This problem is a masterclass in proof architecture. Rather than attacking ∣sin(nx)∣≤n∣sinx∣ directly with induction — which forces you to wrestle with trigonometric identities and induction logic simultaneously — the scaffold decouples the two: Part (i) isolates all the trigonometry into a single lemma, and Part (ii) turns induction into pure algebra. Parts (iii) and (iv) then show how one general result unlocks a family of corollaries.
Problem Statement
Let x1,x2,…,xn be real numbers, and let n∈N with n≥1.
i. Prove that for any two real numbers x1 and x2:
∣sin(x1+x2)∣≤∣sinx1∣+∣sinx2∣
ii. Use mathematical induction, or otherwise, to prove that for any n≥1:
sin(i=1∑nxi)≤i=1∑n∣sinxi∣
iii. Deduce that for any x∈R and n∈N:
∣sin(nx)∣≤n∣sinx∣
iv. Using the results above, prove that for any x∈R and n∈N:
1−cos(nx)≤n2(1−cosx)
Hints
Part (i): Use the angle addition formula for sine, apply the triangle inequality (∣a+b∣≤∣a∣+∣b∣), then bound the cosine factors using ∣cosθ∣≤1.
Part (ii): For the induction step (k⇒k+1), group the first k terms as a single angle X=x1+⋯+xk, and apply the lemma from Part (i) to (X+xk+1).
Part (iii): Set x1=x2=⋯=xn=x in the result from Part (ii).
Part (iv): Use the half-angle identity 1−cos(2θ)=2sin2θ. Write 1−cos(nx) in this form, then apply Part (iii) and square the resulting inequality.
Solutions
Part (i)
Using the angle addition formula and the triangle inequality:
Base case (n=1):∣sin(x1)∣≤∣sinx1∣ holds trivially. (Note: the case n=2 is exactly Part (i).)
Inductive step: Suppose the inequality holds for some k≥1, i.e.
sin(i=1∑kxi)≤i=1∑k∣sinxi∣
We must show it holds for k+1. Write:
sin(i=1∑k+1xi)=sin((i=1∑kxi)+xk+1)
Treat ∑i=1kxi as a single angle and apply the lemma from Part (i):
≤sin(i=1∑kxi)+∣sinxk+1∣
Applying the inductive hypothesis to the first term:
≤(i=1∑k∣sinxi∣)+∣sinxk+1∣=i=1∑k+1∣sinxi∣□
Note how the induction step contains no trigonometry at all — all the trig work was encapsulated in the lemma from Part (i).
Part (iii)
Set xi=x for all i∈{1,2,…,n} in Part (ii):
sin(i=1∑nx)≤i=1∑n∣sinx∣∣sin(nx)∣≤n∣sinx∣□
Part (iv)
Recall the half-angle identity 1−cos(2θ)=2sin2θ. Setting 2θ=nx, i.e. θ=2nx:
1−cos(nx)=2sin2(2nx)
From Part (iii) applied with x↦2x:
sin(n⋅2x)≤nsin(2x)
Since both sides are non-negative, squaring preserves the inequality:
sin2(2nx)≤n2sin2(2x)
Multiply both sides by 2:
2sin2(2nx)≤2n2sin2(2x)
Substituting the half-angle identity back on both sides:
1−cos(nx)≤n2(1−cosx)□
Takeaways
The Paradox of Generalisation: Proving the harder, more general statement (a sum of n different angles) is logically cleaner than proving the specific case directly. Generalising decouples the trigonometry from the induction logic, making both steps simpler.
Modularity in Proofs: Breaking a proof into lemmas creates reusable "black boxes." In Part (ii), the induction step is pure algebra — there are no sines or cosines in sight. The trigonometric heavy-lifting was done once in Part (i) and then simply invoked.
Translating Between Sine and Cosine: Inequalities bounding ∣sin∣ can often be translated into inequalities bounding cosine via the Pythagorean identity or half-angle formulas. Recognising that 1−cosθ=2sin2(θ/2) is the key bridge in Part (iv).