WorkbenchDrag x1,x2,x3. The product on the left keeps changing, yet (Σuivi)2 does not move at all. Why? Then make the slack 0.
Loading figure…
What this move does
The Cauchy inequality (∑ui2)(∑vi2)≥(∑uivi)2 is not hard in itself. The hard part is seeing where the two lists are. The move:
Find two factors in the problem that can play the role of “sums of squares” (often two brackets multiplied together, or a sum you are given).
Split each term into ui2 and vi2 so that in the products uivithe variables cancel, leaving only constants or known quantities.
Apply Cauchy, then check whether the equality condition (all ratios ui:vi equal) can be met under the constraints.
On the workbench above, ui=pi/xi and vi=xi: their sums of squares are ∑xipi and ∑xi, while the product uivi=pi does not depend on xi. When the variables vanish from the products, the lower bound is a constant.
Predict first
a,b,c>0. What is the minimum of (a1+b9+c16)(a+b+c)?
Choose an answer first; then you can see the result.
✓You predicted it!✗Not what you expected — which is exactly why it is worth a pause.
Take the lists (a1,b3,c4) and (a,b,c). Their products are 1,3,4, so the lower bound is (1+3+4)2=64.
Equality needs a1/a=b3/b=c4/c, i.e. a:b:c=1:3:4, and that can happen, for example at (a,b,c)=(1,3,4). 26=1+9+16 comes from adding the numerators and forgetting the square roots; 36=(1+2+3)2 comes from misremembering the numerators as 1,4,9.
Work through one
Example: prove (1/a + 9/b + 16/c)(a + b + c) ≥ 64
Step 1 / 5
Loading figure…
Spot the two sums of squares
The left side is two brackets multiplied together, exactly the shape of the right side of Cauchy, ∑ui2⋅∑vi2. Try a=b=c=1 first: the left side is 26×3=78.
(a1+b9+c16)(a+b+c)
Split into u² and v²
The variables vanish from the products
Equality: every pair of bars has the same shape
The general pattern
Another split: let 1 be the second list
Not every problem has two brackets. When there is only one sum, the other list is often just 1,1,…,1.
Example: a + b + c = 1, a, b, c > 0; prove √(3a+1) + √(3b+1) + √(3c+1) ≤ 3√2
Step 1 / 3
Loading figure…
We want an upper bound for a sum of products
The left side is a sum of three square roots; read it as ∑uivi with ui=1 and vi=3ai+1. The picture uses some arbitrary vi.
1⋅3a+1+1⋅3b+1+1⋅3c+1
Both sums of squares are known
Equality: all the v have the same length
In the picture the vi can only be multiples of 0.5, so the third step uses 1.5 instead of 2. What matters is that they are all equal: the slack is 0.
Practice
Exercise 1Basic
a,b>0. Find the minimum of (a+b)(a1+b4).
Hint · direction
Two brackets multiplied together: exactly two sums of squares.
Hint · method
Take (a,b) and (a1,b2); the products are 1 and 2.
Hint · first step
The lower bound is (1+2)2.
Full solution
(a+b)(a1+b4)≥(1+2)2=9, with equality if and only if b=2a (for example a=1,b=2: 3×3=9). The minimum is 9.
Exercise 2Basic
Positive numbers x,y,z satisfy x+y+z=1. Find the maximum of (x+2y+3z)2.
Hint · direction
Inside the bracket is a “sum of products”, and what you know is x+y+z.
Hint · method
2y=2⋅y. Use the lists (x,y,z) and (1,2,3).
Hint · first step
(x+2y+3z)2≤(x+y+z)(1+2+3).
Full solution
So it is at most 6. Equality needs 1x=2y=3z, i.e. x:y:z=1:2:3, so (x,y,z)=(61,31,21). The maximum is 6.
Exercise 3Advanced
Let x,y,z>0. Prove that (x+y+z)(x+y1+y+z1+z+x1)≥29.
Hint · direction
The denominators in the second bracket are x+y, y+z and z+x. Can you write the first bracket in terms of them too?
Hint · method
x+y+z=21[(x+y)+(y+z)+(z+x)].
Hint · first step
Let s1=x+y, s2=y+z, s3=z+x. Cauchy gives (s1+s2+s3)(s11+s21+s31)≥9.
Full solution
The left side =21(s1+s2+s3)(∑si1)≥29. Equality holds if and only if s1=s2=s3, i.e. x=y=z.
Exercise 4Challenge
Let a,b,c>0 with abc=1. Prove that a1+b4+c9≥a+b+c36, and explain why this does not directly give a1+b4+c9≥12.
Hint · direction
The first part is exactly this stop's construction. For the second part: does a+b+c have an upper bound?
Hint · method
First part: the lists (a1,b2,c3) and (a,b,c). Second part: from abc=1 you only get a+b+c≥3.
Hint · first step
(a1+b4+c9)(a+b+c)≥(1+2+3)2=36; dividing both sides by a+b+c gives the first part.
Full solution
Second part: to get from a+b+c36 to 12 you would need a+b+c≤3, but abc=1 gives exactly the opposite, a+b+c≥3. The chain of bounds is broken here. (In fact, by AM–GM, a1+b4+c9≥33abc36=3336≈9.9, with equality possible when a1=b4=c9, so 12 is not a lower bound at all.)
Carrying on while the products still contain variables
✗
The wrong step: Given a+b+c=3, prove (a2+b2+c2)(a1+b1+c1)≥9: by Cauchy, the left side is ≥(a+b+c)2≥9.
What went wrong? Think first, then open
DiagnosisThe first step is fine, but the right side (a+b+c)2 still contains variables. And it is in fact at most3(a+b+c)=9 (Cauchy again), so the second step goes the wrong way.
✓
The right step: Here bound each factor on its own: a2+b2+c2≥3(a+b+c)2=3 and a1+b1+c1≥a+b+c9=3; multiplying gives ≥9.
“A good construction makes the products constant; if variables are left, check the direction first”
Exhibit
A lower bound that is never reached
✗
The wrong step: a≥2, b>0, a+b=4, so a1+b9≥a+b(1+3)2=4, hence the minimum is 4.
What went wrong? Think first, then open
DiagnosisEquality needs b=3a, i.e. a=1, but the problem requires a≥2. The inequality itself is true; the value 4 simply can never be reached.
✓
The right step: On [2,4) the function f(a)=a1+4−a9 is increasing, so its minimum is at a=2: 21+29=5.
“Before you write “the minimum is”, put the equality condition back into the constraints”
Proving Inequalities with Cauchy ← → step · F full screen · Esc exit