07 Chapter 3

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 30

Chapter 3

An inventory model with uncertain


demand and lost sales reduction under
service level constraint

3.1 Introduction
Inventory systems are usually common and significant in all production processes,
services and business operations. Every organization relentlessly strive to uphold the
optimum inventory to meet its requirements to avoid over or under inventory that can
impact the financial figures. Inventory is always dynamic which requires a very careful
evaluation and control of external and internal factors through planning and reviews for
timely replenishment. In this framework, continuous review inventory system is the most
suitable mathematical model to deal with such type of problems as well as various types of
uncertainties. Uncertainties in inventory problems may be associated with demand, lead
time or various relevant costs. Here in this chapter, we consider that the quantity received
may not match with quantity ordered due to human errors in counting, rejection during
inspection, electricity failure etc.

28
In modern production management, lead time plays an important role and has been a
topic of interest for many authors in inventory management. In most of the inventory
problems, either deterministic or probabilistic model, lead time is viewed as prescribed
constant or stochastic variable, which therefore, is not subject to control. However in
practice, lead time could be shortened by paying an additional crashing cost. In other
words, it is controllable. In a real environment, to reduce lead time, employers may request
the laborers to stay at work longer than usual, employ part-time workers etc., so a high extra
expense ought to be paid to the employees. By shortening lead time, vendors can lower the
safety stock, reduce loss due to out-of-stock and improve the customer service level. Thus,
in present supply chain and inventory management system, controllable lead time is a key to
business achievement and has attracted considerable research attention.

Also, ordering cost can be reduced in various ways through further investment. For
example, ordering cost can be reduced through providing skill development training to
employees, procedural changes and specialized equipment acquisition. In this inventory
model, the ordering cost is reduced through further investment which is in linear functional
form. In real-time markets, many factors may affect customers willingness of accepting
back orders during the stock-out period. Other than the item themselves the image of a
selling shop is one of the potential factors that may propel the customers intention of back
orders. Customer willingness for back orders can be increased by numerous endeavors, for
example, upgrading the servicing facilities, maintaining a high quality of selling products,
increasing expenditure on advertisement etc., could be made to build a good relationship
with customers and enhance customer’s loyalty. This motivates customers to give
preference for backorders. Normally, an extra-added cost must be spent for these efforts.

3.2 Problem description


This chapter explores an integrated single-vendor single-buyer for a single product in the
context of continuous review inventory system. Shortages are allowed and partially back

29
ordered. Moreover, the received quantity doesn’t match with the ordered quantity, but it is
a random variable following a normal distribution. Additionally, the system manager would
like to reduce the ordering cost and lost sales rate. Therefore he/she invests some money
to reduce the ordering cost and lost sales rate. In realistic environment, when an order is
placed, the products will be delivered after a certain period of time (Lead time). Because of
tough competition, the buyer pay an extra amount to reduce the lead time. We consider that
a service level constraint in place of a shortage cost which infers that the stock-out level per
cycle is bounded and the accessibility of stock in a probabilistic or expected sense. The main
contribution of this chapter is to find an optimal solutions of order size, lead time and lost
sales rate so that the joint expected total cost is minimum.

3.3 Notations and assumptions


To develop the proposed model, the following notations and assumptions are adopted.

3.3.1 Notations

A0 Original ordering cost before any investment is made


A(L) Capital investment needed to attain ordering cost A
Y Quantity received, a random variable
α Bias factor which is the expected amount received ÷ amount ordered
h Annual inventory holding cost per unit per year
I(β̃) Capital investment needed to attain lost sales β̃
θ Fractional opportunity cost of capital
γ Proportion of demands that are not met from stock so (1 − γ) is the service level

3.3.2 Assumptions

1. The system deals with a single-vendor and a single-buyer. The order quantity Q is
placed whenever the inventory level falls to the reorder point r.

30
2. The received quantity Y is a random variable depending on the order quantity Q.

3. If the order quantity is greater than or equal to the expected quantity received then
0 ≤ α < 1. Suppose the order quantity is less than the expected quantity due to human
errors in counting etc., leading to the large amount of inventory, α > 1.

4. The reorder point r= expected demand during lead time + safety stock (SS) and SS = k

(standard deviation of lead time demand).i.e., r = DL + kσ L, where k is the safety
factor.

5. The buyer lead time is controllable and reducible by adding additional crashing cost
which is determined by the length of negative exponential lead time. Hence, the lead
time crashing cost is
(
0 if L = Le
R(L) = −ηL
δe , if Le ≤ L < Ls

where δ, η are constants, Le and Ls represent the existing and shortest lead times,
respectively.

6. During the stock out period, a fraction β of the demand will be back ordered and the
remaining fraction (1 − β) ≡ β̃, will be lost. Also the ordering cost A and the lost sales
β̃ can be reduced by capital investments I(β̃) and A(L) respectively.

