HOME/NOTES/CONCEPT

Relation Between AM, GM and HM Complete JEE Notes with Proofs

Ritesh Raj · 1 Aug 2026
Aa
14 min read16 sectionsConcept

Almost every "find the minimum value" question in JEE Main and Advanced is really one inequality wearing a disguise: AM ≥ GM. These notes build the three means from scratch, prove the chain AGHA \ge G \ge H, and then work through the six shapes the question actually takes in a paper — with past questions from IIT-JEE and JEE Main 2026.

The three means, side by side

Let aa and bb be two positive numbers. Each mean is the middle term of a different progression.

Arithmetic mean AAGeometric mean GGHarmonic mean HH
1. Middle term ofan APa GPan HP
2. Defining relationba=cbb-a=c-bba=cb\dfrac{b}{a}=\dfrac{c}{b}1b1a=1c1b\dfrac1b-\dfrac1a=\dfrac1c-\dfrac1b
3. Formula for two numbersA=a+b2A=\dfrac{a+b}{2}G=abG=\sqrt{ab}H=2aba+bH=\dfrac{2ab}{a+b}
4. For nn numbersAn=a1+a2++annA_n=\dfrac{a_1+a_2+\dots+a_n}{n}Gn=(a1a2an)1/nG_n=\left(a_1a_2\cdots a_n\right)^{1/n}Hn=n1a1+1a2++1anH_n=\dfrac{n}{\dfrac1{a_1}+\dfrac1{a_2}+\dots+\dfrac1{a_n}}
5. Built fromthe numbers themselvestheir producttheir reciprocals

Row 5 is the one worth internalising. The harmonic mean is nothing more than "take reciprocals, average them, take the reciprocal back" — so every HM question is an AM question in disguise, and you should reach for 1a\dfrac1a the moment you see the word harmonic.

One formula covers all three

For two positive numbers aba\ne b, the expression

an+1+bn+1an+bn\frac{a^{n+1}+b^{n+1}}{a^{n}+b^{n}}

equals AA when n=0n=0, equals GG when n=12n=-\dfrac12, and equals HH when n=1n=-1.

Why: setting it equal to a+b2\dfrac{a+b}{2} and clearing denominators gives (ab)(anbn)=0(a-b)\left(a^{n}-b^{n}\right)=0. Since aba\ne b we need an=bna^n=b^n, i.e. n=0n=0. The other two drop out the same way. JEE has asked all three versions of this over the years.

Inserting nn means between two numbers

nn arithmetic means

If a,A1,A2,,An,ba,A_1,A_2,\dots,A_n,b are in AP, the list has n+2n+2 terms and therefore n+1n+1 equal gaps:

d=ban+1,Ar=a+r(ban+1).d=\frac{b-a}{n+1},\qquad A_r=a+r\left(\frac{b-a}{n+1}\right).

The sum is the result JEE actually uses:

A1+A2++An=n(a+b2),A_1+A_2+\dots+A_n=n\left(\frac{a+b}{2}\right),

i.e. the sum of nn arithmetic means equals nn times the single arithmetic mean of aa and bb.

nn geometric means

If a,G1,G2,,Gn,ba,G_1,G_2,\dots,G_n,b are in GP, the n+1n+1 equal ratios give

r=(ba)1/(n+1),Gk=a(ba)k/(n+1).r=\left(\frac{b}{a}\right)^{1/(n+1)},\qquad G_k=a\left(\frac{b}{a}\right)^{k/(n+1)}.

And the product is the mirror of the sum result:

G1G2Gn=(ab)n.G_1\cdot G_2\cdots G_n=\left(\sqrt{ab}\right)^{n}.

nn harmonic means

Reciprocals turn HP into AP, so if a,H1,,Hn,ba,H_1,\dots,H_n,b are in HP then 1a,1H1,,1b\dfrac1a,\dfrac1{H_1},\dots,\dfrac1b are in AP with common difference abab(n+1)\dfrac{a-b}{ab(n+1)}, giving

1Hk=1a+k(abab(n+1)),\frac{1}{H_k}=\frac1a+k\left(\frac{a-b}{ab(n+1)}\right),

and, matching the pattern,

1H1+1H2++1Hn=n(a+b2ab)=nHM of a and b.\frac{1}{H_1}+\frac{1}{H_2}+\dots+\frac{1}{H_n}=n\left(\frac{a+b}{2ab}\right)=\frac{n}{\text{HM of }a\text{ and }b}.

