You are on page 1of 50

# Using the Arithmetic Mean-Geometric Mean Inequality in Problem Solving.

Using the Arithmetic Mean-Geometric Mean Inequality in Problem Solving
by JIM WILSON
A Presentation to the Annual Meeting of School Mathematics and Science Association, Birmingham, November 8, 2012 , was prepared using some parts of this paper.

PDF version Arithmetic Mean-Geometric Mean Inequality (AM-GM Inquality) is a fundamental relationship in mathematics.     It is a useful tool for problems solving and building relationships with other mathematics.     It should find more use in school mathematics than currently.   It what follows I present an introduction to the theorem,  some background and generalizations,   alternative demonstrations of the proof, and examples of problems that can be explored by using the AM-GM Inquality.   I will rely heavily on a collection of problems and essays on my web site: Http://jwilson.coe.uga.edu and the sub-directory for my mathematics problem solving course at the University of Georgia.
http://jwilson.coe.uga.edu/EMT725/AMGM/SSMA.Bnghm.html[11/9/12 3:34:14 PM]

The

Using the Arithmetic Mean-Geometric Mean Inequality in Problem Solving.

I will concentrate on the theorem for two positive numbers in my examples, but I mention the generalizations below and occasionally use the case for three positive numbers.

Arithmetic Mean - Geometric Mean Inequality
For real positive numbers a and b, the AM-GM Inequality for two numbers is:

http://jwilson.coe.uga.edu/EMT725/AMGM/SSMA.Bnghm.html[11/9/12 3:34:14 PM]

Using the Arithmetic Mean-Geometric Mean Inequality in Problem Solving.

Clearly, the AM-GM Inequality can be generalized for n positive numbers.  The link poses the problem of generalizing the proof following the lines of argument advanced by Courant and Robbins (1942).   That is,

http://jwilson.coe.uga.edu/EMT725/AMGM/SSMA.Bnghm.html[11/9/12 3:34:14 PM]

Using the Arithmetic Mean-Geometric Mean Inequality in Problem Solving.

The AM-GM Inequality can also be generalized to its inclusion in relation to other means such as the Harmonic Mean (HM)  or the Root Mean Square (RMS -- sometimes called the Quadratic Mean).  In particular, for two positive numbers a  and b,

http://jwilson.coe.uga.edu/EMT725/AMGM/SSMA.Bnghm.html[11/9/12 3:34:14 PM]

Using the Arithmetic Mean-Geometric Mean Inequality in Problem Solving.

Geometric demonstration of the RMS-AM-GM-HM Inequality   At a more advanced level  (perhaps more fundamental?)   all of these means are instances of  Power Means where the power parameter  p  takes on different values for the different means.   These may also be called Generalized Means.

What is the Value of this Theorem? An Example.
The AM-GM for two positive numbers can be a useful tool in examining some optimization problems.    For example, it is well known that for rectangles with a fixed perimeter, the maximum area is given by a square having that perimeter.

http://jwilson.coe.uga.edu/EMT725/AMGM/SSMA.Bnghm.html[11/9/12 3:34:14 PM]

The fixed perimeter is  2a + 2b = 10.Bnghm.   Perhaps some insight is given by the graph at the right. The red graph represents the area function of the graph as either of   a   or   b varies from  0 to 5.edu/EMT725/AMGM/SSMA. http://jwilson.     Let  b  represent the length of one side and  5 .b the length of the other.html[11/9/12 3:34:14 PM] .Using the Arithmetic Mean-Geometric Mean Inequality in Problem Solving.uga.coe.

The area is always less than the constant  AND it is equal to that constant when a =  b. Area = 6.edu/EMT725/AMGM/SSMA.   The utility of the AM-GM Inequality is that the replacement function after the application of the AM-GM inequality is a constant line tangent to the previous function. let the Perimeter of the rectangle equal 10 and let one side be  x.          From my web page.   a = b = 2.    In our example for P  = 10.   The the other side is 5 - x.html[11/9/12 3:34:14 PM] .b)  for 0 <  b < 5 When the AM-GM Inequality is applied we get  Area ≤ Points on the blue curve.25) and the blue curve (a horizontal line representing a constant) is tangent to the red curve if and only if  b = (5 .25 are always greater than points on the red curve (That is.b).Using the Arithmetic Mean-Geometric Mean Inequality in Problem Solving. b =2.e. i..5 when the rectangle is a square.uga.Bnghm.5. Setting this up with the usual function notation.coe. the area of the rectangle is always less that 6.            Area =  ab = b(5 . here is a problem posed for students: http://jwilson.

coe.Using the Arithmetic Mean-Geometric Mean Inequality in Problem Solving.html[11/9/12 3:34:14 PM] .  The second problem.uga.edu/EMT725/AMGM/SSMA. where ab is a constant rather than  a + b. is a nice contrast to the first and it also follows quickly from the AM-GM Inequality: http://jwilson.Bnghm.

Seeing multiple approaches to this relation can help with understanding it. I will limit the exploration to the simplest case:    The arithmetic mean and geometric mean of two positive numbers. seeing its importance.Bnghm.coe.         Some algebra http://jwilson.    Alternative Proofs and Demonstrations of the Arithmetic Mean-Geometric Mean Inequality In this section.Using the Arithmetic Mean-Geometric Mean Inequality in Problem Solving. and finding it useful as a problem solving tool.html[11/9/12 3:34:14 PM] .    In my Problem Solving course I pose AS AN EXPLORATION that the student find a least 5 demonstrations or proofs.edu/EMT725/AMGM/SSMA.uga.

edu/EMT725/AMGM/SSMA.uga.Using the Arithmetic Mean-Geometric Mean Inequality in Problem Solving.coe.html[11/9/12 3:34:14 PM] . A Geometric Demonstration http://jwilson.Bnghm.

http://jwilson.  The radius will be the Arithmetic Mean of  a and  b.uga.  This segment will always have a length less than or equal to the radius of the circle and it will be equal only if a = b.edu/EMT725/AMGM/SSMA.Bnghm.   From the intersection of this perpendicular with the semicircle construct the red segment. Construct a perpendicular to the diameter from common endpoint of the segments of length  a   and  b. Another Geometric Suggestion Construct a semicircle with a diameter a + b.html[11/9/12 3:34:14 PM] .Using the Arithmetic Mean-Geometric Mean Inequality in Problem Solving.coe.

