AMC 10 · 2010 · #24

Grade 11 algebra
sign-analysispolynomial-rootsvieta-formulasinterval-arithmetic identify-subproblemssymmetry-argument ↑ Prerequisites: sign-analysisvieta-formulas
📏 Long solution 💡 3 insights
Problem
A sum of three reciprocals reaches at least one on a union of intervals. Find the total length.

Pick an answer.

(A)
$\dfrac{1003}{335}$
(B)
$\dfrac{1004}{335}$
(C)
3
(D)
$\dfrac{403}{134}$
(E)
$\dfrac{202}{67}$
How to solve
Strategy Change Focus / Count the Complement

The endpoints of these intervals are roots of a cubic that does not factor, so anyone trying to find them one at a time is stuck. Tool #16 (Change Focus) is the whole idea: the question asks only for a total length, and the total length turns out to be the sum of those roots — and a sum of roots is exactly what Vieta's formulas hand over without ever solving anything. Tool #4 (Introduce a Variable) sets t=x-2010 so the poles become -1,0,1; sliding the picture sideways cannot change any length, and symmetric poles keep the algebra clean. Tool #1 (Draw a Diagram) supplies the shape of f: the poles cut the line into four pieces and on each piece f falls steadily, so each piece can meet the level y=1 at most once. Tool #7 (Identify Subproblems) then treats the four pieces separately, because they do not all behave the same way. Tool #14 (Extreme Principle) handles the piece the eye tends to skip — the far-left one, where every term is negative — and that omission is not cosmetic: the argument only works because the count of contributing pieces is exactly the degree of the cubic. Two things have to be nailed down rather than assumed. First, that each contributing piece supplies exactly one interval, closed at its right end, which is what makes the answer a clean sum. Second, that the three endpoints found are all the roots of the cubic — if even one root were complex, Vieta's sum would include it and would not equal the total length.

1STEP 1

Slide the poles to -1,0,1

Shifting puts the poles at three symmetric spots.

t=x-2010 ⟹ g(t)=1/(t+1)+1/t+1/(t-1) ≥ 1
2STEP 2

Four pieces, falling on each

The function falls across every piece.

g strictly decreasing on each of (-∞,-1), (-1,0), (0,1), (1,∞)
3STEP 3

The far-left piece gives nothing

The far-left piece stays negative throughout.

t < -1 ⟹ 1/(t+1) < 0, 1/t < 0, 1/(t-1) < 0 ⟹ g(t) < 0
4STEP 4

Each other piece gives one interval

Each remaining piece yields exactly one interval.

(-1,p], (0,q], (1,r] with g(p)=g(q)=g(r)=1
5STEP 5

Total length is p+q+r

The pole offsets cancel in the total length.

(p+1)+q+(r-1)=p+q+r
6STEP 6

Turn g(t)=1 into a cubic

The endpoints solve one cubic.

3t²-1=t³-t ⟺ t³-3t²-t+1=0
7STEP 7

These three are all the roots

Those three are all of its roots.

h(t)=t³-3t²-t+1=(t-p)(t-q)(t-r)
8STEP 8

Vieta finishes it

The sum of roots is 3, choice (A).

p+q+r=--3/1=3
Answer
3
Three checks. Numerically, the roots of t³-3t²-t+1 are about -0.6751, 0.4608, 3.2143, which do land one in each of (-1,0), (0,1), (1,4) as claimed, and they add to 3.000; the intervals are then about (-1,-0.675], (0,0.461], (1,3.214] with lengths 0.325+0.461+2.214=3. Sampling the original inequality densely on 2005 ≤ x ≤ 2020 measures the solution set at ≈ 2.99999, so the shift by 2010 hid nothing. Structurally, the wrong answers are the tell: 1003/335≈ 2.994, 1004/335≈ 2.997, 403/134≈ 3.007, 202/67≈ 3.015 all cluster just around 3, built from 335=2010/6, 134=2010/15, 67=2010/30 — they are manufactured to punish anyone who believes the year numbers survive into the answer, and Step 1 shows they cannot. One warning worth keeping: a popular shortcut cross-multiplies the inequality to get -t³+3t²+t-1 ≥ 0 and reads the intervals off that. It is not equivalent — at t=-2 that polynomial equals 17 ≥ 0, yet g(-2)=-1-1/2-1/3 < 1, so the polynomial inequality picks up a whole extra branch on the far left. It reaches the right total only because the sum of roots does not notice the error. The derivation above avoids it by clearing denominators only at the equality.
💡Key takeaway

The interval endpoints are ugly roots of a cubic, but their left ends cancel out, so the total length is just the sum of the roots — and a monic cubic tells you that sum straight from its t² coefficient.

  • Slide the poles to -1,0,1
  • Four pieces, falling on each
  • The far-left piece gives nothing
  • Each other piece gives one interval
  • Total length is p+q+r
  • Turn g(t)=1 into a cubic
  • These three are all the roots
  • Vieta finishes it