OR : Module-4

The Mathematical Programming Problems can be classified into two types:

1. Linear programming problems, and

2. Non-linear programming problems

Prepared by Ranbir Kalita

In Linear Programming models, the characteristic assumption is the Linearity of the
Objective and Constraint Functions. But there are many situations where the Objective
function and some or all of the Constraints are Non-linear functions. For example,

• Decrease in transportation cost with larger lot sizes may not be linear. (Transportation
problem)
• Decrease in production cost with increase in production volume may not be linear.
(Inventory problem)

An LPP is generally expressed as,

Maximize / Minimize Z = f (x1 , x2 , x3 , ……… , xn)

Prepared by Ranbir Kalita

Subject to,

g1 (x1 , x2 , x3 , ……… , xn) ≤ = ≥ b1

g2 (x1 , x2 , x3 , ……… , xn) ≤ = ≥ b2

g3 (x1 , x2 , x3 , ……… , xn) ≤ = ≥ b3

……………………………………

gm (x1 , x2 , x3 , ……… , xn) ≤ = ≥ bm

Where,

xj ≥ 0 ; j = 1 , 2 , 3 , ……… , n
Prepared by Ranbir Kalita

The above LPP becomes non-linear if the Objective Function and / or one or more of the
Constraints are non-linear in X (x1 , x2 , x3 , ……… , xn). Such a problem is called as Non-
linear Programming Problem (NLPP).

∴ The General Form of an NLPP is,

Maximize / Minimize Z = f (X)

Page - 1
Mechanical Engineering Department Jorhat Engineering College
 OR : Module-4

Subject to,
gi (X) ≤ = ≥ bi ; i = 1 , 2 , 3 , ……… , m

Where, X≥0

Prepared by Ranbir Kalita

Here, f (X) or some or all of gi (X) or both are non-linear.

∴ In the NLPP, we have to determine the n-tuple X = (x1 , x2 , x3 , ……… , xn) to minimize /
maximize the Objective Function subject to the constraints.

Finding the optimum solution in an LPP is comparatively easy because of the certainty that
the optimum solution lies at one or more extreme points of the Feasible Region. Therefore,
number of iterations to reach the optimal point is limited. But, in case of NLPP, the optimal
solution may be located at any point along the boundary of the feasible region or even within
the feasible region.

• Also due to non-linearity of the objective function, it becomes difficult to distinguish

Prepared by Ranbir Kalita

between the Local and Global Solution.
• Again, the optimality test on an NLPP becomes difficult when the feasible region is
not convex.

The main complexity in the solution of NLPP comes from the fact that, there is no single
algorithm, like the Simplex algorithm in case of LPP, which can be employed to solve
efficiently all the NLPPs. An algorithm that performs well on one type of problem may
perform very poorly on a problem with a different structure.

The optimization techniques to solve the NLPPs can be categorized into two types:

1. Classical optimization techniques (analytical process based on Differential Calculus).

2. Non-classical optimization techniques (based on Numerical Process).
Prepared by Ranbir Kalita

As already stated, there is no single methodology or algorithm to tackle every non-linear

programming problem.

The main limitation of the Classical Optimization Techniques (COTs) is that, the
techniques are not applicable to non-differential / discontinuous functions.

• The COTs are useful in finding the optimum solution, i.e. Maxima or Minima of
continuous and differentiable functions.

Page - 2
Mechanical Engineering Department Jorhat Engineering College
 OR : Module-4

• These are analytical methods and make use of the differential calculus in locating the
optimum solution.

The kind of problems that COTs address can be broadly divided into two types:

Prepared by Ranbir Kalita

1. Un-constrained optimization: Such problems involve only the objective function
and no constraints. That is, the Saddle Point at which the maximum or minimum
value occurs need not to satisfy any condition (constraint). The objective function in
un-constrained optimization problems can be:
• Single-variable functions, or
• Multi-variable functions.
2. Constrained optimization: Such problems involve an objective function as well as
one or more constraints. The constraint(s) could be of equality type or inequality
type. The Saddle Point at which the maximum or minimum value occurs also need to
satisfy the given condition(s) [constraint(s)]. The Objective function and the

Prepared by Ranbir Kalita

constraint(s) are multi-variable functions.

In non-linear programming, a function may have more than one maximum or minimum
values. That is, when a non-linear function is plotted, it will have more than one Peak and
more than one Velley. Each of these peaks is a Local Maxima, and each of the valleys is a
Local Minima.

Since an NLPP can have more than one maximum or minimum values, therefore, these
maximum or minimum values can be categorized into Relative or Local Maximum /
Minimum and Global or Absolute Maximum / Minimum. The following definitions are given
for a single variable NL function. But they are equally application to multi-variable NL
functions too.
Prepared by Ranbir Kalita

Fig. 4.4

Page - 3
Mechanical Engineering Department Jorhat Engineering College
 OR : Module-4

Let, f (x) be a continuous function defined in [a , b] ∈ R.

• Relative / Local Maximum: A function of one variable f (x) is said to have a relative
/ local maximum at x = x* if f (x*) ≥ f (x + h) for all sufficiently small positive or

Prepared by Ranbir Kalita

negative values of h.

• Relative / Local Minimum: A function of one variable f (x) is said to have a relative /
local minimum at x = x* if f (x*) ≤ f (x + h) for all sufficiently small positive or
negative values of h.

• Global / Absolute Maximum: A function of one variable f (x) is said to have a global
/ absolute maximum at x = x* if f (x*) ≥ f (x) ∀ x in the interval [a , b]. That is, a
Global or Absolute Maximum of a function is a value which is higher than all values
of the function.

• Global / Absolute Minimum: A function of one variable f (x) is said to have a global
/ absolute minimum at x = x* if f (x*) ≤ f (x) ∀ x in the interval [a , b]. A Global or

Prepared by Ranbir Kalita

Absolute Minimum of a function is a value which is the lowest of all values of the
function.

The point x* mentioned in the definitions for Local and Global Maximum / Minimum is said
to be a Stationery Point / Point of Inflection if the function f (x*) has neither maximum or
minimum.

Graphically, it can be observed that,

• At a point when maximum occurs (local / global), the function changes its trend
from increasing to decreasing.
• At a point when minimum occurs (local / global), the function changes its trend from
decreasing to increasing.
Prepared by Ranbir Kalita

• At the point of inflection, the function has either increasing or decreasing trend on
both sides of the point.

Page - 4
Mechanical Engineering Department Jorhat Engineering College
 OR : Module-4

Prepared by Ranbir Kalita

Fig. 4.1

As can be seen in Fig. 4.1, the tangents drawn at the maximum, minimum and point of
inflections have Zero Slopes or they are parallel to the x-axis.

Prepared by Ranbir Kalita

Fig. 4.2

For example, as shown in Fig. 4.2, when a single variate NL function is plotted within
interval x = a and x = b, the function is seen to have Three Local Maxima at x1 , x3 and x6.
The Global Maximum is al x6.

Similarly, there are Three Local Minima at x2 , x4 and x7. The Global Minimum is at x7.
Prepared by Ranbir Kalita

The function f (x5) corresponding to x5 which has a Zero Slope is called the Point of
Inflection. If a point with Zero Slope is not a Maximum or Minimum, then it must
automatically be a Point of Inflection.

Page - 5
Mechanical Engineering Department Jorhat Engineering College
 OR : Module-4

Concave and Convex Functions

When the plot of a single variable function results in a curve, always curving downward or
not curving at all is called a Concave Function.

Prepared by Ranbir Kalita

Figure 4.2

Prepared by Ranbir Kalita

As shown in Fig. 4.2, the slope of a concave function (single variable) is such that for any
two points on the curve in the Feasible Region (S), the line joining the points is always below
the function. There is always a Unique Global Maximum of such a function.

Also, for a concave function of several variables, if it has a Local Maxima, then the same is
the Global Maximum for the function.

When the plot of a single variable function results in a curve always curving upward is called
a Convex Function.

Prepared by Ranbir Kalita

Figure 4.3

Page - 6
Mechanical Engineering Department Jorhat Engineering College
 OR : Module-4

The Fig. 4.3 shows a single variable Convex Function. A line joining any two points on the
curve in the Feasible Region (S) is always above the function. There is always a Unique
Global Minimum of such a function

Prepared by Ranbir Kalita

Also, for a concave function of several variables, if it has a Local Minima, then the same is
the Global Minimum for the function.

Important:

For a doubly differentiable single variable function f (x) without any constraint
(unconstrained optimization), then,

𝑑2 𝑓(𝑥)
• If ≤ 0 ∀ 𝑥 ; then the local or relative maxima is also the global maxima.
𝑑𝑥 2
Such a function is the concave function.
𝑑2 𝑓(𝑥)
• If ≥ 0 ∀ 𝑥 ; then the local or relative minima is also the global minima.

Prepared by Ranbir Kalita

𝑑𝑥 2
Such a function is the convex function.

Single Variable Un-Constrained Optimization

A single variable optimization problem is the one in which the value of x = x* is to be found
in the interval [a , b] such that x* maximizes / minimizes f (x).

Prepared by Ranbir Kalita

Fig. 4.4

Page - 7
Mechanical Engineering Department Jorhat Engineering College
 OR : Module-4

Necessary and Sufficient Condition for Maxima / Minima at a point

If a function f (x) is defined in the interval a ≤ x ≤ b and has a Saddle Point (local maxima or
minima or point of inflection) at x = x*, where a ≤ x* ≤ b and if the First Order Derivative

Prepared by Ranbir Kalita

f ′ (x) exists as a finite number at x = x*, then f ′ (x*) = 0.

Let, f ′ (x*) = f ′′ (x*) = f ′′′ (x*) = ……… = f n-1 (x*) = 0 but f n (x*) ≠ 0, where f n is the n-th
order derivative, then x = x* is a:

(i) Local maximum value of f (x) if f n (x*) < 0 and n is even.

(ii) Local minimum value of f (x) if f n (x*) > 0 and n is even.
(iii) Neither maximum nor minimum if n is odd. In this case x* is a Point of Inflection.