Another Geometry Example          http://jwilson. This example is closely related to the well known geometric theorem that the altitude of a right triangle from the 90 degree vertex to the hypothenuse will be the geometric mean of the two segments cut off on the hypotenuse.Bnghm.Using the Arithmetic Mean-Geometric Mean Inequality in Problem Solving.coe.html[11/9/12 3:34:14 PM] .edu/EMT725/AMGM/SSMA.uga.

Find the length indicatec by   ?   along the common tanget in terms of   a   and  b.      we have             with equality if and only if   a Solution:   by constructing a segment parallel to the one under study.   Construct a common external tangent to the two circl3s and draw radii of each circle to the common tangent. Let AM = MB = z and let CM = x.edu/EMT725/AMGM/SSMA. a right triangle with legs  of length  ?  and  a-b.Bnghm. Another  Algebra Demonstration This demonstration begins with the identity:    Since = b.coe. http://jwilson.Using the Arithmetic Mean-Geometric Mean Inequality in Problem Solving. MD = y. Let the CD pass through the midpoint M of AB. and hypotenuse of length  a+b. Another Geometry Example Given two tangent circles of radii   a   and   b. The lengths are all positive values.uga. the segment along the common tangent has length twice the geometric mean of  a and b. More Geometry Consider a circle with chords AB and CD.html[11/9/12 3:34:14 PM] .

By an elementary theorem of geometry the products of the parts of the chords are equal.uga. Algebra Something Different http://jwilson.coe.html[11/9/12 3:34:14 PM] .Using the Arithmetic Mean-Geometric Mean Inequality in Problem Solving. We know that  x + y ≥2z with equality only if M is the midpoint of CD.edu/EMT725/AMGM/SSMA.Bnghm.

How much of each type of fence will he have to buy in order to keep expenses to a minimum?  What is the minimum expense? Solution:   (This is a typical calculus problem but no calculus is needed!) http://jwilson.uga. while the fence for the other three sides costs \$1 per foot.000 square feet of land in a rectangular plot along a straight highway.html[11/9/12 3:34:14 PM] . Solving Problems with the AM-GM Inequality   Cost of Fencing a Field  Problem:  A farmer wants to fence in 60.coe.Using the Arithmetic Mean-Geometric Mean Inequality in Problem Solving.edu/EMT725/AMGM/SSMA.   The fence he plans to use along the highway costs \$2 per foot.Bnghm.

Bnghm.               Minimum of The problem is to find minimum values for this function in the range or   0 <  x  <  π.edu/EMT725/AMGM/SSMA.uga. however.Using the Arithmetic Mean-Geometric Mean Inequality in Problem Solving.   This one requires some change in the form of the equation in order to apply the AM-GM Inequality. First. http://jwilson. it is helpful to see a graph and have some sense of the equation.coe.   The graph at the right suggests that there may be two minimum values in the range  0 < x < π.html[11/9/12 3:34:14 PM] .

Using the Arithmetic Mean-Geometric Mean Inequality in Problem Solving.html[11/9/12 3:34:14 PM] .uga. Construct a Square with Same Area as a Given Rectangle The problem is to construct a square with straightedge and compass  that is the same area as a given rectangle.coe.edu/EMT725/AMGM/SSMA.   http://jwilson. The area will be    ab   and so the length of the side of the square is the geometric mean.Bnghm.

html[11/9/12 3:34:14 PM] . the radii become shorter because more of the fixed perimeter is in the arc.uga. Compare the areas of three sectors -.   Maximum Area of a Sector of a Circle With Fixed Perimeter Understanding the problem.Bnghm.Using the Arithmetic Mean-Geometric Mean Inequality in Problem Solving. As the angles increas. These are sectors that are an eighth circle.edu/EMT725/AMGM/SSMA. and 180 degrees.central angles of 45 degrees. quarter circle.coe. and semicircle.each with P = 100 -. A sector of a circle has a perimeter made up of two radii and an arc of the circle connecting the endpoints of the two radii. Consider: The red segment is the geometric mean of    a and   b. Note these angles are When these three areas are computed we get  http://jwilson. 90 degrees.

depending on a subsitution of   r   or  k  from the perimeter equation.html[11/9/12 3:34:14 PM] .    For them the approach might be as follows:        Let    k    be the fraction of a circle represented by the sector.     Either one can be used with the AM-GM Inequality to reach closure on the problem: http://jwilson.uga.    Where?  Sector as Fraction of a Circle Some students (and teachers) have an aversion to working with radian measure.      The area of the sector in the fraction notation can be formed as either a function of r or a function of k.   Somewhere there is a maximum area. as the angle increases from 45 to 90 to 180 the area increases and then decreases.Bnghm.   0 < k < 1.coe.Using the Arithmetic Mean-Geometric Mean Inequality in Problem Solving.  The  Perimeter is a constant and so we can represent it as  twice the radius plus the fraction of the circumference. Clearly.edu/EMT725/AMGM/SSMA.

Using the Arithmetic Mean-Geometric Mean Inequality in Problem Solving.edu/EMT725/AMGM/SSMA.uga.coe.html[11/9/12 3:34:14 PM] .Bnghm.         Sector in Radian Measure http://jwilson.   The angle is a little less than 120 degrees.  the maximum area would be when r = 25 and therefore the arc length is 50. Looking back at the example when  P = 100.

Inequalities Problems http://jwilson.Using the Arithmetic Mean-Geometric Mean Inequality in Problem Solving.edu/EMT725/AMGM/SSMA.Bnghm.uga. The parallel of this solution with the one where the square was the maximum area for rectangles with fixed perimeter is worth noting.html[11/9/12 3:34:14 PM] .coe.

