PDA

View Full Version : inequality


kakarukeys
Oct11-04, 09:47 AM
:confused:
a^2 - b^2 - c^2 - d^2 > 0
A^2 - B^2 - C^2 - D^2 = 1
A > 1
a > 0
Prove or disprove Aa + Bb + Cc + Dd > 0

after 3 days of trying I give up
can anyone give a clue?

kakarukeys
Oct11-04, 10:10 AM
what I have done:
I used (a + b)^2 = a^2 + b^2 + 2ab, but it leads to nothing.
I have tried Lagrange Multipliers, but the only extremum is a saddle point.

kakarukeys
Oct11-04, 10:11 AM
trying to disprove it
I found no counter-example.

TenaliRaman
Oct11-04, 10:12 AM
(a+A)^2 + (b+B)^2 + (c+C)^2 + (d+D)^2 > 0

-- AI

kakarukeys
Oct11-04, 10:26 AM
I can't see.
then?

Motifs
Oct11-04, 11:19 AM
Are b,c,d >0 and B,C,D>1 ?? if so, no need to prove anything since it is obvious.
If not, I think we can't say the given formula >0 or <0
You can try subsitute some values of b,c,d to guess a, then values for B,C,D to calculate A, you will see it.

Am I incorrect ?

Wong
Oct11-04, 11:37 AM
Let me try to give some motivation about the inequaility.

Consider the equivalent problem in 2D. That is,
a^{2}-b^{2}>0, a>1
A^2-B^2=1, A>1
Prove or disprove aA+bB>0
In 2D, the region represented by the first equation is like a "quadrant" rotated by 45 degrees. (Try to plot it.) The set represented by the second equation is one component of a hyperbola included in the first region. Note that Aa+Bb is just the usual *dot product* between vector (a, b) and (A, B). Can you see why the inequality holds in the case?

Try to do the problem for 3D and then generalise it.

Motifs
Oct11-04, 12:20 PM
Wow, my gal bit me .:redface:

kakarukeys
Oct12-04, 07:34 AM
To Wong,

The angle between the vectors is always less than 90?
I can see it in 2D and 3D, but not in 4D, but how to write a proof?

Wong
Oct12-04, 10:13 AM
hi kakarukeys.

To prove it, first note that the set represented by a^{2}-b^{2}-b^{2}-c^{2}=1, a>1 is in fact a subset of A^2-B^2-C^2-D^2>0, A>0. Then what we need to prove becomes
Prove Aa+Bb+Cc+Dd>0, where (A, B, C, D) and (a, b, c, d) both satisfies,
h^2-i^2-j^2-k^2>0, h>0
A^2>B^2+C^2+D^2
A>(B^2+C^2+D^2)^{\frac{1}{2}}
Similarly,
a>(b^2+c^2+d^2)^{\frac{1}{2}}
Put those expression into Aa+Bb+Cc+Dd, does it remind you of some inequality?
This can be generalised to any dimensions.

Somtimes it can be quite difficult to think of a proof for inequalities, even though it is quite trivial geometrically.

kakarukeys
Oct14-04, 08:52 AM
Intuitive guide, I have spotted the 'subset' clue.
but I still can't see the solution.

Aa > \sqrt{B^2 + C^2 + D^2}\sqrt{b^2 + c^2 + d^2} \geq Bb + Cc + Dd
(Cauchy-Schwarz inequality)

And so Aa - Bb - Cc - Dd > 0
that's not we want.

Note Bb + Cc + Dd can be negative.

kakarukeys
Oct14-04, 08:58 AM
ok COOL
Aa > \sqrt{B^2 + C^2 + D^2}\sqrt{b^2 + c^2 + d^2} = \sqrt{(-B)^2 + (-C)^2 + (-D)^2}\sqrt{b^2 + c^2 + d^2}
\geq - Bb - Cc - Dd

kakarukeys
Oct14-04, 08:59 AM
Thumbs up!

TenaliRaman
Oct14-04, 10:09 AM
Interesting indeed!!!!
I was lost with my initial thoughts !!!!
Bravo!!

-- AI