Problem 4.1: Determine the Maximum and Minimum values of the function:

f (x) = 3x4 – 4x3 – 24x2 + 48x +15

Prepared by Ranbir Kalita

Solution: Find the first order derivative of the given non-linear function, equate to zero and
find the value(s) of the decision variable x.

f ′ (x) = 0 => 12x3 – 12x2 – 48x + 48 = 0 => x3 – x2 – 4x + 4 = 0 => x = –2 , 2 , 1

This means, that f ′ (x) = 0 at x = –2 , 1 and 2. For checking whether the function f (x) has
local maximum or local minimum at x = –2 , 1 and 2, calculate the Second Order Derivative
f ′′ (x) for the function f (x). therefore,

f ′′ (x) = 36x2 – 24x – 48

Here n = 2 (even), and

∵ at x = –2 , f ′′ (–2) = 144 ( > 0 ), ∴ f (x) has a local minima at x = –2 and the value is
Prepared by Ranbir Kalita

–97 (convex curve).

at x = 1 , f ′′ (1) = –36 ( < 0 ), ∴ f (x) has a local maxima at x = 1 and the value is 38
(concave curve).

at x = 2 , f ′′ (2) = 48 ( > 0 ), ∴ f (x) has a local minima at x = 2 and the value is 31

(convex curve).

Page - 8
Mechanical Engineering Department Jorhat Engineering College
 OR : Module-4

Alternate method:

f ′ (x) = 0 => 12x3 – 12x2 – 48x + 48 = 0 => x3 – x2 – 4x + 4 = 0 => x = –2 , 1 , 2

Now find two values of f ′ (x), one < x and one > x for each value of x = –2 , 1 , 2 and show

Prepared by Ranbir Kalita

them on the number line. Observe the signs (+ve / –ve) on either side of the Saddle Point (x).

• If the value of f ′ (x) is changing from +ve to –ve, this means it is a concave curve and
as such Local Maximum occurs at x.

• If the value of f ′ (x) is changing from–ve to +ve, this means it is a convex curve and
as such Local Minima occurs at x.

Here,

f ′ (–3) = –240

f ′ (–1) = 72

Prepared by Ranbir Kalita

f ′ (1.5) = –10.5

f ′ (3) = 120

On putting the values of x = –2 , 1 and 2 in f (x), we get the local maxima and local minima
values as f′ (–2) = –97 , f (1) = 38 and f′ (2) = 31.

Problem 4.2: Determine the Maximum and Minimum values of the function:

f (x) = 2x6 – 6x4 + 6x2 + 10

Solution: Find the first order derivative of the given non-linear function, equate to zero and
find the value(s) of the decision variable x.
Prepared by Ranbir Kalita

f ′ (x) = 0 => 12x5 – 24x3 + 12x = 0 => 12x (x2 – 1)2 = 0 => 12x = 0 and (x2 – 1)2 = 0
=> x = 0 => x2 = 1
=> x = ±1

Therefore, x = –1 , 0 , 1

Page - 9
Mechanical Engineering Department Jorhat Engineering College
 OR : Module-4

Here,

f ′ (–2) = –216

f ′ (–0.5) = –3.375

Prepared by Ranbir Kalita

f ′ (0.5) = 3.375

f ′ (2) = 216

Therefore, at x = 0, the local minimum value is f (0) = 10.

Problem 4.3:
2 − (𝑥 − 1)2 ; 0 ≤ 𝑥 < 3
Maximize 𝑓(𝑥) = {
−3 + (𝑥 − 4)2 , 3 ≤ 𝑥 < 6

Prepared by Ranbir Kalita

Solution: Find the first order derivative of the given function.

−2(𝑥 − 2) , 0≤𝑥<3
𝑓 ′ (𝑥) = {
2(𝑥 − 2) , 3≤𝑥<6

Equating the first f ′ (x) and second f ′ (x) values separately to zero, we get,

–2(x – 1) = 0 and 2(x – 4) = 0

=> x = 1 and => x = 4

Therefore,

f ′ (0) = –2(0 – 1) = 2

f ′ (2) = –2(2 – 1) = –2
Prepared by Ranbir Kalita

f ′ (3) = 2(3 – 4) = –2

f ′ (5) = 2(5 – 4) = 2

Page - 10
Mechanical Engineering Department Jorhat Engineering College
 OR : Module-4

Therefore, the maximum value of the given function occurs at x = 1 and the corresponding
value is f (1) = 2 – (1 – 1)2 = 2.

The minimum value of the given function occurs at x = 4 and the corresponding value is f (4)

Prepared by Ranbir Kalita

= –3 + (4 – 4)2 = –3.

Problem 4.4: Find the maximum or minimum values for the functions:

(a) f (x) = x2 – 4x (b) f (x) = 2x3 + 3x2 – 12x (c) f (x) = 3x4 – 16x3 + 24x2

Solution: (a) f (x) = x2 – 4x

f ′ (x) = 2x – 4 = 0 => x = 2

Therefore,

f ′ (1) = –2 and f ′ (3) = 2

Prepared by Ranbir Kalita

Therefore, the local minimum value is f (2) = 22 – (4×2) = –4 and the corresponding location
of the local minimum point is (2 , –4).

(b) f (x) = 2x3 + 3x2 – 12x

f ′ (x) = 6x2 + 6x – 12 = 0 => x2 + x – 2 = 0 => x = –2 , 1

Therefore,

f ′ (–3) = 4

f ′ (–1) = –2

f ′ (2) = 4

Hence, the local minimum value is f (1) = (2×13) + (3×12) – (12×1) = –7 corresponding to
Prepared by Ranbir Kalita

point (1 , –7).

And, the local maximum value is f (–2) = [2×(–23)] + [3×(–22)] – [12×(–2)] = 20

corresponding to point (–2 , 20).

Page - 11
Mechanical Engineering Department Jorhat Engineering College
 OR : Module-4

Problem 4.5: [Example 16.5-1, pp. 1403, Hira & Gupta]

For each of the following functions, show whether it is convex, concave or neither.

(a) f (x) = 10 – x2 (b) f (x) = x4 + 6x2 + 12x (c) f (x) = 2x3 – 3x2

Prepared by Ranbir Kalita

(d) f (x) = x4 + x2 (e) f (x) = x3 + x4

Solution: (a) f (x) = 10 – x2

𝑑𝑓(𝑥) 𝑑 2 𝑓(𝑥)
∴ = −2𝑥 and = −2 < 0 always
𝑑𝑥 𝑑𝑥 2

∵ d2f (x)/dx2 is always < 0 ∀ x , therefore, the given function is Concave.

(b) f (x) = x4 + 6x2 + 12x

𝑑𝑓(𝑥) 𝑑 2 𝑓(𝑥)
∴ = 4𝑥 3 + 16𝑥 + 12 and = 12𝑥 2 + 16
𝑑𝑥 𝑑𝑥 2

Prepared by Ranbir Kalita

∵ d2f (x)/dx2 is always > 0 ∀ x , therefore, the given function is Convax.

(c) f (x) = 2x3 – 3x2

𝑑𝑓(𝑥) 2
𝑑2 𝑓(𝑥)
∴ = 6𝑥 − 6𝑥 and = 12𝑥 − 16
𝑑𝑥 𝑑𝑥 2

𝑑2 𝑓(𝑥) 𝟏
Put = 0 ⇒ 12𝑥 − 6 = 0 ⇒ 𝒙 =
𝑑𝑥 2 𝟐

At x = 0, d2f (x)/dx2 = – 6 < 0, therefore, the given function is Concave at x = 0.

For x > 1/2, d2f (x)/dx2 is always > 0, therefore, the given function is Convex ∀ x > 1/2.

(d) f (x) = x4 + x2
Prepared by Ranbir Kalita

𝑑𝑓(𝑥) 3
𝑑 2 𝑓(𝑥)
= 4𝑥 + 2𝑥 and = 12𝑥 2 + 2
𝑑𝑥 𝑑𝑥 2

∵ d2f (x)/dx2 is always > 0 ∀ x , therefore, the given function is Convax.

Page - 12
Mechanical Engineering Department Jorhat Engineering College
 OR : Module-4

(e) f (x) = x3 + x4

𝑑𝑓(𝑥) 𝑑2 𝑓(𝑥)
∴ = 3𝑥 2 + 4𝑥 3 and = 6𝑥 + 12𝑥 2
𝑑𝑥 𝑑𝑥 2

Prepared by Ranbir Kalita

𝑑2 𝑓(𝑥) 𝟏
Put 2
= 0 ⇒ 6𝑥 + 12𝑥 2 = 0 ⇒ 𝒙 = 𝟎 , −
𝑑𝑥 𝟐

At x = 0, d2f (x)/dx2 = 0.

At x = –1/2 = – 0.5, d2f (x)/dx2 = 0

At x = – 0.6, d2f (x)/dx2 = 0.72

Therefore, the given function is Convex for all x > – 0.5.

Problem 4.6: Find the maximum and minimum values of the following function:

Prepared by Ranbir Kalita

f (x) = 3x4 – 4x3 – 24x2 + 48x + 15

Solution: f ′ (x) = 12x3 – 12x2 – 48x + 48 = 0 => x = – 2 , 1 , 2

Therefore,

f ′ (–3) = – 240

f ′ (–1) = 72

f ′ (1.5) = – 10.5

f ′ (3) = 120

Hence,
Prepared by Ranbir Kalita

• The given function is Convex at x = – 2 and the corresponding local minimum value
is f (–2) = – 97.
• The given function is Concave at x = 1 and the corresponding local maximum value is
f (1) = 38.
• The given function is Convex at x = 2 and the corresponding local minimum value is
f (2) = 31.

The graph for the function is shown below:

Page - 13
Mechanical Engineering Department Jorhat Engineering College
 OR : Module-4

Prepared by Ranbir Kalita

Prepared by Ranbir Kalita
Prepared by Ranbir Kalita

Page - 14
Mechanical Engineering Department Jorhat Engineering College
 OR : Module-4

Two-variable Un-constrained Optimization

Here, we have to find the maximum and minimum values of a function having two
independent variables. The function is not subjected to any constraint.