Using the Arithmetic Mean-Geometric Mean Inequality in Problem Solving.Bnghm.html[11/9/12 3:34:14 PM] .   http://jwilson. Comparison of altitude and median in a right triangle Construct segments from the endpoints of the diameter to complete a right triangle.coe.uga.edu/EMT725/AMGM/SSMA.

5 only when x = 1.edu/EMT725/AMGM/SSMA.coe.         Maximum and Minimum of Rewrite For x > 0. For x < 0. using the arithmetic mean-geometric mean inequality.html[11/9/12 3:34:14 PM] . a similar argument leads to finding the minimum of the function at x = -1.5 and it is equal to .Bnghm.uga.Using the Arithmetic Mean-Geometric Mean Inequality in Problem Solving.Geometric Mean Inequality holds only for http://jwilson. Therefore the value of the function is always less than or equal to . Since the Arithmetic Mean -.

uga.edu/EMT725/AMGM/SSMA. postitive values. Multiplying each side of the inequality by -1 gives and equality occurs when x = -1.5 only when x = -1. See Graphs Maximum of f(x) = (1-x)(1+x)(1+x) We have three factors in the function and we want to know when the function reaches a maximum in the interval [0. we need the sum of these factors to equal a constant.    We can get that by the following   http://jwilson.5 and it is equal to -0.1/x.1].Using the Arithmetic Mean-Geometric Mean Inequality in Problem Solving.coe.Bnghm.x and .html[11/9/12 3:34:14 PM] . when x < 0 we have to apply the inequality to .   That is we need   -2x  rather  than   -x. Therefore the value of the function is always more than or equal to -0. In order to take advantage of the AM-GM Inequality. We know Keep in mind this is for x< 0 so -x and -1/x are postive.

html[11/9/12 3:34:14 PM] .Using the Arithmetic Mean-Geometric Mean Inequality in Problem Solving. soda cans have the same shape -. The function reaches this maximum value for x in [0. various factors such as tradition and supposed customer preferences may enter into decisions about what shape (e.a height of about 5 inches http://jwilson.coe.x) = 1 + x   2 .Bnghm.edu/EMT725/AMGM/SSMA. short and fat vs. Note. Minimum Surface Area of a Can of Fixed Volume In packaging a product in a can the shape of right circular cylinder. all 12 oz.2x = 1 + x          1 = 3x That is. for example. tall and skinny) can might be used for a fixed volume.uga.g.1] if and only if 2(1 .

What is the minimum S? http://jwilson. the radius and the height) would be determined by the minimum surface area for the can.g.coe. and a radius of about 1.Using the Arithmetic Mean-Geometric Mean Inequality in Problem Solving. What is the relationship between the radius and the height in order to minimize the surface area for a fixed volume? What would be the shape of a 12 ounce soda can that minimizes the amount of aluminum in the can? The Volume is fixed.edu/EMT725/AMGM/SSMA. The Surface area is a function of r and h.html[11/9/12 3:34:14 PM] .uga. Why? What if the decison was based on minimizing the material used to make the can? This would mean that for a fixed volume V the shape of the can (e.25 inches.Bnghm.

edu/EMT725/AMGM/SSMA.html[11/9/12 3:34:14 PM] .Using the Arithmetic Mean-Geometric Mean Inequality in Problem Solving.coe.Bnghm.uga.               ( Note that we have use the AM-GM Inequality for three positive numbers ) http://jwilson.

You want a rectangular pen with a partition parallel to a side.html[11/9/12 3:34:14 PM] .uga.7 sq. with maxim area of approximately 417.   Prove the Maximum Area of a Triangle with Fixed Perimeter is Equilateral http://jwilson. ft. with Partitions Parallel to a Side Suppose you have 100 feet of fencing. by  16. fixed length of fencing.  What is the shape of the pen and what is the maximum area? Show that the dimensions of the maximum area pen   is   25 ft. Maximum area of a Pen.coe.67 ft.edu/EMT725/AMGM/SSMA.   The 100 feet of fencing must enclose the four sides and the partition.Using the Arithmetic Mean-Geometric Mean Inequality in Problem Solving.Bnghm.

edu/EMT725/AMGM/SSMA. http://jwilson.coe.html[11/9/12 3:34:14 PM] .Using the Arithmetic Mean-Geometric Mean Inequality in Problem Solving.Bnghm.uga.

html[11/9/12 3:34:14 PM] .Bnghm.   Find the Maximum Area. http://jwilson.Using the Arithmetic Mean-Geometric Mean Inequality in Problem Solving. (use link for discussion)   The Box Problem Use the Arithmetic Mean-Geometric Mean Inequality to find the maximum volume of a box made from a 25 by 25 square sheet of cardboard by removing a small square from each corner and folding up the sides to form a lidless box.uga.edu/EMT725/AMGM/SSMA.coe. Maximum Area of a Triangle with Sides of 9. Minimum Distance from (0. 40x. and 41x Given a triangle with one side of length 9 units and the ratio of the other two sides is 40/41.0) to Find a point on the graph of            (follow the link to the discussion) that is closest to the origin.

Bnghm. http://jwilson.Using the Arithmetic Mean-Geometric Mean Inequality in Problem Solving.coe.edu/EMT725/AMGM/SSMA.html[11/9/12 3:34:14 PM] .uga.

Theorem:     In the Product of  n  positive numbers is equal to   1   then the sum if the numbers is greater than or equal to n.coe.Bnghm.Using the Arithmetic Mean-Geometric Mean Inequality in Problem Solving.html[11/9/12 3:34:14 PM] .uga.   http://jwilson.edu/EMT725/AMGM/SSMA.

the proof of  3  implies the first two.html[11/9/12 3:34:14 PM] .coe.uga.edu/EMT725/AMGM/SSMA. Clearly.Using the Arithmetic Mean-Geometric Mean Inequality in Problem Solving. http://jwilson.Bnghm.