3.4 Mathematical model


We have assumed that the lead time demand X follows a normal distribution with mean

DL and standard deviation σ L. We assume that X has a cumulative distribution function f

and the reorder point r = DL + kσ L. If X > r, then shortage occurs. Hence, the expected
shortage at the end of the cycle is
Z∞ √
+
E(X − r) = (X − r)df (x) = σ Lψ(k)
r

31
where ψ(k) = φ(k) − k[1 − Φ(k)] > 0 and φ, Φ are the standard normal probability density
function and cumulative distribution function respectively. Thus the expected number of
back orders is βE(X − r)+ and the expected lost sale is (1 − β)E(X − r)+ . Hence the stock out
cost per cycle is [π+(1−β)π0 ]. The expected net inventory level just before the order arrives
is [r − DL + (1 − β)E(X − r)+ ] and the expected net inventory level at the beginning of the
cycle is [Q +r −DL+(1−β)E(X −r)+ ]. Therefore, the expected annual holding cost is h[ Q2 +
r − DL + (1 − β)E(X − r)+ ] and we consider the crashing cost R(L) is a negative exponential
function of lead time. Consequently, the joint expected total cost per cycle comprised of
ordering cost, holding cost, stock out cost and lead time crashing cost is expressed by,
√ √
" #
Q
EAC(Q, L) =A + h + kσ L + (1 − β)σ Lψ(k)
2

+(π + (1 − β)π0 )σ Lψ(k) + R(L) (3.1)

Subsequently, this chapter examines the amount received is uncertain, that is, if a quantity
Q is ordered each time, the quantity received will be Y units which is a random variable
with E(Y /Q) = αQ and the variance of the quantity received after amount Q is ordered and
is given by var(Y /Q) = σ0 2 + σ1 2 Q2 , where α, σ0 2 and σ1 2 are non-negative constants. If
σ1 2 = 0 then the standard deviation of the quantity received is independent of the quantity
ordered, and if σ0 2 = 0 then the standard deviation of the quantity received is proportional
to the quantity ordered.
Y
Let C(Y , L) be the current total cost per cycle and D be the time duration of this cycle. Under
the hypothesis, the total cost per cycle with a variable lead time can be obtained as equation
3.1 given that Y units are received is,
√ √
" #
Y Y
C(Y , L) =A + h + kσ L + (1 − β)σ Lψ(k)
D 2

+(π + (1 − β)π0 )σ Lψ(k) + R(L) (3.2)
 
Y Y
Using the value of E Q = αQ and var( Q ) = σ02 + (σ12 + α 2 )Q2 (see Appendix A), we obtain
the expected total cost per cycle with variable lead time for the uncertain received quantity

32
system is,

αQ √ √ h 2
E[C(Y , L)|Q] =A + h [kσ L + (1 − β)σ Lψ(k)] + [σ + (σ12 + α 2 )Q2 ]
D 2D 0

+(π + (1 − β)π0 )σ Lψ(k) + R(L) (3.3)

Moreover, the expected cycle time is

E(Y |Q) αQ
= (3.4)
D D

By using this equations 3.3 and 3.4, we obtain the joint expected total cost with controllable
lead time denoted by JET C(Q, L) for the uncertain received quantity system can be
articulated by,

AD √ √ h
JET C(Q, L) = + h[kσ L + (1 − β)σ Lψ(k)] + [σ 2 + (σ12 + α 2 )Q2 ]
αQ 2αQ 0
D(π + (1 − β)π0 ) √ R(L)D
+ σ Lψ(k) + (3.5)
αQ αQ

3.4.1 Investment on ordering cost and lost sales rate

Building upon equation 3.5, we desire to study the effects of investment on ordering
cost and lost sales rate. The above model is further evaluated by considering two types
of investment functions. We assume that the lead time and ordering cost reductions act
dependently with the following relationship ( Chen et al. [31])

L0 − L A −A
= α∗ 0
L0 A0

where α ∗ (> 0) is a constant scaling parameter to describe the linear relationship between
percentages of reductions in lead time and ordering cost.
The ordering cost A can be written as a linear function of L, that is given by,

A(L) = u + vL (3.6)

A0
where u = (1 − α1∗ )A0 and v = α ∗ L0 .
We examine the case where the capital investment I(β̃) for lost sales rate reduction is

33
described by the logarithmic function such as:

β˜0
!
I(β̃) = c1 ln f or 0 < β̃ ≤ β˜0 , (3.7)
β̃

where β˜0 is the original fraction of the excess demands that will be lost (before extra
investment is made), c1 = 1/δ1 and δ1 = the percentage decrease in β̃ per dollar increase in
I(β̃).

Now, our problem is to minimize the investment in ordering cost, lost sales reduction,
and the inventory relevant costs (as expressed in 3.5) by simultaneously optimizing Q, β̃
and L constrained on 0 < β̃ ≤ β˜0 . That is, the objective of our problem is to minimize the
following joint expected total cost:

A(L)D √ √ h
JET C(Q, β̃, L) = + h[kσ L + (1 − β)σ Lψ(k)] + [σ 2 + (σ12 + α 2 )Q2 ]
αQ 2αQ 0
D(π + (1 − β)π0 ) √ R(L)D
+ σ Lψ(k) + + I(β̃) (3.8)
αQ αQ

for 0 < β̃ ≤ β˜0 .