Prepared by Ranbir Kalita

Conditions for Maxima and Minima:

Let f is a function of two independent variables x and y. Since two independent variables are
involved, therefore, the differentiation will be partial.

𝜕𝑓 𝜕𝑓
Put =0 ; = 0 and find the values of 𝑥 and 𝑦.
𝜕𝑥 𝜕𝑦

These obtained values of x and y gives the Saddle Point at which maxima or minima will
occur.

Find the values of r , s and t , where,

Prepared by Ranbir Kalita

𝜕 2𝑓 𝜕 2𝑓 𝜕 2𝑓
𝑟 = ( 2) ; 𝑠=( ) ; 𝑡 = ( 2)
𝜕𝑥 (𝑥,𝑦) 𝜕𝑥𝜕𝑦 (𝑥,𝑦) 𝜕𝑦 (𝑥,𝑦)

(1) If rt – s2 > 0 and r > 0, then f (x , y) is minimum at (x , y).

(2) If rt – s2 > 0 and r < 0, then f (x , y) is maximum at (x , y).

(3) If rt – s2 < 0 and r > 0, then neither maximum nor minimum at (x , y).

(4) If rt – s2 = 0 and r > 0, then the case is doubtful and need further investigation.

Problem 4.7: Find the minimum / maximum for the function: f (x , y) = x2 + y2 + 6x + 12

Solution:
Prepared by Ranbir Kalita

𝜕𝑓 𝜕𝑓
= 2𝑥 + 6 = 0 ⇒ 𝑥 = −3 and = 2𝑦 = 0 ⇒ 𝑦 = 0
𝜕𝑥 𝜕𝑦

Therefor, the saddle point is (x , y) = (–3 , 0)

𝜕 2𝑓 𝜕 2𝑓 𝜕 2𝑓
𝑟= = 2 ; 𝑠 = = 0 ; 𝑡 = =2
𝜕𝑥 2 𝜕𝑥𝜕𝑦 𝜕𝑦 2

At (x , y) = (–3 , 0): r = 2 ; s = 0 ; t = 2

Page - 15
Mechanical Engineering Department Jorhat Engineering College
 OR : Module-4

Now, rt – s2 = (2×2) – 02 = 4 > 0 and r = 2 > 0

Since, rt – s2 > 0 and r > 0, then f (x , y) is minimum at (–3 , 0). And the minimum
value is: f (–3 , 0) = (–3)2 + 0 + [6×(–3)] + 12 = 9 – 18 + 12 = 3.

Prepared by Ranbir Kalita

Problem 4.7: Find the minimum / maximum for the function: u = x13 + x23 + x12 + 4x22 + 6

Solution:

𝜕𝑢 −2
= 3𝑥1 2 + 2𝑥1 = 0 ⇒ 𝑥1 (3𝑥1 + 2) = 0 ⇒ 𝑥1 = 0 and 3𝑥1 + 2 = 0 ⇒ 𝑥1 =
𝜕𝑥1 3
𝜕𝑢 −8
= 3𝑥2 2 + 8𝑥2 = 0 ⇒ 𝑥2 (3𝑥2 + 8) ⇒ 𝑥2 = 0 and 3𝑥2 + 8 = 0 ⇒ 𝑥2 =
𝜕𝑥2 3

Therefore, the saddle points are (0 , 0) and (–2/3 , –8/3).

Now,

Prepared by Ranbir Kalita

𝜕 2𝑢 𝜕 2𝑢 𝜕 2𝑢
𝑟= = 6𝑥1 + 2 ; 𝑠 = = 0 ; 𝑡 = = 6𝑥2 + 8
𝜕𝑥1 2 𝜕𝑥1 𝜕𝑦1 𝜕𝑥2 2

At (0 , 0): r = 2 ; s = 0 ; t = 8 and since rt – s2 = 16 > 0 and r = 2 > 0 , therefore, the

function is minimum at (0 , 0) and the minimum value is: u = 03 + 03 + 02 + (4×0) + 6 = 6.

Again, at (–2/3 , –8/3): r = –2 ; s = 0 ; t = –8 and since rt – s2 = 16 > 0 and r = –2 < 0 ,

therefore, the function is minimum at (–2/3 , –8/3) and the minimum value is: u = (–2/3)3 + (–
8/3)3 + (–2/3)2 + [4×(–8/3)] + 6 = 50/3.

NOTE: Instead of (0 , 0) and (–2/3 , –8/3), the points (0 , –8/3) and (–2/3 , 0) can also be
tested as the saddle points. But at these points, there will be neither maximum nor minimum.
Prepared by Ranbir Kalita

Problem 4.8: Find the minimum / maximum for the function:

u = x3 + 3xy2 – 15x2 – 15y2 + 72x
Solution:

𝜕𝑢
= 3𝑥 2 + 3𝑦 2 − 30𝑥 + 72 = 0 … … (1)
𝜕𝑥

𝜕𝑢
= 6𝑥𝑦 − 30𝑦 = 0 ⇒ 𝑦(6𝑥 − 30) = 0 ⇒ 𝑦 = 0 and 6𝑥 − 30 = 0 ⇒ 𝑥 = 5
𝜕𝑦

Page - 16
Mechanical Engineering Department Jorhat Engineering College
 OR : Module-4

Put y = 0 in (1): 3x2 + [3×0] – 30x + 72 = 0 => x = (6 , 4)

Put x = 5 in (1): [3×52] + 3y2 – (30×5) + 72 = 0 => y = ±1

Therefore, the saddle points are: (6 , 0) ; (4 , 0) and (5 , 1) ; (5 , –1)

Prepared by Ranbir Kalita

Now,
𝜕 2𝑢 𝜕 2𝑢 𝜕 2𝑢
𝑟 = 2 = 6𝑥 − 30 ; 𝑠 = = 6𝑦 ; 𝑡 = 2 = 6𝑥 − 30
𝜕𝑥 𝜕𝑥𝜕𝑦 𝜕𝑦

At (4 , 0) At (6 , 0) At (5 , 1) At (5 , –1)
r = –6 , s = 0 , t = –6 r=6,s=0,t=6 r=0,s=6,t=0 r = 0 , s = –6 , t = 0
rt – s2 = 36 > 0 rt – s2 = 36 > 0 rt – s2 = –36 < 0 rt – s2 = –36 < 0
and r = –6 < 0 and r = 6 > 0 and r = 0 and r = 0
Therefore, the Therefore, the
Therefore, the Therefore, the
function u is neither function u is neither
function u is function u is
maximum nor maximum nor

Prepared by Ranbir Kalita

maximum at (4 , 0) minimum at (6 , 0)
minimum at (5 , 1) minimum at (5 , –1)

Problem 4.9: [Example 16.5-2(a), pp. 1403, Hira & Gupta]

Find the minimum / maximum for the function: f(X) = x1x2 – x12 – x22
Solution:

𝜕𝑓 𝜕𝑓
= 𝑥2 − 2𝑥1 = 0 … … (1) = 𝑥1 − 2𝑥2 = 0 ⇒ 𝑥1 = 2𝑥2 … … (2)
𝜕𝑥1 𝜕𝑥2

Put (2) in (1): x2 – 2×2x2 = 0 => x2 = 0 …… (3)

Put (3) in (1): x1 = 2×0 => x1 = 0

Prepared by Ranbir Kalita

Therefore, (0 , 0) is the Saddle Point.

𝜕 2𝑓 𝜕 2𝑓 𝜕 2𝑓
𝑟= = −2 ; 𝑠 = =1 ; 𝑡= = −2
𝜕𝑥1 2 𝜕𝑥1 𝑥2 𝜕𝑥2 2
At (0 , 0): r = –2 ; s = 1 ; t = –2. Therefore, rt – s2 = 3 > 0 and r = –2 < 0

Therefore, f(X) is maximum at (0 , 0). That is Concave function and the maximum value is

Max. f(X) = 0.

Page - 17
Mechanical Engineering Department Jorhat Engineering College
 OR : Module-4

Strictly Strictly
Entity Convex Concave
Convex Concave
𝜕 2𝑓 𝜕 2𝑓 𝜕 2𝑓
[ × ] − [ ] ≥0 >0 ≥0 >0
𝜕𝑥1 2 𝜕𝑥2 2 𝜕𝑥1 𝜕𝑥2

Prepared by Ranbir Kalita

𝜕 2𝑓
≥0 >0 ≤0 <0
𝜕𝑥1 2
𝜕 2𝑓
≥0 >0 ≤0 <0
𝜕𝑥2 2

Problem 4.9: [Example 16.5-2(b), pp. 1403, Hira & Gupta]

Find the minimum / maximum for the function: f(X) = 3x1 + 2x12 + 4x2 + x22 – 2x1x2
Solution:
𝜕𝑓 𝜕 2𝑓
= 3 + 4𝑥1 − 2𝑥2 and 𝑟 = =4
𝜕𝑥1 𝜕𝑥1 2
𝜕 2𝑓

Prepared by Ranbir Kalita

𝑠= = −2
𝜕𝑥1 𝜕𝑥2

𝜕𝑓 𝜕 2𝑓
= 4 + 2𝑥2 − 2𝑥1 and 𝑡 = =2
𝜕𝑥2 𝜕𝑥2 2
𝜕 2𝑓 𝜕 2𝑓 𝜕 2𝑓 𝜕 2𝑓 𝜕 2𝑓
Since [ × ] − [ ] = 4 > 0 and = 4 > 0 , =2>0
𝜕𝑥1 2 𝜕𝑥2 2 𝜕𝑥1 𝜕𝑥2 𝜕𝑥1 2 𝜕𝑥2 2

Therefore, the function f(X) is Strictly Convex.