Bnghm.html[11/9/12 3:34:14 PM] .Using the Arithmetic Mean-Geometric Mean Inequality in Problem Solving.edu/EMT725/AMGM/SSMA.uga.                  http://jwilson.coe.

then a + c > b + c. if a ≥ b and c ≥ d. and ac < bc if c < 0. number theory. with equality if and only if a = 0. But ﬁrstly. if a ≥ b. c. 1. Theorem 1 For real numbers a. and functional equations. subject to some constraint. One subarea of inequalities involves problems of the following form: “Maximise or minimise some expression. 1. b > 0. If a2 > b2 . then a > b. This area is not covered well at schools mainly because there is a large variety of methods involved. (c) If a > b. (b) If a > b and c > d. If a > b and c > d. then ac > bd. Three major examples are geometry. (f ) Let a.Inequalities and ‘Maximum-Minimum’ Problems Henry Liu. then ac ≥ bc if c > 0. and then discuss these socalled ‘max-min’ problems. then a + c > b + d.” We shall discuss some basic facts about inequalities. b. 26 February 2007 There are many olympiad level problems in mathematics which belong to areas that are not covered well at all at schools.2 Proving inequalities: some well-known techniques How do we prove an inequality at olympiad level? We will discuss many diﬀerent techniques. if a ≥ b and c > d. then ac > bc if c > 0. (d) Let a. (a) If a > b. then a + c ≥ b + c. Well-known inequalities 1. b. Such areas must be learned outside class time if one wants to be successful at solving olympiad style problems. and ac ≤ bc if c < 0. d. c. if a ≥ b and c > d. 1 . then a + c > b + d. (e) a2 ≥ 0 always holds. if a ≥ b and c ≥ d.1 The basics We begin with some obvious rules. note that such a problem usually has two parts: (a) Prove the inequality. (b) If possible. state when we have equality. then ac ≥ bd. then a + c ≥ b + d. then a ≥ b. if a2 ≥ b2 . if a ≥ b. d > 0. This set of notes considers another one of these areas: inequalities. we have the following. then ac > bd.

+ · · · + x1n x1 Then. Example 2. so 2 ≥ ab. . 2 . Let a. Prove that x2 + y 2 + z 2 ≥ xy + yz + zx for all real x. we need both x2 − 4 = 0 and x + 2 = 0. we may have x1 . we have QM ≥ AM ≥ GM ≥ HM. ie: when a = b since a. This is only possible when x = −2. . . . . n 2 x2 1 + · · · + xn . so a + b ≥ 2 ab. . For equality. Prove that x4 − 7x2 + 4x + 20 ≥ 0 for every real x. So 2x2 + 2y 2 + 2z 2 ≥ 2xy + 2yz + 2zx. We have 0 ≤ (x − y )2 + (y − z )2 + (z − x)2 = 2x2 + 2y 2 + 2z 2 − 2xy − 2yz − 2zx. ie: when x = y = z . and we have the result by dividing by 2. b ≥ 0. Example 1. Equality holds if and only if √ √ a − b = 0. We have QM = AM = GM = HM if and only if x1 = · · · = xn . a and b are well-deﬁned. 1. xn ≥ 0 instead. . xn ≥ 0. We deﬁne their quadratic mean to be QM = Deﬁne their arithmetic mean to be AM = Deﬁne their geometric mean to be GM = √ n x1 · · · xn . .1 Squares are non-negative A complicated looking inequality can be proved if we can show that it is equivalent to an inequality of the form ‘LHS ≥ 0’.2. Example 3. We have 0 ≤ ( a − b) = √ √ √ a+b a + b − 2 ab. . . we have QM ≥ AM ≥ GM.2. . xn > 0. If we remove ‘ HM’ from above. xn ≥ 0. b ≥ 0. . deﬁne their harmonic mean to be n HM = 1 . . xn > 0. √ b ≥ ab. y. . . n For x1 .It is a common mistake to ignore part (b). . where ‘LHS’ is a sum of squares. . Prove that a+ 2 √ √ 2 √ √ Since a.2 The QM-AM-GM-HM inequalities We now describe a chain of inequalities which will be extremely useful. x1 + · · · + xn . Theorem 2 (The QM-AM-GM-HM inequalities) Let x1 . 1. if x1 . When do we have equality? We have x4 −7x2 +4x+20 = (x4 −8x2 +16)+(x2 +4x+4) = (x2 −4)2 +(x+2)2 ≥ 0. b ≥ 0. ie: For x1 . Equality holds if and only if x − y = y − z = z − x = 0. . . . z .

Applying the AM-HM inequality. yn are arbitrary. . . Note that Example 2 above is the AMGM inequality for two terms. . And so on. with equality if and only if x1 = · · · = xn . Example 4. Let a. For example. . then the ﬁrst statement above is true with k = 0. . n with equality if and only if x1 = · · · = xn . y. xn = kyn ’ and ‘there exists a real number such that y1 = x1 . . The ‘AM-HM inequality’ says: For x1 . . xn . . c. We will have a problem if either all xi = 0 or all yi = 0. . . we have √ x1 + · · · + xn ≥ n x1 · · · xn . but the second is false! 3 . z > 0. x2 = ky2 . 1. For example. xn > 0. . b. yn = xn ’ are not quite equivalent. Theorem 3 (Cauchy-Schwarz inequality) Let x1 . Prove that a b c + + x y z x y z + + a b c ≥ 9. This easily rearranges to the required inequality. .3 Cauchy-Schwarz inequality This is another well-known inequality which is extremely useful. z satisfy x = b y c =z . x. . . . . . . Then 2 2 2 (x1 y1 + · · · + xn yn )2 ≤ (x2 1 + · · · + xn )(y1 + · · · + yn ). . x. c. if x1 = · · · = xn = 0 and y1 . x2 = ky2 . . . . Note: Be careful! The statements ‘there exists a real number k such that x1 = ky1 . or there exists a real number such that y1 = x1 . yn = xn . b. y1 . . . yn be real numbers. a Equality holds if and only if a. we have 1 a b c + + 3 x y z ≥ 3 x a y b z c + + . . y. . . not all zero. . . Equality holds if and only if either there exists a real number k such that x1 = ky1 . y2 = x2 .Taking any two of these means gives us a useful inequality. . . . xn ≥ 0.2. y2 = x2 . . . we have x1 + · · · + xn ≥ n 1 x1 n + ··· + 1 xn . . . xn = kyn . the ‘AM-GM inequality’ says: For x1 .

