WorkbenchStart with the first rewrite and drag a,b,c. Does the bound stay safely to the right of 3/2? Then switch to the other rewrites and compare.
Loading figure…
What this move does
The fraction form of the Cauchy inequality, ∑yixi2≥∑yi(∑xi)2 (yi>0), is made for “a row of fractions added together”. When a numerator is not a square, first make it one:
Multiply the top and bottom of each term by a positive expression so that every term has the form (⋯)(⋯)2. Choose the multiplier so that the merged numerator and denominator are easy to handle (ideally symmetric expressions or known quantities).
Merge once: ∑yixi2≥∑yi(∑xi)2.
What remains is a polynomial inequality. Check that it holds, and that its equality case agrees with the equality condition of step 2.
Step 1 is the crux. The same fraction can be “squared up” in many ways, and each gives a different lower bound after merging: some stay above the target everywhere, others drop below it.
Predict first
We want to prove b+ca+c+ab+a+bc≥23 (a,b,c>0). If we write each term as b+c(a)2 and merge, does the proof go through?
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.
Merging gives 2(a+b+c)(a+b+c)2. By Cauchy the numerator is ≤3(a+b+c), so this lower bound is ≤23, with equality only when a=b=c. It is below the target almost everywhere and proves nothing. (The option “when a,b,c are all equal” is not right either: then the expression simply equals 23, and there is nothing to prove.)
On the workbench above, use the first rewrite and drag the sliders: the “bound” dot almost always lands to the left of 3/2.
Work through one
This is the famous Nesbitt inequality (1903).
Example: a, b, c > 0; prove a/(b+c) + b/(c+a) + c/(a+b) ≥ 3/2
Step 1 / 5
Loading figure…
The laziest rewrite fails
The numerator is a=(a)2; merge directly. At a,b,c=1,2,3 the lower bound is about 1.43, already below 1.5.
S≥2(a+b+c)(a+b+c)2≤23
Multiply top and bottom by a
The polynomial inequality that is left
The two equality conditions agree
Higher powers are not always better
When it fails
Break it
Counterexample lab: remove one condition
Conclusion: the merged lower bound is at least 23 for all positive a,b,c← still true?
This route works everywhere: the reading “Lowest bound over all a, b, c” is exactly 1.5, reached at a=b=c.
Loading figure…
Switch to multiplying by a5. Near a=b=1, make c a little larger (say 1.8), and the lower bound falls below 1.5.
Loading figure…
The merge itself is not wrong; the problem is that it throws away too much: equality in the merge needs the ratios a5(b+c)a3 to be equal, and away from a=b=c that is very far from the truth.
Practice
Exercise 1Basic
Positive numbers a,b,c satisfy a+b+c=3. Find the minimum of a+11+b+11+c+11.
The wrong step: To prove ∑b+ca≥23: ∑b+c(a)2≥2(a+b+c)(a+b+c)2≥23.
What went wrong? Think first, then open
DiagnosisThe first step is right, the second is wrong: by Cauchy, (a+b+c)2≤3(a+b+c), so this lower bound is at most23 and only reaches it when a=b=c. The merge threw away too much.
✓
The right step: Use b+ca=a(b+c)a2 instead. Merging gives 2(ab+bc+ca)(a+b+c)2, and (a+b+c)2≥3(ab+bc+ca) always holds.
“Before merging, ask whether the step that remains will still be true”
Cauchy and Fractional Inequalities ← → step · F full screen · Esc exit