Determination of Saddle Point and Minimum Value:

𝜕𝑓 𝜕𝑓
= 3 + 4𝑥1 − 2𝑥2 = 0 … … (1) and = 4 + 2𝑥2 − 2𝑥1 = 0
𝜕𝑥1 𝜕𝑥2
2𝑥1 − 4
⇒ 𝑥2 = … … (2)
2
Prepared by Ranbir Kalita

Put (2) in (1):

2𝑥1 − 4 𝟕
3 + 4𝑥1 − 2 ( ) = 0 ⇒ 𝒙𝟏 = … … (3)
2 𝟔

Putting (3) in (2):

−𝟓
𝒙𝟐 =
𝟔

Page - 18
Mechanical Engineering Department Jorhat Engineering College
 OR : Module-4

Therefore, the saddle point is (7/6 , –5/6).

At (7/6 , –5/6),

𝜕 2𝑓 𝜕 2𝑓 𝜕 2𝑓

Prepared by Ranbir Kalita

𝑟= = 4 ; 𝑠 = = −2 ; 𝑡= =2
𝜕𝑥1 2 𝜕𝑥1 𝜕𝑥2 𝜕𝑥2 2

Since rt – s2 = 4 > 0 and r = 4 > 0, therefore, the function f(X) has a minimum value at (7/6 ,
–5/6) and the minimum value is 199/36 = 5.528.

Problem 4.10: [Example 16.5-2(c), pp. 1403, Hira & Gupta]

Find the minimum / maximum for the function: f(X) = x12 + 3x1x2 + 2x22
Solution:
𝜕𝑓 𝜕 2𝑓
= 2𝑥1 + 3𝑥2 and 𝑟 = =2
𝜕𝑥1 𝜕𝑥1 2
𝜕 2𝑓
𝑠= =3

Prepared by Ranbir Kalita

𝜕𝑥1 𝜕𝑥2

𝜕𝑓 𝜕 2𝑓
= 3𝑥1 + 4𝑥2 and 𝑡 = =4
𝜕𝑥2 𝜕𝑥2 2
𝜕 2𝑓 𝜕 2𝑓 𝜕 2𝑓 𝜕 2𝑓 𝜕 2𝑓
Since [ × ] − [ ] = −1 < 0 and = 2 > 0 , =4>0
𝜕𝑥1 2 𝜕𝑥2 2 𝜕𝑥1 𝜕𝑥2 𝜕𝑥1 2 𝜕𝑥2 2

Therefore, the function f(X) is neither concave nor convex.

Determination of Saddle Point and Minimum Value:

𝜕𝑓 𝜕𝑓 −3𝑥1
= 2𝑥1 + 3𝑥2 = 0 … … (1) and = 3𝑥1 + 4𝑥2 = 0 ⇒ 𝑥2 = … … (2)
𝜕𝑥1 𝜕𝑥2 4

Putting (2) in (1), we get,

Prepared by Ranbir Kalita

−3𝑥1
2𝑥1 + 3 ( ) = 0 ⇒ 𝒙𝟏 = 𝟎 … … (3)
4

Putting (3) in (2), we get, 𝒙𝟐 = 𝟎

Therefore, the saddle point is (0 , 0)

At (0 , 0), r = 2 ; s = 3 ; t= 4 and since rt – s2 = –1 < 0 , therefore, the function f(X) is

neither maximum nor minimum at (0 , 0).

Page - 19
Mechanical Engineering Department Jorhat Engineering College
 OR : Module-4

Lagrange’s Method of Un-determined Multiplier

Unlike finding the maximum or minimum value at a point (saddle point) that we did in the
previous classes, in this method, the point at which the maximum / minimum is to be

Prepared by Ranbir Kalita

calculated should also satisfy one constraint (equality type). That is,

Max. / Min. f(x, y, z) ; subject to g(x, y, z) = k

L = f + λ.g

Solution Steps:

(1) f ′(x) = 0 ; f ′(y) = 0 ; f ′(z) = 0 ; f ′(λ) = 0

(2) Find x, y, z and λ using the above equations. (x, y, z) is the saddle point.

(3) Construct the following matrix to check whether it is a maxima / minima at (x, y, z).

Prepared by Ranbir Kalita

𝐿11 − 𝑘 𝐿12 𝐿13 𝑔1
𝐿 𝐿22 − 𝑘 𝐿23 𝑔2
[ 21 ]
𝐿31 𝐿32 𝐿33 − 𝑘 𝑔3
𝑔1 𝑔2 𝑔3 0

Equate the determinant of this matrix to zero and find the value of k.

If k < 0 , then maxima occurs at (x, y, z)

If k > 0 , then maxima occurs at (x, y, z)

𝜕 2𝐿 𝜕 2𝐿 𝜕 2𝐿
𝐿11 = 2 ; 𝐿12 = ; 𝐿13 =
𝜕𝑥 𝜕𝑥𝜕𝑦 𝜕𝑥𝜕𝑧

𝜕 2𝐿 𝜕 2𝐿 𝜕 2𝐿
𝐿21 = ; 𝐿22 = 2 ; 𝐿23 =
𝜕𝑦𝜕𝑥 𝜕𝑦 𝜕𝑦𝜕𝑧
Prepared by Ranbir Kalita

𝜕 2𝐿 𝜕 2𝐿 𝜕 2𝐿
𝐿31 = ; 𝐿32 = ; 𝐿33 = 2
𝜕𝑧𝜕𝑥 𝜕𝑧𝜕𝑦 𝜕𝑧

𝜕𝑔 𝜕𝑔 𝜕𝑔
𝑔1 = ; 𝑔2 = ; 𝑔3 =
𝜕𝑥 𝜕𝑦 𝜕𝑧

Property: The determinant of any matrix is zero when any two rows are equal.

Page - 20
Mechanical Engineering Department Jorhat Engineering College
 OR : Module-4

Problem 4.11: Given f(x) = x1 x2 x3 subject to g = x1 + x2 + x3 = 1 . Use Lagrange’s

multipliers to find any maximum / minimum values.

Solution: We know, L = f + λg => L = x1 x2 x3 + λ (x1 + x2 + x3 – 1)

Prepared by Ranbir Kalita

𝜕𝐿
= 𝑥2 𝑥3 + 𝜆 = 0 ⇒ 𝜆 = −𝑥2 𝑥3 … … (1)
𝜕𝑥1

𝜕𝐿
= 𝑥1 𝑥3 + 𝜆 = 0 ⇒ 𝜆 = −𝑥1 𝑥3 … … (2)
𝜕𝑥2

𝜕𝐿
= 𝑥1 𝑥2 + 𝜆 = 0 ⇒ 𝜆 = −𝑥1 𝑥2 … … (3)
𝜕𝑥3

𝜕𝐿
= 𝑥1 + 𝑥2 + 𝑥3 − 1 = 0 … … (4)
𝜕𝜆

From (1) & (2): x1 = x2 from (2) & (3): x2 = x3

Prepared by Ranbir Kalita

Putting the above values in equation (4), we get,

𝑥1 + 𝑥2 + 𝑥3 − 1 = 0 ⇒ 𝑥1 + 𝑥2 + 𝑥3 = 1 ⇒ 𝑥1 + 𝑥1 + 𝑥1 = 1 ⇒ 3𝑥1 = 1 ⇒ 𝑥1 = 1⁄3

Therefore, 𝑥1 = 𝑥2 = 𝑥3 = 1⁄3 Hence, the saddle point is (1/3 , 1/3 , 1/3)

Now,
𝐿11 − 𝑘 𝐿12 𝐿13 𝑔1
𝐿 𝐿22 − 𝑘 𝐿23 𝑔2
[ 21 ]
𝐿31 𝐿32 𝐿33 − 𝑘 𝑔3
𝑔1 𝑔2 𝑔3 0

Therefore,

𝜕 2𝐿 𝜕 2𝐿 𝜕 2𝐿
Prepared by Ranbir Kalita

𝐿11 = = 0 ; 𝐿12 = = 𝑥3 ; 𝐿13 = = 𝑥2

𝜕𝑥 2 𝜕𝑥𝜕𝑦 𝜕𝑥𝜕𝑧

𝜕 2𝐿 𝜕 2𝐿 𝜕 2𝐿
𝐿21 = = 𝑥3 ; 𝐿22 = 2 = 0 ; 𝐿23 = = 𝑥1
𝜕𝑦𝜕𝑥 𝜕𝑦 𝜕𝑦𝜕𝑧

𝜕 2𝐿 𝜕 2𝐿 𝜕 2𝐿
𝐿31 = = 𝑥2 ; 𝐿32 = = 𝑥1 ; 𝐿33 = 2 = 0
𝜕𝑧𝜕𝑥 𝜕𝑧𝜕𝑦 𝜕𝑧

Page - 21
Mechanical Engineering Department Jorhat Engineering College
 OR : Module-4

𝜕𝑔 𝜕𝑔 𝜕𝑔
𝑔1 = = 1 ; 𝑔2 = = 1 ; 𝑔3 = =1
𝜕𝑥 𝜕𝑦 𝜕𝑧

Putting the values, we get,

Prepared by Ranbir Kalita

0−𝑘 𝑥3 𝑥2 1 0−𝑘 1⁄3 1⁄3 1
𝑥 0−𝑘 𝑥1 1 1⁄3 0−𝑘 1⁄3 1
[ 3 ]=[ ]
𝑥2 𝑥1 0−𝑘 1 1⁄3 1⁄3 0−𝑘 1
1 1 1 0 1 1 1 0

Now, to find out the value of k in the above matrix, equate the determinant of the above
matrix to zero.

Property: Determinant of a matrix is zero, when any two rows are identical.

Considering Rows-1 & 2, the last two elements are equal; and the first two elements will be
equal when,

Prepared by Ranbir Kalita

0 − 𝑘 = 1⁄3 and 1⁄3 = 0 − 𝑘

⇒ 𝑘 = − 1⁄3 and ⇒ 𝑘 = − 1⁄3

This means 𝑘 = − 1⁄3 . Since 𝑘 < 0 , therefore, Maximum will occur at (1/3 , 1/3 , 1/3)

Therefore, f (1/3 , 1/3 , 1/3) = 1/3 × 1/3 × 1/3 = 1/27

Problem 4.12: Obtain the extreme points for the function:

f = x1 + x2 + x3 subject to x12 + x22 + x32 = 1

Find whether the extreme points are maximum / minimum?