Four arithmetic means and four geometric means inserted between 1 and 32
The pattern across all three

Sum of nn AMs =n×= n \times (single AM). Product of nn GMs =(single GM)n= (\text{single GM})^n. Sum of the reciprocals of nn HMs =n×= n \times (reciprocal of the single HM). Each mean answers to its own operation — addition, multiplication, addition-of-reciprocals.

The relation AGHA \ge G \ge H

This is the engine of the whole chapter. Both halves come from the same square.

Step 1: AGA \ge G.

AG=a+b2ab=a+b2ab2=(ab)22  0.A-G=\frac{a+b}{2}-\sqrt{ab}=\frac{a+b-2\sqrt{ab}}{2}=\frac{\left(\sqrt a-\sqrt b\right)^2}{2}\ \ge\ 0.

Step 2: GHG \ge H.

GH=ab2aba+b=ab(a+b2aba+b)=ab(ab)2a+b  0,G-H=\sqrt{ab}-\frac{2ab}{a+b}=\sqrt{ab}\left(\frac{a+b-2\sqrt{ab}}{a+b}\right)=\frac{\sqrt{ab}\left(\sqrt a-\sqrt b\right)^2}{a+b}\ \ge\ 0,

since a,b>0a,b>0 make both ab\sqrt{ab} and a+ba+b positive.

Step 3: Chain them. AGA\ge G and GHG\ge H give

A  G  H\boxed{A\ \ge\ G\ \ge\ H}

with equality throughout exactly when a=ba=b, because (ab)2=0\left(\sqrt a-\sqrt b\right)^2=0 forces a=ba=b.

The three means for a = 1 and b = 9 on a number line
The gap A minus G closes only when a equals b
The equality condition is half the marks

Every optimisation question using AM–GM has two parts: the bound, and the check that the bound is attained. A minimum of 2k2\sqrt{k} is only the answer if the variables can actually be made equal within the given domain. State where equality holds, every time.

The geometric picture

The inequality is a fact about a circle. Lay segments PR=aPR=a and RQ=bRQ=b end to end, so PQ=a+bPQ=a+b, and draw the circle on PQPQ as diameter. Its radius is a+b2\dfrac{a+b}{2}. The chord through RR perpendicular to PQPQ meets the circle at AA, and the intersecting-chords relation gives

AR2=PRRQ=ab  AR=ab.AR^2=PR\cdot RQ=ab\ \Longrightarrow\ AR=\sqrt{ab}.

Half a chord can never exceed the radius, so aba+b2\sqrt{ab}\le\dfrac{a+b}{2} — and the two are equal exactly when RR is the centre, i.e. when a=ba=b.

Semicircle proof that root ab is at most the average of a and b

The key results worth memorising

For two positive numbers aa and bb with means AA, GG, HH:

  1. AA, GG, HH are themselves in GP: G2=AHG^2=AH. (Check: AH=a+b22aba+b=ab=G2AH=\dfrac{a+b}{2}\cdot\dfrac{2ab}{a+b}=ab=G^2.)
  2. aa and bb are the roots of x22Ax+G2=0x^2-2Ax+G^2=0, since their sum is 2A2A and their product is G2G^2.
  3. Recovering the numbers from the means: a=AA2G2a=A-\sqrt{A^2-G^2} and b=A+A2G2b=A+\sqrt{A^2-G^2}.
  4. For three numbers with means AA, GG, HH, the cubic having them as roots is x33Ax2+3G3HxG3=0x^3-3Ax^2+\dfrac{3G^3}{H}x-G^3=0.

Result 3 is why AGA\ge G must hold: if A<GA<G the square root would be imaginary, and two positive reals cannot have imaginary values.

Standard inequalities that follow

Each of these is AM–GM applied once, and each shows up in papers in its own right.