....... .... .. ..... f (a)) and (b.... . ....... ............ .... .. . ..... . . .... .... . ........4 Jensen’s inequality This is yet another very useful inequality. .. ...... . . . and for every λ with 0 < λ < 1... . .. ...... .f (b)) .... .... ..... .. .... .. .. . we have (32 + 42 + 122 )(x2 + y 2 + z 2 ) ≥ (3x + 4y + 12z )2 ... ... .. .............. .. ... f (b)) lies strictly below the graph of the function in the interval (a...... . Graphically. .... .. ........ Equality holds if and only if x = 3k .......... . .. . . z > 0........ . . ... this just says that........ .... y = 4k and z = 12k for some real number k > 0. . . .... .... 169 Since both sides are positive... f (a)) and (b.. .....Remarkably. ... .. ... ... We can now state Jensen’s inequality.......... . .. ... .. .... b).... . .. . ........ .. .. ....... .. . ....... .. . .... ........ ... (1−λ)f (a)+λf (b) . .. ... ... . . .. . . we have f ((1 − λ)a + λb) < (1 − λ)f (a) + λf (b)......... . 4 .... . we must ﬁrst deﬁne a property of functions...... ... 1 (3x + 4y + 12z ). ... ........... g (x) = sin x is strictly convex on [π........ . ..... f (x) = x2 is strictly convex on (−∞.... . ..... ..... ......... .. .. ... .. .... f ((1 − λ ) a + λb ) . A function f (x) deﬁned on an interval I is said to be strictly convex on I if for every a. ... ...... .... ... b).. ............... . ......... ....... y =f (x) . To state this inequality. ....2... ... ..f (a)) ..... .. .. the result follows by taking square roots... the chord joining (a.. . .... ..... ... .... the chord joining (a.. b ∈ I with a < b.. ... ... . . ........ ... .... . . ... ... ............... . Let x... .... . . ......... ...... . no matter where we choose a and b within I ................. . a (1−λ)a+λb b ........ . ...... .. ........ ............. . . is strictly concave on I if −f (x) is strictly convex on I ................ ... . . .... ........ (a.... .... .... deﬁned on interval I ........ .... ... for all a... .... ........ ....... . .. . .. . ... ... ... For example.. ........ .. ∞). ....... 1 x2 + y 2 + z 2 ≥ (3x + 4y + 12z )2 .. .. .. .... ........ . ............... . ...... . .. .... merely just stating one of these statements as the condition for equality! Example 5.. ..... Prove that (x2 + y 2 + z 2 )1/2 ≥ Applying Cauchy-Schwarz. (b.... ... ..... ........................ ... y. ......... f (b)) lies strictly above the graph of the function in the interval (a... .. .. ... ... b ∈ I with a < b. . ........ .... . almost all literature about the Cauchy-Schwarz inequality are not too careful about this.... 13 1.... ...... Graphically.......... .......... .. 2π ]... .. ... ... We say that f (x)... . ... ..

then calculus can oﬀer solutions. and reverse the inequality. A big disadvantage is that calculus only locates local minima/local maxima/points of inﬂexion. Equality holds if and only if w = x = y = z. ugly solution (but still a solution!). n Equality holds if and only if x1 = · · · = xn . Then. xn ∈ I . namely there may be elegant solutions to a certain problem. A detailed analysis of the function concerned is needed to show whether a local minimum/local maximum is also indeed global. but calculus oﬀers an easy. Theorem 5 Suppose f : I → R is twice diﬀerentiable on the interval I . we have 1 (f (x1 ) + · · · + f (xn )) ≥ f n 1 ( x1 + · · · + xn ) . which we will come to next. 2. . Then. and f (c) is the global maximum of f (x) if f (c) < 0. it has advantages and disadvantages. we may need to compute the second derivative to determine the nature of stationary values. Since f (x) = x3 is strictly convex on (0. ∞). Unfortunately. . Suppose that c ∈ I is the only critical point of f (x). A similar statement holds when we replace ‘convex’ by ‘concave’. . where is the function f (x) = x4 + 4x3 − 18x2 + 5x strictly convex/concave? One way to determine this would be to use calculus. by Jensen’s inequality. and is strictly convex on I . y. There is the following important principle in calculus as well. Example 6. prove that 16(w3 + x3 + y 3 + z 3 ) ≥ (w + x + y + z )3 . Moreover. z > 0. 5 . for every x1 . Nevertheless. This rearranges to the required inequality. Calculus 2.1 Calculus or no calculus? Despite the fact that a lot of references suggest that the knowledge of calculus is never required in olympiad style problems.Theorem 4 (Jensen’s inequality) Let f (x) be a function deﬁned on an interval I . For example. x. Another is that calculus oﬀers a way to determine where a function is strictly convex/concave. calculus is still a useful tool to have. . For w. it is not always easy to decide whether a function is strictly convex/concave on a certain interval. if the function concerned is simple enough and a detailed analysis is carried out. A big advantage is that it is quite powerful and sometimes it oﬀers a ‘cheap way out’. we have 1 3 (w + x3 + y 3 + z 3 ) ≥ 4 w+x+y+z 4 3 . f (c) is the global minimum of f (x) if f (c) > 0. and this is not always pleasant. Of course.