Solution: We know, 𝐿 = 𝑓 + 𝜆𝑔 = 𝑥1 + 𝑥2 + 𝑥3 + 𝜆(𝑥1 2 + 𝑥2 2 + 𝑥3 2 − 1)

Prepared by Ranbir Kalita

𝜕𝐿
= 1 + 2𝑥1 𝜆 = 0 ⇒ 𝑥1 = − 1⁄2𝜆 … … (1)
𝜕𝑥1

𝜕𝐿
= 1 + 2𝑥2 𝜆 = 0 ⇒ 𝑥2 = − 1⁄2𝜆 … … (2)
𝜕𝑥2

𝜕𝐿
= 1 + 2𝑥3 𝜆 = 0 ⇒ 𝑥3 = − 1⁄2𝜆 … … (3)
𝜕𝑥3

Page - 22
Mechanical Engineering Department Jorhat Engineering College
 OR : Module-4

𝜕𝐿
= 𝑥1 2 + 𝑥2 2 + 𝑥3 2 − 1 = 0 … … (4)
𝜕𝜆

1 1 1 3 √3
⇒ + 2 + 2 − 1 = 0 ⇒ 𝜆2 = ⇒ 𝜆 = ±

Prepared by Ranbir Kalita

4𝜆 2 4𝜆 4𝜆 4 2

√3 1 2 1
For 𝜆 = , from (1) => 𝑥1 = − 2 × ⇒ 𝑥1 = −
2 √3 √3

1
Therefore, 𝑥1 = 𝑥2 = 𝑥3 = −
√3

1 1 1 √3
Hence, saddle point is (− ,− ,− ) for 𝜆 =
√3 √3 √3 2

√3 1 2 1
For 𝜆 = − , from (1) => 𝑥1 = (− 2) × (− ) ⇒ 𝑥1 =
2 √3 √3

1
Therefore, 𝑥1 = 𝑥2 = 𝑥3 =

Prepared by Ranbir Kalita

√3

1 1 1 √3
Hence, saddle point is ( , , ) for 𝜆 = −
√3 √3 √3 2

Now,
𝐿11 − 𝑘 𝐿12 𝐿13 𝑔1
𝐿 𝐿22 − 𝑘 𝐿23 𝑔2
[ 21 ]
𝐿31 𝐿32 𝐿33 − 𝑘 𝑔3
𝑔1 𝑔2 𝑔3 0

Therefore,
𝜕 2𝐿 𝜕 2𝐿 𝜕 2𝐿
𝐿11 = = 2𝜆 ; 𝐿12 = = 0 ; 𝐿13 = =0
𝜕𝑥1 2 𝜕𝑥1 𝜕𝑥2 𝜕𝑥1 𝜕𝑥3

𝜕 2𝐿 𝜕 2𝐿 𝜕 2𝐿
Prepared by Ranbir Kalita

𝐿21 = = 0 ; 𝐿22 = = 2𝜆 ; 𝐿23 = =0

𝜕𝑥2 𝜕𝑥1 𝜕𝑥2 2 𝜕𝑥2 𝜕𝑥3

𝜕 2𝐿 𝜕 2𝐿 𝜕 2𝐿
𝐿31 = = 0 ; 𝐿32 = = 0 ; 𝐿33 = = 2𝜆
𝜕𝑥3 𝜕𝑥1 𝜕𝑥3 𝜕𝑥2 𝜕𝑥3 2

𝜕𝑔 𝜕𝑔 𝜕𝑔
𝑔1 = = 2𝑥1 ; 𝑔2 = = 2𝑥2 ; 𝑔3 = = 2𝑥3
𝜕𝑥1 𝜕𝑥2 𝜕𝑥3

Page - 23
Mechanical Engineering Department Jorhat Engineering College
 OR : Module-4

Therefore,
𝐿11 − 𝑘 𝐿12 𝐿13 𝑔1 2𝜆 − 𝑘 0 0 2𝑥1
𝐿 𝐿22 − 𝑘 𝐿23 𝑔2 0 2𝜆 − 𝑘 0 2𝑥2
[ 21 ]=0⇒[ ]=0
𝐿31 𝐿32 𝐿33 − 𝑘 𝑔3 0 0 2𝜆 − 𝑘 2𝑥3
𝑔1 𝑔2 𝑔3 0 2𝑥1 2𝑥2 2𝑥3 0

Prepared by Ranbir Kalita

Considering Rows-1 & 2, the last two elements are equal; and the first two elements will be
equal when,

2𝜆 − 𝑘 = 0 and 0 = 2𝜆 − 𝑘

⇒ 𝑘 = 2𝜆 and ⇒ 𝑘 = 2𝜆

√3 √3
For 𝜆 = ; 𝑘 =2× ⇒ 𝑘 = √3 > 0
2 2

1 1 1
Therefore, the function has a minima at (− ,− ,− ) and the minimum value is,
√3 √3 √3

Prepared by Ranbir Kalita

1 1 1
𝑓𝑚𝑖𝑛 = 𝑥1 + 𝑥2 + 𝑥3 = − − − ⇒ 𝒇𝒎𝒊𝒏 = −√𝟑
√3 √3 √3

√3 √3
For 𝜆 = − ; 𝑘 = 2 × (− ) ⇒ 𝑘 = −√3 < 0
2 2

1 1 1
Therefore, the function has a maxima at ( , , ) and the maximum value is,
√3 √3 √3

1 1 1
𝑓𝑚𝑎𝑥 = 𝑥1 + 𝑥2 + 𝑥3 = + + ⇒ 𝒇𝒎𝒂𝒙 = √𝟑
√3 √3 √3

Problem 4.13: A company produces TV sets at two different factories A & B. suppose x
number of TVs are produced at factory A and y number of TVs are produced at factory B.
The total cost of production of x + y TVs is given by C(x , y) = 6x2 + 12y2
Prepared by Ranbir Kalita

If 90 numbers of TV sets are produced in a month, then calculate how much TV sets should
be produced at each of the factories?

Solution: Given, C(𝑥, 𝑦) = 6𝑥 2 + 12𝑦 2 or 𝑓 = 6𝑥 2 + 12𝑦 2

Subject to, 𝑥 + 𝑦 = 90 ⇒ 𝑔 = 𝑥 + 𝑦 − 90

We know, 𝐿 + 𝑓 + 𝜆. 𝑔 ⇒ 𝐿 = 6𝑥 2 + 12𝑦 2 + 𝜆(𝑥 + 𝑦 − 90)

Page - 24
Mechanical Engineering Department Jorhat Engineering College
 OR : Module-4

𝜕𝐿
= 12𝑥 + 𝜆 = 0 ⇒ 12𝑥 = −𝜆 ⇒ 𝑥 = − 𝜆⁄12 … … (1)
𝜕𝑥

𝜕𝐿
= 24𝑦 + 𝜆 = 0 ⇒ 12𝑥 = −𝜆 ⇒ 𝑦 = − 𝜆⁄24 … … (2)
𝜕𝑦

Prepared by Ranbir Kalita

𝜕𝐿 𝜆 𝜆 𝜆
= 𝑥 + 𝑦 − 90 = 0 ⇒ − − − 90 = 0 ⇒ − = 90 ⇒ 𝝀 = −𝟕𝟐𝟎
𝜕𝜆 12 24 8

𝜆 −720
From (1): 𝑥 = − 12 = − ( ) ⇒ 𝒙 = 𝟔𝟎
12

𝜆 −720
From (2): 𝑦=− = −( ) ⇒ 𝒚 = 𝟑𝟎
24 24

Therefore, Saddle Point (60 , 30)

Therefore,
𝜕2𝐿 𝜕 2𝐿
𝐿11 = 2 = 12 ; 𝐿12 = =0

Prepared by Ranbir Kalita

𝜕𝑥 𝜕𝑥𝜕𝑦

𝜕 2𝐿 𝜕 2𝐿
𝐿21 = = 0 ; 𝐿22 = 2 = 24
𝜕𝑦𝜕𝑥 𝜕𝑦

𝜕𝐿 𝜕𝐿
𝑔1 = = 1 ; 𝑔2 = =1
𝜕𝑥 𝜕𝑦

Now, putting the values, we get,

𝐿11 − 𝑘 𝐿12 𝑔1 12 − 𝑘 0 1
[ 𝐿21 𝐿22 − 𝑘 𝑔2 ] = 0 ⇒ [ 0 24 − 𝑘 1] = 0
𝑔1 𝑔2 0 1 1 0

12 − 𝑘 = 0 and 0 = 24 − 𝑘

⇒ 𝑘 = 12 > 0 and 𝑘 = 24 > 0

Prepared by Ranbir Kalita

This means Minima at (60 , 30) and the minimum value is:

𝐶𝑚𝑖𝑛 = 6𝑥 2 + 12𝑦 2 = (60 × 602 ) + (12 × 302 ) ⇒ 𝑪𝒎𝒊𝒏 = 𝟑𝟐𝟒𝟎𝟎

Page - 25
Mechanical Engineering Department Jorhat Engineering College
 OR : Module-4

Karush-Kuhn-Tucker (KKT) Conditions

Whenever we need to find out the minimum or maximum value at a point w.r.t some
constraints (one / two) of inequality type, then Karush-Kuhn-Tucker (KKT) conditions are to

Prepared by Ranbir Kalita

be used.

Solution steps:

(1) Express the given NLPP into Canonical Form.

(2) Find / Identify f (x1, x2); h1 (x1, x2) and h2 (x1, x2).