#ResultWhere it comes from
(i)(a1++an)(1a1++1an)n2\left(a_1+\dots+a_n\right)\left(\dfrac1{a_1}+\dots+\dfrac1{a_n}\right)\ge n^2AM–GM on the aia_i, then on the 1ai\dfrac1{a_i}, multiplied
(ii)(a+b+c)(1a+1b+1c)9(a+b+c)\left(\dfrac1a+\dfrac1b+\dfrac1c\right)\ge 9result (i) with n=3n=3
(iii)1a+1b+1c  1ab+1bc+1ca\dfrac1a+\dfrac1b+\dfrac1c\ \ge\ \dfrac1{\sqrt{ab}}+\dfrac1{\sqrt{bc}}+\dfrac1{\sqrt{ca}}AM–GM on each pair of reciprocals
(iv)a1m++anmn  (a1++ann)m\dfrac{a_1^m+\dots+a_n^m}{n}\ \ge\ \left(\dfrac{a_1+\dots+a_n}{n}\right)^{m} for m<0m<0 or m>1m>1; the inequality reverses for 0<m<10<m<1power mean inequality
(v)m1a1++mnanm1++mn  (a1m1anmn)1m1++mn\dfrac{m_1a_1+\dots+m_na_n}{m_1+\dots+m_n}\ \ge\ \left(a_1^{m_1}\cdots a_n^{m_n}\right)^{\frac{1}{m_1+\dots+m_n}}weighted AM–GM

Result (v) is the one that unlocks "unequal coefficients" problems: split 2an2a_n into an+ana_n+a_n, or split 3a33a^{-3} into three copies of a3a^{-3}, and plain AM–GM applies again.

Now the six question types.

Type 1 — Inserting means and summing them

Example 1 (JEE Main 2026, 05 Apr Shift 2). Let A1,A2,,A39A_1,A_2,\dots,A_{39} be 3939 arithmetic means between 5959 and 159159. Find the mean of A25,A28,A31A_{25},A_{28},A_{31} and A36A_{36}.

SHOW SOLUTION

Step 1: Count the terms. The AP is 59,A1,,A39,15959,A_1,\dots,A_{39},159 — that is 4141 terms, so 4040 gaps.

Step 2: Common difference.

d=1595940=10040=2.5.d=\frac{159-59}{40}=\frac{100}{40}=2.5.

Step 3: Write the four means. Using Ar=59+rdA_r=59+rd:

A25=59+62.5=121.5,A28=59+70=129,A_{25}=59+62.5=121.5,\qquad A_{28}=59+70=129, A31=59+77.5=136.5,A36=59+90=149.A_{31}=59+77.5=136.5,\qquad A_{36}=59+90=149.

Step 4: Average them.

121.5+129+136.5+1494=5364=134\frac{121.5+129+136.5+149}{4}=\frac{536}{4}=\boxed{134}

Faster route. The four subscripts 25,28,31,3625,28,31,36 average to 3030, and ArA_r is linear in rr, so the mean of the four values is A30=59+30(2.5)=134A_{30}=59+30(2.5)=134 directly.

Example 2 (JEE Main 2026, 28 Jan Shift 2). Let the arithmetic mean of 1a\dfrac1a and 1b\dfrac1b be 516\dfrac{5}{16} with a>2a>2. If a,4,α,ba,4,\alpha,b are in AP, locate the roots of αx2ax+2(α2b)=0\alpha x^2-ax+2(\alpha-2b)=0.

SHOW SOLUTION

Step 1: Use the AP to name everything by one letter. With 44 as the second term and common difference dd:

a=4d,α=4+d,b=4+2d.a=4-d,\qquad \alpha=4+d,\qquad b=4+2d.

Step 2: Translate the mean condition.

12(14d+14+2d)=516  (4+2d)+(4d)(4d)(4+2d)=58.\frac12\left(\frac{1}{4-d}+\frac{1}{4+2d}\right)=\frac{5}{16}\ \Longrightarrow\ \frac{(4+2d)+(4-d)}{(4-d)(4+2d)}=\frac58.

Step 3: Clear and solve. With (4d)(4+2d)=16+4d2d2(4-d)(4+2d)=16+4d-2d^2,

8(8+d)=5(16+4d2d2)  10d212d16=0  5d26d8=0,8(8+d)=5\left(16+4d-2d^2\right)\ \Longrightarrow\ 10d^2-12d-16=0\ \Longrightarrow\ 5d^2-6d-8=0,

so (5d+4)(d2)=0(5d+4)(d-2)=0 and d=45d=-\dfrac45 or d=2d=2.

Step 4: Pick the branch. d=2d=2 gives a=2a=2, α=6\alpha=6, b=8b=8. (The other branch d=45d=-\tfrac45 gives a=4.8a=4.8, and both branches land the roots in the same two intervals, so the answer is unaffected.)