. .. . this problem is crying out for a calculus solution.... −3) and on (1.. The graph of f (x) looks like: .. f (x) = x4 + 4x3 − 18x2 + 5x is strictly convex on (−∞.... . ....... .. and 60 is the required minimum. by considering the graph of f (x). . and avoid calculus if the function is too complicated. .. . .. ... ..... and it behaves like 81x3/2 for large x......... 2 x .. .... ... ‘Maximum-Minimum Problems’ 3...... .... . . . ......... Let f (x) = 16x−1 + 81x3/2 ... Determine where the function f (x) = x4 + 4x3 − 18x2 + 5x is strictly convex and strictly concave. . ...... .. ... ..... 2 9 4 243 −1/2 −3 We ﬁnd that f ( 9 ) = 60. 60 ..... 3.... ... .... Hence by Theorem 5.. ...... . . f (x) = x + 81x3/2 . ......... f ( 4 9 minimum of f (x). in fact. We have f (x) = −16x−2 + 243 x1/2 . ..... Then...... .. .. As mentioned............... .. So by Theorem 6.. use the AM-GM inequality to solve this problem: apply AM16 16 16 81 3/2 81 3/2 .. ..... . .. .... ... ...... ....... .. use calculus only when the function is suﬃciently simple..... ..+ 81x3/2 .. ....... . y .. . 2 x ... ... . .. . . We have equality if and only if x= 4 .. Example 8. ... so in particular. . ....... We see that f (x) > 0 for 4 ) > 0. .. ....... . . . For x > 0.. and f (x) < 0 when −3 < x < 1.. . We have the following result.. . . .. .... .....1 How do we tackle them? Problems of the type 6 ...... Solving f (x) = 0 gives x = 4 ... ... .. . .. . . This does not seem as obvious as the calculus GM on 3 x 3x 3x approach... ... We ﬁnd easily that f (x) = 12x2 + 24x − 36 = 12(x + 3)(x − 1). . .. ∞).... and f (x) is strictly concave on I if f (x) < 0 for all x ∈ I ...... . . .............. ..... 16 ..... Indeed f (x) has the y -axis as an asymptote.. . . .. . . ... . ... .. . . . ... . . ... .. . minimise 16 x At ﬁrst sight. another thing that calculus can do is to determine where a function is strictly convex/concave... .. . Also.. ... .. f (x) = 32x + 4 x ..... ... . f ( 9 ) = 60 is the global all x > 0. 1).. 4 x 0 9 . We have f (x) > 0 when x < −3 or when x > 1. ... .. ..... . . .. and strictly concave on (−3.. 9 We may also argue without Theorem 5....... We could.... ..... ... Example 7.. .. .. ....... ... In a nutshell. .... . f (x) is strictly convex on I if f (x) > 0 for all x ∈ I ...... Theorem 6 Suppose f : I → R is twice diﬀerentiable on the interval I ... Observe that f (x) → ∞ when x → 0 and when x → ∞. ... . . ..

4 where we have used the condition 4x2 + y 2 + 16z 2 = 1 to get the last equality. and satisfying the constraint. so xz = 1. xz where we have used the condition xyz (x + y + z ) = 1 to get the last equality. Example 9. z where this minimum can be attained. one must then show that it/they may be attained by giving an example. y and z are positive real numbers satisfying 4x2 + y 2 + 16z 2 = 1. inequality above. for example. We have (7x + 2y + 8z )2 ≤ ((2x)2 + y 2 + (4z )2 ) 7 2 2 + 2 2 + 22 = 81 .y= 4 and z = 1 . subject to (constraint B)” can be diﬃcult to solve. we take the root with the positive sign. We 9 see that 7x + 2y + 8z ≤ 2 . and one would have to think of another approach. We have (x + y )(y + z ) = xy + xz + y 2 + yz = xz + y (x + y + z ) = xz + 1 . Substituting these into 4x + y + 16z = 1 gives 7 49 2 k + 4k 2 + 4k 2 = 1.“Maximise and/or minimise (expression A). Since y > 0. y and z are positive real numbers satisfying xyz (x + y + z ) = 1. This gets even more diﬃcult if. Maximise the expression 7x + 2y + 8z . we must give an example of x. once the maximum and/or minimum has/have been decided. Example 10. Then y must satisfy y (2 + y ) = 1. Finally. where x. solving this gives y = −1 ± 2. Minimise the expression (x + y )(y + z ). we have 2x = 2 k. The diﬃculty usually seems to be to decide how to incorporate constraint B into the problem of maximising and/or minimising expression A. 2 2 2 y = 2k . z satisfying 2 7 the given constraint. √ It is then easy to check that (x + y )(y + z ) = 2 when x = z = 1 and y = −1 + 2. where x. expression A involves more than two variables and constraint B is only a single equation relating the variables: there may be little or no hope if we try to ‘eliminate one variable’. To make things simple. giving k = 2 . One just needs to be a little clever in deciding what technique(s) to use. 4z = 2k for some k > 0. In the 1 . y. The techniques that we have already discussed more or less cover many of the strategies on how to solve such problems. It is 4 9 9 9 7 . xz xz To claim that ‘2’ is indeed the required minimum value. Again we must show that ‘ 9 ’ can be attained by some values of x. This leads to x = 18 . We try to cleverly apply the Cauchy-Schwarz inequality. For equality to hold in Cauchy-Schwarz. Applying the AM-GM inequality gives (x + y )(y + z ) = xz + 1 1 ≥ 2 xz · = 2. y. we have equality when xz = xz √ choose x = z = 1.

y and z be positive real numbers satisfying 1 ≤ xy + yz + zx ≤ 3. 10. 0 ≤ y ≤ 1 and 0 ≤ z ≤ 1. Let x. 9. z . Let x and y be real numbers. x y 5. y and z are real numbers satisfying x + y + z = 10. Find the minimum value of x2 + 4xy + 4y 2 + 2z 2 . 8. 3 Determine the range of values of (a) xyz and (b) x + y + z . (b) Maximise the expression x2 y + y 2 z + z 2 x − y 2 x − z 2 y − x2 z when 0 ≤ x ≤ 1. y and z be positive real numbers satisfying xyz = 32. (a) Maximise the expression x2 y − y 2 x when 0 ≤ x ≤ 1 and 0 ≤ y ≤ 1. Does x2 + 5y 2 + 8z 2 have a greatest value subject to the same constraint? 7. Prove that. 6. Prove that (x3 + y 3 )(x5 + y 5 ) ≤ 2(x8 + y 8 ). Show that x2 y 2 (x2 + y 2 ) ≤ 2. 4. 1. Minimise 2x2 + y 2 + z 2 . Let x. Let x. x x 2. 3. we do have 4x2 + y 2 + 16z 2 = 1 9 and 7x + 2y + 8z = 2 . Find the minimum value of 1 1 x4 + 4 + y 4 + 4 . where x. if x is a real number and x = 0. 8 . y and z are real numbers satisfying yz + zx + xy = −1. y and z be real numbers. y. Prove that x6 + y 6 + z 6 + 3x2 y 2 z 2 ≥ 2(x3 y 3 + y 3 z 3 + z 3 x3 ). Let x and y be non-negative real numbers satisfying x + y = 2. where x. then x8 − x 5 − 1 1 + 4 ≥ 0. Let x and y be non-zero real numbers satisfying x2 + y 2 = 4. Determine the smallest value of x2 + 5y 2 + 8z 2 .then easy to check that with these values of x. Problems Here are some inequalities and ‘maximum-minimum’ problems. 4.

