$$

We first see that $t$ is positive, $C^{\infty}$, and $t(a)\to+\infty$ as $a\to\pm\infty$, so there must be a minimum, and the first order condition means that at the minimum $t'(a)=0$. Also, geometrically, it's obvious that $a\in[x_1,x_2]$, otherwise there is a point in $[x_1,x_2]$ which would take less time (if $a_{min}<x_1$ take $a=x_1$, and if $a_{min}>x_2$ take $a=x_2$).

Here,

$$t'(a)=\frac{1}{\nu_1} \frac{a-x_1}{\sqrt{(a-x_1)^2+y_1^2}}+\frac{1}{\nu_2} \frac{a-x_2}{\sqrt{(x_2-a)^2+y_2^2}}=\frac{1}{\nu_1}\frac{a-x_1}{d_1}+\frac{1}{\nu_2}\frac{a-x_2}{d_2}$$

The equation to solve is thus

$$\frac{1}{\nu_1} \frac{a-x_1}{d_1}=\frac{1}{\nu_2} \frac{x_2-a}{d_2}$$

Let's have a look at what it means. In the right triangle $AIP$ with $P(a,y_1)$, the angle $\theta_1=AIP$ has sine:

$$\sin \theta_1=\frac{a-x_1}{d_1}$$

Likewise, in the right triangle $BIQ$ with $Q(a,y_2)$, the sine of $\theta_2=\widehat{BIQ}$ is:

$$\sin \theta_2=\frac{x_2-a}{d_2}$$

That is, at any local minimum of $t$, we must have

$$\frac{\sin\theta_1}{\nu_1}=\frac{\sin\theta_2}{\nu_2}$$

Which also proves that there is a unique local minimum (hence it's a global minimum), since $\sin\theta_1$ is an increasing function of $a\in[x_1,x_2]$, while $\sin\theta_2$ is a decreasing function of $a$.

Jean-Claude Arbaut
  • 23,601
  • 7
  • 53
  • 88
  • Thanks for the effort taken to explain how Snell's Law. However I have given this as start point already in my question and am looking to an algebraic simplification further on. – Narasimham Oct 19 '19 at 12:59
  • @Narasimham Maybe I didn't understand: you want to find something simpler than $\nu_2\sin\theta_1=\nu_1\sin\theta_2$? How much simpler? – Jean-Claude Arbaut Oct 19 '19 at 13:08
  • When we are given two points one inside a dense medium and another outside.it ..what is required is $x_I= f(a,b,p,q,\mu),, y_I= g(a,b,p,q,\mu), $ – Narasimham Oct 19 '19 at 13:15
  • @Narasimham It's possible to write $x_I$ using a root of a polynomial of degree $4$. While the polynomial is not complicated, the roots are not nice. But it seems that answer was already given by Christian Blatter in this thread, where you also posted an answer. – Jean-Claude Arbaut Oct 19 '19 at 14:53