Step 5: Build and factorise the quadratic.

6x22x+2(616)=6x22x20=0  3x2x10=0  (3x+5)(x2)=0.6x^2-2x+2(6-16)=6x^2-2x-20=0\ \Longrightarrow\ 3x^2-x-10=0\ \Longrightarrow\ (3x+5)(x-2)=0.

Step 6: Read off. x=2x=2 and x=531.67x=-\dfrac53\approx-1.67.

Answer: one root in (1,4)(1,4) and the other in (2,0)(-2,0).

Type 2 — Minimum of a sum when the product is fixed

This is the single most common use of AM–GM. If the product of the terms is a constant, their sum has a minimum, and the minimum is nn times the nnth root of that product.

Example 3 (IIT-JEE 1991). The product of nn positive numbers is unity. What can be said about their sum?

SHOW SOLUTION

Step 1: Name the numbers. Let x1,x2,,xn>0x_1,x_2,\dots,x_n>0 with x1x2xn=1x_1x_2\cdots x_n=1.

Step 2: Apply AM–GM to all nn of them.

x1+x2++xnn  (x1x2xn)1/n=11/n=1.\frac{x_1+x_2+\dots+x_n}{n}\ \ge\ \left(x_1x_2\cdots x_n\right)^{1/n}=1^{1/n}=1.

Step 3: Multiply out.

x1+x2++xn  n.x_1+x_2+\dots+x_n\ \ge\ n.

Step 4: Equality. All xix_i equal, i.e. every xi=1x_i=1, which does satisfy the product condition — so the bound is attained.

Answer: the sum is never less than nn.

Example 4 (IIT-JEE 2002). If a1,a2,,ana_1,a_2,\dots,a_n are positive reals with product a fixed number cc, find the minimum value of a1+a2++an1+2ana_1+a_2+\dots+a_{n-1}+2a_n.

SHOW SOLUTION

Step 1: Make the product match the sum. The sum contains 2an2a_n, not ana_n, so treat the nn quantities as

a1, a2, , an1, 2an,a_1,\ a_2,\ \dots,\ a_{n-1},\ 2a_n,

whose product is 2(a1a2an)=2c2\left(a_1a_2\cdots a_n\right)=2c.

Step 2: Apply AM–GM to those nn quantities.

a1+a2++an1+2ann  (2c)1/n.\frac{a_1+a_2+\dots+a_{n-1}+2a_n}{n}\ \ge\ \left(2c\right)^{1/n}.

Step 3: Multiply out.

a1+a2++an1+2an  n(2c)1/n.a_1+a_2+\dots+a_{n-1}+2a_n\ \ge\ n\,(2c)^{1/n}.

Step 4: Equality. All nn quantities equal, i.e. a1==an1=2ana_1=\dots=a_{n-1}=2a_n, which is achievable.

Answer: n(2c)1/n\boxed{n\,(2c)^{1/n}}

The trick in Example 4 is the whole technique

AM–GM only bounds a sum by the nnth root of the product of the very terms being added. So when a coefficient appears, absorb it: read 2an2a_n as one quantity, or split 3a33a^{-3} into a3+a3+a3a^{-3}+a^{-3}+a^{-3}. Getting the grouping right is the entire difficulty in these questions.

Example 5 (IIT-JEE 2011). Find the minimum value of a5+a4+3a3+1+a8+a10a^{-5}+a^{-4}+3a^{-3}+1+a^{8}+a^{10} for a>0a>0.

SHOW SOLUTION

Step 1: Split the coefficient. Write 3a33a^{-3} as three separate terms. The expression becomes a sum of eight quantities:

a5, a4, a3, a3, a3, 1, a8, a10.a^{-5},\ a^{-4},\ a^{-3},\ a^{-3},\ a^{-3},\ 1,\ a^{8},\ a^{10}.

Step 2: Check their product. Add the exponents:

54333+0+8+10=0,-5-4-3-3-3+0+8+10=0,

so the product is a0=1a^0=1. This is why the question was built with these particular powers.

Step 3: Apply AM–GM to the eight quantities.

a5+a4+3a3+1+a8+a108  (1)1/8=1.\frac{a^{-5}+a^{-4}+3a^{-3}+1+a^{8}+a^{10}}{8}\ \ge\ \left(1\right)^{1/8}=1.