least positive value of x + y 2 14. x y z 12.11. y and z be non-negative real numbers satisfying x + y + z = 1. Let a. 5 5 + b + ab b + c + bc c + a5 + ca 9 . Show that the 2 y2 is 1 . Let x. c. Prove that 4 x2 y + y 2 z + z 2 x ≤ . 3 16. Prove that √ a2 + b2 + c2 + 2 3abc ≤ 1. Let a. 27 15. Let x. y and z be positive real numbers satisfying x + y + z = 1. b. Find the minimum value of 1 1 1 1+ 1+ 1+ . Prove that a2 a b c 9 + 2 + 2 ≤ . Let x. Let x and y be real numbers such that 7x2 + 3xy + 3y 2 = 1. c be real numbers such that a. c be positive real numbers satisfying abc = 1. b. Let a. Let a. b. 13. +1 b +1 c +1 10 17. b+c+d c+d+a d+a+b a+b+c 3 18. b. Prove that a5 ab bc ca + 5 + 5 ≤ 1. b. d be positive real numbers satisfying ab + bc + cd + da = 1. Prove that 7(xy + yz + zx) ≤ 2 + 9xyz. c be positive real numbers satisfying a + b + c = 1. c ≥ − 4 and a + b + c = 1. y and z be non-negative real numbers satisfying x + y + z = 1. Prove that b3 c3 d3 1 a3 + + + ≥ .

Dipak Singh.. sin2 4θ . use + } 2. 1+ tan2 θ = sec2 θ how did we get these formulas? Already explained. cos 4θ . a sin θ ± b cos θ = ±√ (a2 + b2 ) { for min. for max. sin2 9θ …sin2 nθ Min value -1 Max value +1 Negative Signs inside out   Sin (. cos2 8θ … cos2 nθ 0 ≤ Cos2 nθ ≤ 1 Sin θ Cos θ -1/2 +1/2 -1/2 ≤ Sin θ Cos θ ≤ ½ observe that in case of sin2θ and cos2θ.This is a guest article by Mr.Sin (θ) Cos (-θ) = Cos (θ) Ratta-fication formulas 1. the minimum value if 0 and not (-1). First Remember following identities: Trig-Identities 1. Today we’ll see how to find the maximum value (greatest value ) or the minimum value (least value) of a trigonometric function without using differentiation.θ) = . sin2 θ + cos2 θ = 1 2.. use . cos 7θ … cos nθ sin2 θ . sin 9θ …. keep doing the steps while reading this article. use . for max. click me Min-Max table Can be written as -1 ≤ Sin nθ ≤ 1 -1 ≤ Cos nθ ≤ 1 0 +1 Can be written as 0 ≤ Sin2 nθ ≤ 1 cos2 θ . sin nθ cos θ. use . 1+ cot2 θ = cosec2 θ 3. cos2 3θ . for max. Take a pen and note-book. Min. a sin θ ± b sin θ = ±√ (a2 + b2 ) { for min. sin 2θ. a cos θ ± b cos θ = ±√ (a2 + b2 ) { for min. use + } 4.. value of (sin θ cos θ)n = (½)n . use + } 3. Why does this happen? because (-1)2=+1 sin θ.

The AM GM Logic Let A . We can confidently say that my age is not less 25 years. Infact it can not be less than 25. (1) G. sin θ ≥ .M Meaning. cos2 θ = 1/ sec2 θ ). For example: Find minimum value of 4 tan2θ + 9 cot2θ (they’ll not ask maximum value as it is not defined.M ≥ G .B are any two numbers then. Arithmetic Mean (AM)= (A + B) / 2 and Geometric Mean (GM) = √ (A.e. value of sin θ + cosec θ or tan θ + cot θ or cos2 θ + sec2 θ etc.. y ≥ 77 ( minimum value of y = 77 ) If.M of given equation = √ (4 tan2θ .M ≥ G. sin θ = 1/ cosec θ . 50 . Means. Or in other words my minimum age is 25 years. 786 or 1000 years etc… but it can not be 24 or 19 or Sweet 16 .B)      Hence. we get A.M logic. if Age ≥ 25 years ( minimum age = 25 )      Similarly. we come across a special case of trigonometric identities like to find min. A. Showing numerically. strictly. ” What could you conclude about my age from this statement ? Answer : My age can be anywhere between 25 to infinity … means it can be 25 . Arithmetic mean is always greater than or equal to geometric mean. ) We know that tan2θ = 1/ cot2θ . so tan x cot =1) .1 ( minimum value of Sin θ = -1 ) If. the first trigonometric term is opposite of the second term or vice-versa ( tan θ = 1/ cot θ . . These identities have one thing in common i. If I say x ≥ 56 ( minimum value of x = 56 ) If. These type of problems can be easily tackled by using the concept of A. 9 cot2θ ) = √ 4 * 9 # ( tan2θ and cot2θ inverse of each other.99.M of given equation = (4 tan2θ + 9 cot2θ) / 2 ….M ≥ G. hence applying A. x + y ≥ 133 ( minimum value of x + y = 133 ) If. tan θ + cot θ ≥ 2 (minimum value of tan θ + cot θ = 2 ) ]] Sometimes.M ( We can check it by putting any values of A and B ) Consider the following statement “ My age is greater than or equal to 25 years .