(3) Write the KKT conditions:

𝜕𝑓 𝜕ℎ1 𝜕ℎ2
− 𝜆1 − 𝜆2 = 0 … … … (1)
𝜕𝑥1 𝜕𝑥1 𝜕𝑥1

𝜕𝑓 𝜕ℎ1 𝜕ℎ2

Prepared by Ranbir Kalita

− 𝜆1 − 𝜆2 = 0 … … … (2)
𝜕𝑥2 𝜕𝑥2 𝜕𝑥2

𝜆1 ℎ1 (𝑥1 , 𝑥2 ) = 0 … … … (3)

𝜆2 ℎ2 (𝑥1 , 𝑥2 ) = 0 … … … (4)

ℎ1 (𝑥1 , 𝑥2 ) ≤ 0 (for max. ) | ℎ1 (𝑥1 , 𝑥2 ) ≥ 0 (for min. ) … … … (5)

ℎ2 (𝑥1 , 𝑥2 ) ≤ 0 (for max. ) | ℎ2 (𝑥1 , 𝑥2 ) ≥ 0 (for min. ) … … … (6)

𝑥1 , 𝑥2 ≥ 0 … … … (7)

𝜆1 , 𝜆2 ≥ 0 (for max. ) | 𝜆1 , 𝜆2 ≤ 0 (for min. ) … … … (8)

(4) Solve equations (1) to (4) to find out the values of x1 , x2 , λ1 and λ2 . These values must
satisfy the inequalities (5) to (8).
Prepared by Ranbir Kalita

For solving equations (1) to (4), following four cases may be considered:

Case-1: 𝜆1 = 𝜆2 = 0 Case-2: 𝜆1 ≠ 0 ; 𝜆2 ≠ 0

Case-3: 𝜆1 = 0 ; 𝜆2 ≠ 0 Case-4: 𝜆1 ≠ 0 ; 𝜆2 = 0

Page - 26
Mechanical Engineering Department Jorhat Engineering College
 OR : Module-4

Problem 4.14: Solve the following using KKT conditions:

Max. Z = 8𝑥1 + 10𝑥2 − 𝑥1 2 − 𝑥2 2

Subject to: 3𝑥1 + 2𝑥2 ≤ 6 Where, 𝑥1 , 𝑥2 ≥ 0

Prepared by Ranbir Kalita

Solution: Given 𝑓(𝑥1 , 𝑥2 ) = 8𝑥1 + 10𝑥2 − 𝑥1 2 − 𝑥2 2

And ℎ(𝑥1 , 𝑥2 ) = 3𝑥1 + 2𝑥2 − 6

Therefore, the KKT conditions are:

𝜕𝑓 𝜕ℎ
−𝜆 = 0 ⇒ 8 − 2𝑥1 − 𝜆(3) = 0 ⇒ 8 − 2𝑥1 − 3𝜆 = 0 … … … (1)
𝜕𝑥1 𝜕𝑥1

𝜕𝑓 𝜕ℎ
−𝜆 = 0 ⇒ 10 − 2𝑥2 − 𝜆(2) = 0 ⇒ 10 − 2𝑥2 − 2𝜆 = 0 … … … (2)
𝜕𝑥2 𝜕𝑥2

Prepared by Ranbir Kalita

𝜆ℎ(𝑥1 , 𝑥2 ) = 0 ⇒ 𝜆(3𝑥1 + 2𝑥2 − 6) = 0 … … … (3)

ℎ(𝑥1 , 𝑥2 ) ≤ 0 ⇒ 3𝑥1 + 2𝑥2 − 6 ≤ 0 … … … (4)

𝑥1 , 𝑥2 ≥ 0 … … … (5) and 𝜆 ≥ 0 … … … (6)

Case-1: when λ = 0

Putting λ = 0 in equation (1) & (2):

(1) ⇒ 8 − 2𝑥1 = 0 and (2) ⇒ 10 − 2𝑥2 = 0

⇒ 𝒙𝟏 = 𝟒 and ⇒ 𝒙𝟐 = 𝟓

Putting these values of x1 and x2 in equation (4), we get,

Prepared by Ranbir Kalita

(4) ⇒ (3 × 4) + (2 × 5) − 6 = 𝟏𝟔 ≮ 𝟎

Therefore, even though the values of x1 and x2 satisfies the inequalities (4) & (5), yet
inequality (6) is not satisfied. Hence, the solution x1 = 4 and x2 = 5 must be rejected.

Page - 27
Mechanical Engineering Department Jorhat Engineering College
 OR : Module-4

Case-2: when λ ≠ 0

From equation (3), if λ ≠ 0, then, 3𝑥1 + 2𝑥2 − 6 must be equal to zero. Therefore,

3𝑥1 + 2𝑥2 − 6 = 0 … … … (7)

Prepared by Ranbir Kalita

Now, solving equations (1) , (2) & (7), we get,

4 33 32
𝑥1 = ; 𝑥2 = ; 𝜆=
13 13 13

Since, these values of x1 , x2 and λ are satisfying the inequalities (4) , (5) & (6), therefore,
these set of values can be accepted. Hence,

4 33 4 2 33 2 𝟐𝟕𝟕
Max. Z = (8 × ) + (10 × ) − ( ) − ( ) ⇒ 𝐌𝐚𝐱. 𝐙 =
13 13 13 13 𝟏𝟑

Prepared by Ranbir Kalita

Problem 4.15: Solve the following using KKT conditions:

Max. Z = 10𝑥1 + 10𝑥2 − 𝑥1 2 − 𝑥2 2

Subject to: 𝑥1 + 𝑥2 ≤ 14 and − 𝑥1 + 𝑥2 ≤ 6 Where, 𝑥1 , 𝑥2 ≥ 0

Solution: Given 𝑓(𝑥1 , 𝑥2 ) = 10𝑥1 + 10𝑥2 − 𝑥1 2 − 𝑥2 2

ℎ1 (𝑥1 , 𝑥2 ) = 𝑥1 + 𝑥2 − 14 and ℎ2 (𝑥1 , 𝑥2 ) = −𝑥1 + 𝑥2 − 6

The KKT conditions are:

𝜕𝑓 𝜕ℎ1 𝜕ℎ2
− 𝜆1 − 𝜆2 = 0 ⇒ (10 − 2𝑥1 ) − 𝜆1 (1) − 𝜆2 (−1) = 0
𝜕𝑥1 𝜕𝑥1 𝜕𝑥1

⇒ 2𝑥1 + 𝜆1 − 𝜆2 = 10 … … … (1)
Prepared by Ranbir Kalita

𝜕𝑓 𝜕ℎ1 𝜕ℎ2
− 𝜆1 − 𝜆2 = 0 ⇒ (10 − 2𝑥2 ) − 𝜆1 (1) − 𝜆2 (1) = 0
𝜕𝑥2 𝜕𝑥2 𝜕𝑥2

⇒ 2𝑥2 + 𝜆1 + 𝜆2 = 10 … … … (2)

𝜆1 ℎ1 (𝑥1 , 𝑥2 ) = 0 ⇒ 𝜆1 (𝑥1 + 𝑥2 − 14) = 0 … … … (3)

𝜆2 ℎ2 (𝑥1 , 𝑥2 ) = 0 ⇒ 𝜆1 (−𝑥1 + 𝑥2 − 6) = 0 … … … (4)

Page - 28
Mechanical Engineering Department Jorhat Engineering College
 OR : Module-4

ℎ1 (𝑥1 , 𝑥2 ) ≤ 0 ⇒ 𝑥1 + 𝑥2 − 14 ≤ 0 … … … (5)

ℎ2 (𝑥1 , 𝑥2 ) ≤ 0 ⇒ −𝑥1 + 𝑥2 − 6 ≤ 0 … … … (6)

Prepared by Ranbir Kalita

𝑥1 , 𝑥2 ≥ 0 … … … (7) and 𝜆1 , 𝜆2 ≥ 0 … … … (8)

Case-1: when λ1 = λ2 = 0

Putting λ1 = λ2 = 0 in (1) and (2), we get,

(1) ⇒ 2𝑥1 = 10 and (2) ⇒ 2𝑥2 = 10

⇒ 𝑥1 = 5 and ⇒ 𝑥2 = 5

Now, these values of x1 and x2 must satisfy the inequalities (5), (6) & (7).

(5) ⇒ 𝑥1 + 𝑥2 − 14 = 5 + 5 − 14 = −4 < 0 [satisfied]

Prepared by Ranbir Kalita

(6) ⇒ −𝑥1 + 𝑥2 − 6 = −5 + 5 − 6 = −6 < 0 [satisfied]

(7) ⇒ 𝑥1 = 5 > 0 and 𝑥2 = 5 > 0 [satisfied]

Therefore,

𝑍𝑚𝑎𝑥 = 10𝑥1 + 10𝑥2 − 𝑥1 2 − 𝑥2 2 = (10 × 5) + (10 × 5) − 52 − 52 ⇒ 𝒁𝒎𝒂𝒙 = 𝟓𝟎

Though the solution to the problem stops here, yet, lets try Case-2: λ1 ≠ 0 & λ2 ≠ 0

If λ1 ≠ 0 & λ2 ≠ 0, then from equations (3) & (4), we get,

(3) ⇒ 𝑥1 + 𝑥2 − 14 = 0 … … … (A) and (4) ⇒ −𝑥1 + 𝑥2 − 6 = 0 … … … (B)

Solving equations (1), (2), (A) & (B), we will get the values of x1 , x2 , λ1 & λ2 .
Prepared by Ranbir Kalita

Solving (A) and (B), we get: 𝑥1 = 4 and 𝑥2 = 10

Putting these values of x1 and x2 in equations (1) & (2), we get.

(1) ⇒ (2 × 4) + 𝜆1 − 𝜆2 = 0 ⇒ 𝜆1 − 𝜆2 = 2 … … … (C)

(2) ⇒ (2 × 10) + 𝜆1 + 𝜆2 = 0 ⇒ 𝜆1 + 𝜆2 = −10 … … … (D)

Solving equations (C) and (D), we get: 𝜆1 = −4 and 𝜆2 = −6

Page - 29
Mechanical Engineering Department Jorhat Engineering College
 OR : Module-4

Through the values of x1 and x2 obtained in Case-2 satisfies the inequality (7), yet the values
of λ1 & λ2 do not satisfy inequality (8). Therefore. These set of values can be discarded.