Step 4: Multiply out.

a5+a4+3a3+1+a8+a10  8a^{-5}+a^{-4}+3a^{-3}+1+a^{8}+a^{10}\ \ge\ \boxed{8}

Step 5: Equality. All eight equal forces a=1a=1, and substituting a=1a=1 gives 1+1+3+1+1+1=81+1+3+1+1+1=8. The bound is attained, so 88 really is the minimum.

Example 6 (JEE Main 2026, 05 Apr Shift 2). Let (21a+21+a)\left(2^{1-a}+2^{1+a}\right), f(a)f(a), (3a+3a)\left(3^{a}+3^{-a}\right) be in AP and let α\alpha be the minimum value of f(a)f(a). Evaluate

loge(α1)logeαdxe2xe2x.\int_{\log_e(\alpha-1)}^{\log_e \alpha}\frac{dx}{e^{2x}-e^{-2x}}.

SHOW SOLUTION

Step 1: The middle term of an AP is the average of its neighbours.

f(a)=12[(21a+21+a)+(3a+3a)].f(a)=\frac12\Big[\left(2^{1-a}+2^{1+a}\right)+\left(3^{a}+3^{-a}\right)\Big].

Step 2: Minimise each bracket by AM–GM. Each is a sum of two positives with constant product:

21a+21+a  221a21+a=222=4,2^{1-a}+2^{1+a}\ \ge\ 2\sqrt{2^{1-a}\cdot 2^{1+a}}=2\sqrt{2^2}=4, 3a+3a  23a3a=2.3^{a}+3^{-a}\ \ge\ 2\sqrt{3^{a}\cdot 3^{-a}}=2.

Both equalities need a=0a=0, the same value of aa — so both minima are reached together.

Step 3: Minimum of ff.

α=f(0)=12(4+2)=3.\alpha=f(0)=\frac12(4+2)=3.

Step 4: Set up the integral. The limits are loge2\log_e 2 and loge3\log_e 3. Substitute t=e2xt=e^{2x}, so dx=dt2tdx=\dfrac{dt}{2t} and e2xe2x=t21te^{2x}-e^{-2x}=\dfrac{t^2-1}{t}; the limits become t=4t=4 and t=9t=9:

I=49tt21dt2t=49dt2(t21).I=\int_4^9\frac{t}{t^2-1}\cdot\frac{dt}{2t}=\int_4^9\frac{dt}{2\left(t^2-1\right)}.

Step 5: Partial fractions. Using 1t21=12(1t11t+1)\dfrac{1}{t^2-1}=\dfrac12\left(\dfrac{1}{t-1}-\dfrac{1}{t+1}\right),

I=14[loget1t+1]49=14(loge810loge35).I=\frac14\left[\log_e\frac{t-1}{t+1}\right]_4^9=\frac14\left(\log_e\frac{8}{10}-\log_e\frac{3}{5}\right).

Step 6: Simplify.

I=14loge(81053)=14loge(43)I=\frac14\log_e\left(\frac{8}{10}\cdot\frac{5}{3}\right)=\frac14\log_e\left(\frac43\right)

Answer: 14loge43\boxed{\dfrac14\log_e\dfrac43}

Type 3 — Maximum of a product when the sum is fixed

The mirror image. If the sum is fixed, the product has a maximum, reached again at equality.

Fixed sum: the product peaks when the two numbers are equal

Example 7 (IIT-JEE 2000). If a,b,c,da,b,c,d are positive reals with a+b+c+d=2a+b+c+d=2 and m=(a+b)(c+d)m=(a+b)(c+d), find the range of mm.

SHOW SOLUTION

Step 1: Group into two blocks. Treat (a+b)(a+b) and (c+d)(c+d) as two positive numbers. Their sum is 22.

Step 2: Apply AM–GM to the two blocks.

(a+b)+(c+d)2  (a+b)(c+d)  22  m  m1.\frac{(a+b)+(c+d)}{2}\ \ge\ \sqrt{(a+b)(c+d)}\ \Longrightarrow\ \frac22\ \ge\ \sqrt m\ \Longrightarrow\ m\le 1.

Step 3: Get the lower bound. All four numbers are positive, so a+b>0a+b>0 and c+d>0c+d>0, giving m>0m>0. It cannot reach 00.

Step 4: Combine.