values.M ≥ G. min.= √ 36 = 6 …. A function and its reciprocal have same sign. M From equations (1) and (2) above we get.M ≥ G.M (a cos2 θ + b sec2 θ / 2 ) ≥ √ (a cos2 θ . where T1 = 1 / T2 Positive sign in between terms is mandatory. Keeping these tools (not exhaustive) in mind we can easily find Maximum or Minimum values easily. (otherwise how would you calculate mean ? ) Directly apply 2√ab . Extra facts:     The reciprocal of 0 is + ∞ and vice-versa. Rearrange/Break terms if necessary -> priority should be given to direct use of identities -> find terms eligible for A. (2) Now. to sine and cosines -> finally put known max. A.M logic :     Term should be like a T1 + b T2 .M logic -> if any.M ≥ G. The reciprocal of 1 is 1 and -1 is -1. we know that A. b sec2 θ) a cos2 θ + b sec2 θ ≥ 2 √ (ab) ( minimum value 2 √ab ) So. Summary: While using A. SSC CGL 2012 Tier II Question What is The minimum value of sin2 θ + cos2 θ + sec2 θ + cosec2 θ + tan2 θ + cot2 θ A. we can use 2 √ab directly in these kind of problems. 1 B. 3 . apply -> convert remaining identities.M ≥ G.. => (4 tan2 θ + 9 cot2θ) / 2 ≥ 6 Multiplying both sides by 2 => 4 tan2 θ + 9 cot2 θ ≥ 12 ( minimum value of tan2 θ + cot2 θ is 12 ) Deriving a common conclusion:      Consider equation a cos2 θ + b sec2 θ ( find minimum value) As. if any. If a function has a maximum value its opposite has a minimum value.

= 1 + 2 + sec2 θ + cosec2 θ changing into sin and cos values ( Because we know maximum and minimum values of Sin θ.C. and we get = 1 + 2 + (sin2 θ + cos2 θ)/( sin2 θ .C. (sin2 θ + cos2 θ) + sec2 θ + cosec2 θ + tan2 θ + cot2 θ = (1) + sec2 θ + cosec2 θ + tan2 θ + cot2 θ Using A. 5 D.) = 1 + 2 + (1/ cos2 θ) + (1/ sin2 θ) solving taking L. cos2 θ)…. 7 Solution: We know that sin2 θ + cos2 θ = 1 (identitiy#1) Therefore.. B. Cos θ :P and by using simple identities we can convert all trigonometric functions into equation with Sine and Cosine. value of (sin θ cos θ)n = (½)n (Ratta-fication formula #4) Apply them into eq1. D. 1 2 3 5 We can solve this question via two approaches .eq1 but we already know two things sin2 θ + cos2 θ=1 (trig identity #1) Min. C.M ≥ G.M logic for tan2 θ + cot2 θ we get . cos2 θ) = 1 + 2 + (1/1/4) = 1+2+4 = 7 (correct answer D) The least value of 2 sin2 θ + 3 cos2 θ (CGL2012T1) A.M = 1 + 2 + (sin2 θ + cos2 θ)/( sin2 θ .

1 a sin θ ± b cos θ = ±√ (a2 + b2 ) { for min. 1 = 2 sin2 θ + 2 cos2 θ + cos2 θ =2(sin2 θ + cos2 θ) + cos2 θ .(but as per min-max table.(because sin2 θ + cos2 θ=1=>cos2 θ=1. the minimum value of cos2 θ=0) = 2 + 0 = 2 (correct answer B) Approach #2 convert equation into one identity .cos2 θ) + 3 cos2 θ = 2 – 2 cos2 θ + 3 cos2 θ = 2 + cos2 θ = 2 + 0 = 2 ( correct answer B ) The maximum value of Sin x + cos x is A. √2 1/ √2 1 2 Applying Ratta-fication formulae No. D. use + } in the given question.either sin or cos first convert it into a sin equation : = 2 sin2 θ + 3 (1. B.sin2 θ) . use . for max.sin2 θ) = 2 sin2 θ + 3 – 3 sin2 θ = 3 .Approach #1 Break the equation and use identity no. value of sin2 θ in order to get minimum value . Converting into a cos equation : = 2 sin2 θ + 3 cos2 θ = 2 (1.. C. (but sin2 θ + cos2 θ=1) = 2 + cos2 θ . we’ve to find the max value of Sin x + cos x = + √ (12+ 12 ) = √2 ( correct answer A ) .sin2 θ = 3 – ( 1) = 2 (but Min. value of sin2 θ is 0 …confusing ???? ) As sin2 θ is preceded by a negative sign therefore we have to take max.

Sin +1 . 6 Solution: We know that. -5 (D) 4. use . 1 Sin -1 . use + } in the given question.Sin 1 .1 a sin θ ± b cos θ = ±√ (a2 + b2 ) { for min. 6 (B) 4. we’ve to find the max value of 3 Sin x – 4 Cos x = + √ (32+ 42 ) = √25 = 5 ( correct answer B ) Min Max values of sin 4x + 5 are (A) 2. the minimum value is 4 and maximum value is 6 ( correct answer D ) Minimum and maximum value of Sin Sin x is A. Do not exist -1. B. 5 (C) -4. B. Sin 1 . C.. 4 ≤ Sin 4x +5 ≤ 6 Therefore. -1 ≤ Sin nx ≤ 1 = -1 ≤ Sin 4x ≤ 1 Adding 5 throughout. D.The maximum value of 3 Sin x – 4 Cos x is A. for max. D. -1 5 7 9 Solution: Applying Ratta-fication formulae No. C.

[Sin(-θ) is same as – Sin θ ] Therefore. visit mrunal. -1 ≤ Sin nx ≤ 1 = Sin (-1) ≤ Sin x ≤ Sin (1) = .org/aptitude .We know that. For more articles on trigonometry and aptitude.Sin 1 ≤ Sin x ≤ Sin 1 . Minimum value is –Sin 1 and maximum is Sin 1 ( correct answer D) The key to success is Practice! Practice! Practice! Drop your problems in the comment box.