Equation system


By Arun Iyer on Wednesday, May 22, 2002 - 06:23 pm:

Find all real solutions of the system
x2+y2+[2x y/(x+y)]=1


  ___
Öx+y
 
=x2-y


love arun


By David Loeffler on Wednesday, May 22, 2002 - 07:26 pm:

From the last part, we have

x+y = x4 - 2x2 y + y2

This we can rearrange into a quadratic in y, and we get y2 + -(2x2 +1)y + (x4 -x) = 0.

Solving this by the normal methods, a miracle happens, and we see that

2y = (2x2 +1) ± ( (1+2x2 )2 - 4(x4 -x) )1/2

2y = 2x2 +1 ± ( 1 + 4x2 + 4x4 - 4x4 + 4x )1/2

= 2x2 + 1 ± (2x+1)

So y is x2 + x + 1 or x2 -x.

Each of these may be quickly substituted into the first equation, and the resulting equations can be factorised. One leads to the quartic
-3 + 2x + 2x2 - 2x3 + x4 =0
of which ±1 are roots, and the others may be found by dividing through by x2 -1.

The other case leads to the slightly more awkward equation

4x + 10x2 + 14x3 + 9x4 + 4x5 + x6 = 0

which is mildly awkward; 0 and -2 are obviously roots, and this leaves us with a mildly nasty quartic, 2 + 4x + 5x2 + 2x3 + x4

However, as soon as we try to depress this by substituting x = z - 1/2, to get rid of the cubic term, the whole thing falls to pieces and we are left with a quadratic in z2 , z4 + 7/2 z2 + 17/16 = 0, which isn't difficult to solve. (Of course this will have only complex roots for z anyway, which is a bit irritating.)

This will give you 10 solutions for x and y. Six are complex and can be ignored immediately, and the remainder are easily checked to see if they fit the second equation (we might have introduced spurious solutions in the squaring process).

In fact, we are left with only two solutions: (x,y) = (1,0) and (-2,3).

David


By Arun Iyer on Friday, May 24, 2002 - 11:25 am:

Thanks David...

love arun