0<m1\boxed{0<m\le 1}

Step 5: Equality. m=1m=1 needs a+b=c+d=1a+b=c+d=1, which is easily arranged, so the upper bound is attained.

Type 4 — Two terms whose product is constant

Whenever an expression has the shape u+kuu+\dfrac{k}{u} with u>0u>0 and kk constant, AM–GM finishes it in one line: the minimum is 2k2\sqrt k, at u=ku=\sqrt k.

y = x + 1/x has minimum 2 at x = 1 <!-- upload fig05-min-x-plus-1-over-x.png -->

Example 8 (IIT-JEE 2003). If α(0,π2)\alpha\in\left(0,\dfrac{\pi}{2}\right), show that

x2+x+tan2αx2+x\sqrt{x^2+x}+\frac{\tan^2\alpha}{\sqrt{x^2+x}}

is always at least 2tanα2\tan\alpha.

SHOW SOLUTION

Step 1: Check both terms are positive. Wherever the expression is defined we need x2+x>0x^2+x>0, so x2+x>0\sqrt{x^2+x}>0. And α(0,π2)\alpha\in\left(0,\dfrac\pi2\right) makes tanα>0\tan\alpha>0, so tan2α>0\tan^2\alpha>0.

Step 2: Name the pieces. Put u=x2+xu=\sqrt{x^2+x} and v=tan2αx2+xv=\dfrac{\tan^2\alpha}{\sqrt{x^2+x}}. Their product is constant:

uv=tan2α.uv=\tan^2\alpha.

Step 3: Apply AM–GM to uu and vv.

u+v2  uv=tan2α=tanα\frac{u+v}{2}\ \ge\ \sqrt{uv}=\sqrt{\tan^2\alpha}=\tan\alpha

(the positive square root, since tanα>0\tan\alpha>0 on this interval).

Step 4: Multiply out.

x2+x+tan2αx2+x  2tanα\sqrt{x^2+x}+\frac{\tan^2\alpha}{\sqrt{x^2+x}}\ \ge\ \boxed{2\tan\alpha}

Step 5: Equality. u=vu=v means x2+x=tan2αx^2+x=\tan^2\alpha, which has a solution for every α\alpha in the given range, so the bound is sharp. \blacksquare

Example 9 (IIT-JEE 2004). For positive reals a,b,ca,b,c, prove that

{(1+a)(1+b)(1+c)}7>77a4b4c4.\left\{(1+a)(1+b)(1+c)\right\}^{7}>7^{7}a^{4}b^{4}c^{4}.

SHOW SOLUTION

Step 1: Expand the product.

(1+a)(1+b)(1+c)=1+a+b+c+ab+bc+ca+abc.(1+a)(1+b)(1+c)=1+a+b+c+ab+bc+ca+abc.

Step 2: Count the terms. The right-hand side is a sum of seven positive quantities after the leading 11:

a, b, c, ab, bc, ca, abc.a,\ b,\ c,\ ab,\ bc,\ ca,\ abc.

Apply AM–GM to those seven, and their product is

abcabbccaabc=a4b4c4.a\cdot b\cdot c\cdot ab\cdot bc\cdot ca\cdot abc=a^{4}b^{4}c^{4}.

Step 3: Write the inequality.

a+b+c+ab+bc+ca+abc7  (a4b4c4)1/7.\frac{a+b+c+ab+bc+ca+abc}{7}\ \ge\ \left(a^{4}b^{4}c^{4}\right)^{1/7}.

Step 4: Add the leading 11. Since 1>01>0,

(1+a)(1+b)(1+c)=1+(a+b+c+ab+bc+ca+abc)>7(a4b4c4)1/7.(1+a)(1+b)(1+c)=1+\big(a+b+c+ab+bc+ca+abc\big)>7\left(a^{4}b^{4}c^{4}\right)^{1/7}.

The inequality is now strict because of that extra 11, regardless of whether the seven terms are equal.

Step 5: Raise to the seventh power. Both sides are positive, and tt7t\mapsto t^7 is increasing, so

{(1+a)(1+b)(1+c)}7>77a4b4c4.\left\{(1+a)(1+b)(1+c)\right\}^{7}>7^{7}a^{4}b^{4}c^{4}. \qquad \blacksquare

Type 5 — Questions mixing AP, GP and HP

Example 10 (IIT-JEE 1997). Let xx be the arithmetic mean and y,zy,z be two geometric means between two positive numbers. Find y3+z3xyz\dfrac{y^3+z^3}{xyz}.