Problem 4.16: Solve the following using KKT conditions:

Prepared by Ranbir Kalita

Max. Z = 2𝑥1 + 3𝑥2 − 2𝑥1 2

Subject to: 𝑥1 + 4𝑥2 ≤ 4 and 𝑥1 + 𝑥2 ≤ 2 Where, 𝑥1 , 𝑥2 ≥ 0

Solution: Given 𝑓(𝑥1 , 𝑥2 ) = 2𝑥1 + 3𝑥2 − 2𝑥1 2

ℎ1 (𝑥1 , 𝑥2 ) = 𝑥1 + 4𝑥2 − 4 and ℎ2 (𝑥1 , 𝑥2 ) = 𝑥1 + 𝑥2 − 2

The KKT conditions are:

𝜕𝑓 𝜕ℎ1 𝜕ℎ2
− 𝜆1 − 𝜆2 = 0 ⇒ (10 − 4𝑥1 ) − 𝜆1 (1) − 𝜆2 (−1) = 0
𝜕𝑥1 𝜕𝑥1 𝜕𝑥1

Prepared by Ranbir Kalita

⇒ 4𝑥1 + 𝜆1 + 𝜆2 = 2 … … … (1)

𝜕𝑓 𝜕ℎ1 𝜕ℎ2
− 𝜆1 − 𝜆2 = 0 ⇒ 3 − 𝜆1 (4) − 𝜆2 (1) = 0
𝜕𝑥2 𝜕𝑥2 𝜕𝑥2

⇒ 4𝜆1 + 𝜆2 = 3 … … … (2)

𝜆1 ℎ1 (𝑥1 , 𝑥2 ) = 0 ⇒ 𝜆1 (𝑥1 + 4𝑥2 − 4) = 0 … … … (3)

𝜆2 ℎ2 (𝑥1 , 𝑥2 ) = 0 ⇒ 𝜆1 (𝑥1 + 𝑥2 − 2) = 0 … … … (4)

ℎ1 (𝑥1 , 𝑥2 ) ≤ 0 ⇒ 𝑥1 + 4𝑥2 − 4 ≤ 0 … … … (5)

ℎ2 (𝑥1 , 𝑥2 ) ≤ 0 ⇒ 𝑥1 + 𝑥2 − 2 ≤ 0 … … … (6)
Prepared by Ranbir Kalita

𝑥1 , 𝑥2 ≥ 0 … … … (7) and 𝜆1 , 𝜆2 ≥ 0 … … … (8)

Case-1: when λ1 = λ2 = 0

Putting λ1 = λ2 = 0 in (1) and (2), we get,

(1) ⇒ 4𝑥1 = 2 and (2) ⇒ 𝑥2 = 0

Page - 30
Mechanical Engineering Department Jorhat Engineering College
 OR : Module-4

1
⇒ 𝑥1 = and ⇒ 𝑥2 = 0
2

Now, these values of x1 and x2 must satisfy the inequalities (5), (6) & (7).

Prepared by Ranbir Kalita

1
(5) ⇒ 𝑥1 + 4𝑥2 − 4 = + (4 × 10) − 4 = 3.5 < 0 [satisfied]
2

1
(6) ⇒ 𝑥1 + 𝑥2 − 2 = + 0 − 2 = −1.5 < 0 [satisfied]
2

1
(7) ⇒ 𝑥1 = > 0 and 𝑥2 = 0 [satisfied]
2

Therefore,

2
1 1 2 𝟏
𝑍𝑚𝑎𝑥 = 2𝑥1 + 3𝑥2 − 2𝑥1 = (2 × ) + (3 × 0) − [2 × ( ) ] ⇒ 𝒁𝒎𝒂𝒙 = 𝟎. 𝟓 =
2 2 𝟐

Prepared by Ranbir Kalita

Though the solution to the problem stops here, yet, lets try Case-2: λ1 ≠ 0 & λ2 ≠ 0

If λ1 ≠ 0 & λ2 ≠ 0, then from equations (3) & (4), we get,

(3) ⇒ 𝑥1 + 4𝑥2 − 4 = 0 … … … (A) and (4) ⇒ 𝑥1 + 𝑥2 − 2 = 0 … … … (B)

Solving equations (1), (2), (A) & (B), we will get the values of x1 , x2 , λ1 & λ2 .

4 2
Solving (A) and (B), we get: 𝑥1 = 3 and 𝑥2 = 3

Putting these values of x1 and x2 in equations (1) & (2), we get.

4 10
(1) ⇒ (4 × ) + 𝜆1 + 𝜆2 = 0 ⇒ 𝜆1 + 𝜆2 = − … … … (C)
3 3
Prepared by Ranbir Kalita

(2) ⇒ 4𝜆1 + 𝜆2 = 3 … … … (D)

19 49
Solving equations (C) and (D), we get: 𝜆1 = and 𝜆2 = −
9 9

Since, λ2 < 0, therefore, inequality (8) is not satisfied. Hence, these set of values obtained
under Case-2 can be discarded.

Page - 31
Mechanical Engineering Department Jorhat Engineering College
 OR : Module-4

Problem 4.17: Solve the following using KKT conditions:

Max. Z = 3𝑥1 2 + 14𝑥1 𝑥2 − 8𝑥2 2

Subject to: 𝑥1 + 2𝑥2 ≤ 24 Where, 𝑥1 , 𝑥2 ≥ 0

Prepared by Ranbir Kalita

Solution: Given 𝑓(𝑥1 , 𝑥2 ) = 3𝑥1 2 + 14𝑥1 𝑥2 − 8𝑥2 2

And ℎ(𝑥1 , 𝑥2 ) = 𝑥1 + 2𝑥2 − 24

Therefore, the KKT conditions are:

𝜕𝑓 𝜕ℎ
−𝜆 = 0 ⇒ 6𝑥1 + 14𝑥2 − 𝜆(1) = 0 ⇒ 6𝑥1 + 14𝑥2 − 𝜆 = 0 … … … (1)
𝜕𝑥1 𝜕𝑥1

𝜕𝑓 𝜕ℎ
−𝜆 = 0 ⇒ 14𝑥1 − 16𝑥2 − 𝜆(2) = 0 ⇒ 14𝑥1 − 16𝑥2 − 2𝜆 = 0 … … … (2)
𝜕𝑥2 𝜕𝑥2

Prepared by Ranbir Kalita

𝜆ℎ(𝑥1 , 𝑥2 ) = 0 ⇒ 𝜆(𝑥1 + 2𝑥2 − 24) = 0 … … … (3)

ℎ(𝑥1 , 𝑥2 ) ≤ 0 ⇒ 𝑥1 + 2𝑥2 − 24 ≤ 0 … … … (4)

𝑥1 , 𝑥2 ≥ 0 … … … (5) and 𝜆 ≥ 0 … … … (6)

Case-1: when λ = 0

Putting λ = 0 in equation (1) & (2):

(1) ⇒ 6𝑥1 + 14𝑥2 = 0 and (2) ⇒ 14𝑥1 − 16𝑥2 = 0

𝟒 𝟕
⇒ 𝒙𝟏 = and ⇒ 𝒙𝟐 =
𝟕𝟑 𝟏𝟒𝟔

Putting these values of x1 and x2 in equation (4), we get,

Prepared by Ranbir Kalita

4 7 𝟏𝟕𝟒𝟏
(4) ⇒ 𝑥1 + 2𝑥2 − 24 = + (2 × ) − 24 = − < 𝟎 [satisfied]
73 146 𝟕𝟑

Therefore,

4 2 4 7 7 2
𝑍𝑚𝑎𝑥 = 3𝑥1 2 + 14𝑥1 𝑥2 − 8𝑥2 2 = [3 × ( ) ] + (14 × × ) − [8 × ( ) ]
73 73 146 146

Page - 32
Mechanical Engineering Department Jorhat Engineering College
 OR : Module-4

𝟐
⇒ 𝒁𝒎𝒂𝒙 =
𝟕𝟑

Case-2: when λ ≠ 0

Prepared by Ranbir Kalita

From equation (3), if λ ≠ 0, then, 𝑥1 + 2𝑥2 − 24 must be equal to zero. Therefore,

𝑥1 + 2𝑥2 − 24 = 0 … … … (7)

Now, solving equations (1) , (2) & (7), we get,

𝑥1 = 22 ; 𝑥2 = 1 ; 𝜆 = 146

Therefore, (4) ⇒ 𝑥1 + 2𝑥2 − 24 = 22 + (2 × 1) − 24 = 0 [satisfied]

Since, these values of x1 , x2 and λ are satisfying the inequalities (4) , (5) & (6), therefore,
these set of values can be accepted. Hence,

Prepared by Ranbir Kalita

𝑍𝑚𝑎𝑥 = 3𝑥1 2 + 14𝑥1 𝑥2 − 8𝑥2 2 = (3 × 222 ) + (14 × 22 × 1) − (8 × 12 )

⇒ 𝒁𝒎𝒂𝒙 = 𝟏𝟕𝟓𝟐

Since, the value of Zmax obtained from Case-1 is less than that obtained from Case-2,
therefore, the final solution to the given problem is:

𝒙𝟏 = 𝟐𝟐 ; 𝒙𝟐 = 𝟏 and 𝒁𝒎𝒂𝒙 = 𝟏𝟕𝟓𝟐

Problem 4.18: Solve the following using KKT conditions:

Min. Z = (𝑥1 − 2)2 + (𝑥2 − 1)2

Subject to: 𝑥1 2 − 𝑥2 2 ≤ 0 and 𝑥1 + 𝑥2 ≤ 2 Where, 𝑥1 , 𝑥2 ≥ 0

Prepared by Ranbir Kalita

Solution: Given 𝑓(𝑥1 , 𝑥2 ) = 𝑥1 2 + 𝑥2 2 − 4𝑥1 − 2𝑥2 + 5

ℎ1 (𝑥1 , 𝑥2 ) = −𝑥1 2 + 𝑥2 and ℎ2 (𝑥1 , 𝑥2 ) = −𝑥1 − 𝑥2 + 2

The KKT conditions are:

𝜕𝑓 𝜕ℎ1 𝜕ℎ2
− 𝜆1 − 𝜆2 = 0 ⇒ (2𝑥1 − 4) − 𝜆1 (−2𝑥1 ) − 𝜆2 (−1) = 0
𝜕𝑥1 𝜕𝑥1 𝜕𝑥1

Page - 33
Mechanical Engineering Department Jorhat Engineering College
 OR : Module-4