SHOW SOLUTION

Step 1: Set up the GP. Let the numbers be aa and bb. Since a,y,z,ba,y,z,b are in GP with ratio rr:

y=ar,z=ar2,b=ar3.y=ar,\qquad z=ar^2,\qquad b=ar^3.

Step 2: Express aa and bb through yy and zz. From ay=yz=zb\dfrac ay=\dfrac yz=\dfrac zb we get

a=y2z,b=z2y.a=\frac{y^2}{z},\qquad b=\frac{z^2}{y}.

Step 3: Use the AM condition.

x=a+b2=12(y2z+z2y)=y3+z32yz.x=\frac{a+b}{2}=\frac12\left(\frac{y^2}{z}+\frac{z^2}{y}\right)=\frac{y^3+z^3}{2yz}.

Step 4: Rearrange.

2xyz=y3+z3  y3+z3xyz=22xyz=y^3+z^3\ \Longrightarrow\ \frac{y^3+z^3}{xyz}=\boxed{2}

Example 11 (IIT-JEE 2002). Let a,ba,b be positive reals. If a,A1,A2,ba,A_1,A_2,b are in AP; a,G1,G2,ba,G_1,G_2,b are in GP; and a,H1,H2,ba,H_1,H_2,b are in HP, show that

G1G2H1H2=A1+A2H1+H2=(2a+b)(a+2b)9ab.\frac{G_1G_2}{H_1H_2}=\frac{A_1+A_2}{H_1+H_2}=\frac{(2a+b)(a+2b)}{9ab}.

SHOW SOLUTION

Step 1: The two easy products. For the AP, the two means are symmetric about the centre, so

A1+A2=a+b.A_1+A_2=a+b.

For the GP, the two means multiply to give

G1G2=ab.G_1G_2=ab.

Step 2: The harmonic means. Reciprocals of a,H1,H2,ba,H_1,H_2,b form an AP with common difference 13(1b1a)\dfrac{1}{3}\left(\dfrac1b-\dfrac1a\right), which after simplifying gives

H1=3ab2b+a,H2=3abb+2a.H_1=\frac{3ab}{2b+a},\qquad H_2=\frac{3ab}{b+2a}.

Step 3: Use the AP of reciprocals. Since 1H1\dfrac1{H_1} and 1H2\dfrac1{H_2} are symmetric about the centre of that AP,

1H1+1H2=1a+1b  H1+H2H1H2=a+bab=A1+A2G1G2.\frac{1}{H_1}+\frac{1}{H_2}=\frac1a+\frac1b\ \Longrightarrow\ \frac{H_1+H_2}{H_1H_2}=\frac{a+b}{ab}=\frac{A_1+A_2}{G_1G_2}.

Cross-multiplying gives the first equality:

G1G2H1H2=A1+A2H1+H2.\frac{G_1G_2}{H_1H_2}=\frac{A_1+A_2}{H_1+H_2}.

Step 4: Compute the common value.

G1G2H1H2=ab3ab2b+a3abb+2a=ab(2b+a)(b+2a)9a2b2=(2a+b)(a+2b)9ab\frac{G_1G_2}{H_1H_2}=\frac{ab}{\dfrac{3ab}{2b+a}\cdot\dfrac{3ab}{b+2a}}=\frac{ab(2b+a)(b+2a)}{9a^2b^2}=\boxed{\frac{(2a+b)(a+2b)}{9ab}}\qquad\blacksquare

Example 12 (IIT-JEE 1992). The harmonic mean and geometric mean of two positive numbers are in the ratio 4:54:5. Find the ratio of the numbers.

SHOW SOLUTION

Step 1: Name the numbers with one unknown. Let them be kaka and aa with a>0a>0, so the required ratio is k:1k:1.

Step 2: Write both means.

G=kaa=ak,H=2(ka)(a)ka+a=2kak+1.G=\sqrt{ka\cdot a}=a\sqrt k,\qquad H=\frac{2(ka)(a)}{ka+a}=\frac{2ka}{k+1}.

Step 3: Impose the ratio.

HG=2ka(k+1)ak=2kk+1=45.\frac HG=\frac{2ka}{(k+1)\,a\sqrt k}=\frac{2\sqrt k}{k+1}=\frac45.

Step 4: Solve for k\sqrt k.

10k=4k+4  2k5k+2=0  k=5±34=2 or 12,10\sqrt k=4k+4\ \Longrightarrow\ 2k-5\sqrt k+2=0\ \Longrightarrow\ \sqrt k=\frac{5\pm 3}{4}=2\ \text{or}\ \frac12,

so k=4k=4 or k=14k=\dfrac14.

Step 5: Read the ratio. Both give the same pair of numbers in the two possible orders.

Answer: the numbers are in the ratio 4:1\boxed{4:1}

Type 6 — Proving the standard inequalities

Example 13. Prove that (a+b+c)(1a+1b+1c)9(a+b+c)\left(\dfrac1a+\dfrac1b+\dfrac1c\right)\ge 9 for positive a,b,ca,b,c.

SHOW SOLUTION

Step 1: AM–GM on the three numbers.

a+b+c3  (abc)1/3.\frac{a+b+c}{3}\ \ge\ (abc)^{1/3}.

Step 2: AM–GM on the three reciprocals.

1a+1b+1c3  (1abc)1/3.\frac{\dfrac1a+\dfrac1b+\dfrac1c}{3}\ \ge\ \left(\frac{1}{abc}\right)^{1/3}.

Step 3: Multiply the two. Both sides of both inequalities are positive, so multiplying is legal:

(a+b+c)3(1a+1b+1c)3  (abc)1/3(1abc)1/3=1.\frac{(a+b+c)}{3}\cdot\frac{\left(\dfrac1a+\dfrac1b+\dfrac1c\right)}{3}\ \ge\ (abc)^{1/3}\cdot\left(\frac{1}{abc}\right)^{1/3}=1.

Step 4: Clear the nines.

(a+b+c)(1a+1b+1c)  9(a+b+c)\left(\frac1a+\frac1b+\frac1c\right)\ \ge\ \boxed{9}

Step 5: Equality. Both AM–GM steps need all three equal, so equality holds exactly when a=b=ca=b=c. \blacksquare

Check. At a=b=c=1a=b=c=1 the left side is 3×3=93\times3=9. At a=1,b=2,c=3a=1,b=2,c=3 it is 6×116=11>96\times\dfrac{11}{6}=11>9.

The same argument with nn numbers instead of three gives result (i) from the table: (a1++an)(1a1++1an)n2\left(a_1+\dots+a_n\right)\left(\dfrac1{a_1}+\dots+\dfrac1{a_n}\right)\ge n^2.

Quick recall

SituationWhat to reach for
Product fixed, minimise a sumAM–GM: minimum is n(product)1/nn\left(\text{product}\right)^{1/n}
Sum fixed, maximise a productAM–GM the other way: maximum at all terms equal
Expression looks like u+kuu+\dfrac{k}{u}Minimum 2k2\sqrt k at u=ku=\sqrt k
A coefficient spoils the groupingSplit it: 2anan+an2a_n\to a_n+a_n, or use weighted AM–GM
The word harmonic appearsTake reciprocals immediately; HP becomes AP
nn means inserted between aa and bbSum of AMs =na+b2=n\dfrac{a+b}{2}; product of GMs =(ab)n=\left(\sqrt{ab}\right)^n; sum of reciprocals of HMs =n(a+b)2ab=\dfrac{n(a+b)}{2ab}
Given AA and GG, recover the numbersRoots of x22Ax+G2=0x^2-2Ax+G^2=0
Asked to relate AA, GG, HHG2=AHG^2=AH, and AGHA\ge G\ge H with equality only when all numbers are equal
A stray +1+1 or a leftover positive termThe inequality becomes strict — say so
Before you move on

Two habits carry almost all the marks in this chapter. First, match the grouping to the product — AM–GM bounds a sum only by the terms actually being added, so absorb every coefficient before you apply it. Second, always state where equality holds; a bound that cannot be reached is not a minimum, and examiners build options around exactly that gap.


Practise these on doMath, and read the JEE Main 2026 shift-wise analyses to see how often Sequences & Series and inequality questions actually appeared this year.

[ SHARE ] WhatsApp
[ practice these chapters ]
Reinforce this post with live problems on doMath.
Application of Derivatives
PreviousMonotonic Functions Increasing, Decreasing, Concavity and Inflection
← Back to all notes