⇒ 2𝑥1 + 2𝜆1 𝑥1 + 𝜆2 = 4 … … … (1)

𝜕𝑓 𝜕ℎ1 𝜕ℎ2
− 𝜆1 − 𝜆2 = 0 ⇒ (2𝑥2 − 2) − 𝜆1 (1) − 𝜆2 (−1) = 0
𝜕𝑥2 𝜕𝑥2 𝜕𝑥2

Prepared by Ranbir Kalita

⇒ 2𝑥2 − 𝜆1 + 𝜆2 = 2 … … … (2)

𝜆1 ℎ1 (𝑥1 , 𝑥2 ) = 0 ⇒ 𝜆1 (−𝑥1 2 + 𝑥2 ) = 0 … … … (3)

𝜆2 ℎ2 (𝑥1 , 𝑥2 ) = 0 ⇒ 𝜆2 (−𝑥1 − 𝑥2 + 2) = 0 … … … (4)

ℎ1 (𝑥1 , 𝑥2 ) ≤ 0 ⇒ −𝑥1 2 + 𝑥2 ≥ 0 … … … (5)

ℎ2 (𝑥1 , 𝑥2 ) ≤ 0 ⇒ −𝑥1 − 𝑥2 + 2 ≥ 0 … … … (6)

𝑥1 , 𝑥2 ≥ 0 … … … (7) and 𝜆1 , 𝜆2 ≤ 0 … … … (8)

Prepared by Ranbir Kalita

Case-1: when λ1 = λ2 = 0

Putting λ1 = λ2 = 0 in (1) and (2), we get,

(1) ⇒ 2𝑥1 = 4 and (2) ⇒ 2𝑥2 = 2

⇒ 𝑥1 = 2 and ⇒ 𝑥2 = 1

Now, these values of x1 and x2 must satisfy the inequalities (5), (6) & (7).

(5) ⇒ −𝑥1 2 + 𝑥2 = (−2)2 + 1 = −3 ≯ 0 [not satisfied]

(6) ⇒ −𝑥1 − 𝑥2 + 2 = −2 − 1 + 2 = −1 ≯ 0 [not satisfied]

(7) ⇒ 𝑥1 = 2 > 0 and 𝑥2 = 1 > 0 [satisfied]

Prepared by Ranbir Kalita

Though the values of x1 and x2 satisfies the inequality (7), yet, the inequalities (5) & (6)
remains unsatisfied. Therefore, these set of values must be rejected.

Case-2: λ1 ≠ 0 & λ2 ≠ 0

If λ1 ≠ 0 & λ2 ≠ 0, then from equations (3) & (4), we get,

(3) ⇒ −𝑥1 2 + 𝑥2 = 0 … … … (A) and (4) ⇒ −𝑥1 − 𝑥2 + 2 = 0 … … … (B)

Page - 34
Mechanical Engineering Department Jorhat Engineering College
 OR : Module-4

Solving equations (1), (2), (A) & (B), we will get the values of x1 , x2 , λ1 & λ2 .

Solving (A) and (B), we get:

(𝐵) ⇒ −𝑥2 = 𝑥1 − 2 ⇒ 𝑥2 = 2 − 𝑥1

Prepared by Ranbir Kalita

(𝐴) ⇒ −𝑥1 2 + 𝑥2 = 0 ⇒ −𝑥1 2 + (2 − 𝑥1 ) = 0 ⇒ 𝑥1 2 + 𝑥1 − 2 = 0 ⇒ 𝑥1 = 1 & 𝑥1 = −2

Putting the values of x2 in (B), we get:

𝑥2 = 2 − 1 ⇒ 𝑥2 = 1 and 𝑥2 = 2 − (−2) ⇒ 𝑥2 = 4

Therefore, we have the values of x1 and x2 as: (𝒙𝟏 , 𝒙𝟐 ) = (𝟏 , 𝟏) and (−𝟐 , 𝟒)

Putting these values of x1 and x2 in equations (3) & (4), we have to find the values of λ1 & λ2

When x1 = 1 and x2 = 1:

Prepared by Ranbir Kalita

(3) ⇒ 𝜆1 (−𝑥1 2 + 𝑥2 ) = 0 ⇒ 𝜆1 (−1 + 1) ⇒ 𝜆1 = 0

(4) ⇒ 𝜆2 (−𝑥1 − 𝑥2 + 2) = 0 ⇒ 𝜆2 (−1 − 1 + 2) = 0 ⇒ 𝜆2 = 0

Now, theses set of values, i.e. x1 = 1 , x2 = 1 , λ1 = 0 and λ2 = 0 satisfies the inequalities (5),
(6), (7) & (8). Therefore, these set of values can be accepted.

Therefore, 𝑍𝑚𝑎𝑥 = (𝑥1 − 2)2 + (𝑥2 − 1)2 = (1 − 2)2 + (1 − 1)2 ⇒ 𝒁𝒎𝒂𝒙 = 𝟏

When x1 = –2 and x2 = 4:

(3) ⇒ 𝜆1 (−𝑥1 2 + 𝑥2 ) = 0 ⇒ 𝜆1 [(−2)2 + 4] ⇒ 𝜆1 = 0

(4) ⇒ 𝜆2 (−𝑥1 − 𝑥2 + 2) = 0 ⇒ 𝜆2 [−(−2) − 4 + 2] = 0 ⇒ 𝜆2 = 0

Prepared by Ranbir Kalita

Now, theses set of values, i.e. x1 = –2 , x2 = 4 , λ1 = 0 and λ2 = 0 satisfies the inequalities (5),
(6), (7) & (8). Therefore, these set of values can be accepted.

Therefore, 𝑍𝑚𝑎𝑥 = (𝑥1 − 2)2 + (𝑥2 − 1)2 = [(−2) − 2]2 + (4 − 1)2 ⇒ 𝒁𝒎𝒂𝒙 = 𝟏𝟕

Page - 35
Mechanical Engineering Department Jorhat Engineering College
 OR : Module-4

Case-3: λ1 = 0 & λ2 ≠ 0

Putting λ1 = 0 in equations (1) & (2), we get:

(1) ⇒ 2𝑥1 + 𝜆2 = 4 and (2) ⇒ 2𝑥2 + 𝜆2 = 2

Prepared by Ranbir Kalita

If λ2 ≠ 0, then from equation (4):
−𝑥1 − 𝑥2 + 2 = 0

Solving the above three equations, we get the values of x1 , x2 and λ2 as:

3 1
𝑥1 = ; 𝑥2 = ; 𝜆2 = 1
2 2

Now, these set of values must satisfy the inequalities (5), (6), (7) & (8).

2
3 2 1 7
(5) ⇒ −𝑥1 + 𝑥2 = − ( ) + ( ) = − ≯ 0 [not satisfied]
2 2 4

Prepared by Ranbir Kalita

3 1
(6) ⇒ −𝑥1 − 𝑥2 + 2 = − − + 2 = 0 [satisfied]
2 2

Since, inequality (5) is not satisfied, therefore, these set of values are not acceptable.

Case-4: λ1 ≠ 0 & λ2 = 0

Putting λ2 = 0 in (1) & (2), we get,

(1) ⇒ 2𝑥1 + 2𝜆1 𝑥1 = 4 ⇒ 𝑥1 + 𝜆1 𝑥1 − 2 = 0

(2) ⇒ 2𝑥2 − 𝜆1 = 2 ⇒ 𝜆1 = 2𝑥2 − 2

For equation (3) to be valid: −𝑥1 2 + 𝑥2 = 0 ⇒ 𝑥2 = 𝑥1 2

Prepared by Ranbir Kalita

Put (2) in (1) ⇒ 𝑥1 + (2𝑥2 − 2)𝑥1 − 2 = 0 ⇒ 𝑥1 − 2𝑥1 𝑥2 + 2 = 0

⇒ 𝑥1 − 2𝑥1 (𝑥1 2 ) + 2 = 0 ⇒ −2𝑥1 3 + 𝑥1 + 2 = 0 ⇒ −2𝑥1 3 + 0. 𝑥1 2 + 1. 𝑥1 + 2 = 0

The above equation is of the form ax3 + bx2 + cx + d = c. Solving the above equation, we get,

𝑥1 = 1.165

Page - 36
Mechanical Engineering Department Jorhat Engineering College
 OR : Module-4

Putting this value of x1 in equation (3), we get, 𝑥2 = 𝑥1 2 = 1.1652 ⇒ 𝑥2 = 1.357

Putting the obtained values of x1 and x2 in equation (1), we get,

Prepared by Ranbir Kalita

167
𝑥1 + 𝜆1 𝑥1 − 2 = 0 ⇒ 1.165 + 𝜆1 (1.165) − 2 ⇒ 𝜆1 = = 0.72
233

These values of x1 = 1.165 , x2 = 1.357 , λ1 = 0.72 & λ2 = 0 must satisfy the inequalities (5),
(6), (7) & (8).

(5) ⇒ −𝑥1 2 + 𝑥2 = −1.1652 + 1.357 = 0 [satisfied]

(6) ⇒ −𝑥1 − 𝑥2 + 2 = −1.165 − 1.357 + 2 = −0.522 ≯ 0 [not satisfied]

Therefore, these set of values must be rejected.

Hence, the 𝒁𝒎𝒂𝒙 = 𝟏 for x1 = 1 and x2 = 1 obtained from Case-2.

Prepared by Ranbir Kalita

Prepared by Ranbir Kalita

References:

[1] Hamdy A. Taha (2007), Operations Research: An Introduction (8th edition), Pearson,
ISBN 0-13-188923·0, Chapter 16, pp. 605-640.

[2] Prem Kumar Gupta and D. S. Hira (2014), Operations Research (7th revised edition),
S. Chand, ISBN: 81-219-0281-9, Chapter 16, pp. 1396-1474.

Page - 37
Mechanical Engineering Department Jorhat Engineering College