The problem can be written as,

β˜0 √ √
!
(u + vL)D
MinJET C(Q, β̃, L) =θc1 ln + + h[kσ L + (1 − β)σ Lψ(k)]
β̃ αQ
h D(π + (1 − β)π0 ) √
+ [σ02 + (σ12 + α 2 )Q2 ] + σ Lψ(k)
2αQ αQ
R(L)D
+ , L ∈ [Le , Ls ] (3.9)
αQ

subject to 0 < β̃ ≤ β˜0 , where θ is the annual fractional cost of capital investment (e.g.,
interest rate).

3.4.2 Buyer’s service level constraint

To avoid the imprecision and inefficiency of estimation, the penalty costs associated
with a shortage, several researchers use an SLC in their models. SLC is a constraint on the
system that defines at least an assumed proportion of demands should be met from on hand

34
inventory. The proportion of demand which exceeds reorder point, at the end of each
production cycle, should not exceed the desired value of γ. So the service-level constraint
can be written as:

The expected demand shortages at the end of cycle for a given safety factor
≤γ
Quantity available for satisfying the demand per cycle
That is,

E(X − r)+
≤γ
Q

Hence, the lead-time demand follows a normal distribution. Therefore SLC can be written
as,

σ Lψ(k)
≤ γ, L ∈ [Le , Ls ] (3.10)
Q

Thus the joint expected total cost is,

β˜0 √ √
!
(u + vL)D
MinJET C(Q, β̃, L) =θc1 ln + + h[kσ L + (1 − β)σ Lψ(k)]
β̃ αQ
h D(π + (1 − β)π0 ) √
+ [σ02 + (σ12 + α 2 )Q2 ] + σ Lψ(k)
2αQ αQ
R(L)D
+ , L ∈ [Le , Ls ] (3.11)
αQ

σ Lψ(k)
subject to Q ≤ γ, 0 < β̃ ≤ β˜0 .

3.4.3 Solution procedure

The problem expressed in the previous section occurs as constrained non-linear program.
To solve this kind of non-linear problem, we pursue the similar procedure of most of the
literature dealing with nonlinear problem. That is, first we temporarily ignore the constraint
SLC, 0 < β̃ ≤ β˜0 , then determine the optimum solutions of Q, β̃ and L ∈ [Le , Ls ] which
minimizes the joint expected total cost, JET C(Q, β̃, L).

Lemma 3.1. For fixed Q and β̃, JET C(Q, β̃, L) is a concave function of L ∈ [Le , Ls ].
Proof. Taking the first and second order partial derivatives of JET C(Q, β̃, L) with respect to

35
L, we have
ηδDe−ηL
" #
∂JET C(Q, β̃, L) vD σh Dσ ψ(k)
= + √ k + β̃ψ(k) + √ (π + π0 β̃) −
∂L αQ 2 L 2Qα L αQ
∂2 JET C(Q, β̃, L)
" #
σh Dσ ψ(k)
2
= − √ k + β̃ψ(k) − √ (π + π0 β̃)
∂L 4L L 4QLα L
η 2 δDe−ηL
+ <0
αQ
Therefore, for fixed Q and β̃, JET C(Q, β̃, L) is concave in L ∈ [Le , Ls ], then the minimum
joint expected total cost occurs at the end points of the interval [Le , Ls ]. Otherwise, the
minimum joint expected total cost occurs at any point of the interval [Le , Ls ].

For a fixed L ∈ [Le , Ls ], let us take the first order partial derivatives of JET C(Q, β̃, L) with
respect to Q and β̃, we obtain
∂JET C(Q, β̃, L) (u + vL)D hσ02 h 2
=− − + (σ + α 2 )
∂Q αQ 2 2αQ 2 2α 1
D(π + π0 β̃) √ Dδe−ηL
− σ Lψ(k) − (3.12)
αQ2 αQ

∂JET C(Q, β̃, L) θc √ Dπ0 σ Lψ(k)
= − 1 + hσ Lψ(k) + (3.13)
∂β̃ β̃ αQ
respectively.
Now, for fixed L ∈ [Le , Ls ], the values of Q and β̃ are obtained by equating the equations
3.12 and 3.13 to zero as given by,
#  12

 "
2
 σ 0 h 



 2D A + 2D + (π + β̃π 0 )σ Lψ(k) + R(L) 



 
Q = (3.14)
 
2 2




 h(σ1 + α ) 




 

θc1 αQ
β̃ = √ (3.15)
σ Lψ(k)[hαQ + Dπ0 ]
Theoretically, for fixed L ∈ [Le , Ls ], by solving the above equations 3.14 and 3.15 iterating
until convergence, the values of Q and β̃ can be obtained (these values are denoted by Q∗
and β̃ ∗ respectively). From the following proportion, it can be shown that for fixed
L ∈ [Le , Ls ], the joint expected total cost has a minimum at the point (Q∗ , β̃ ∗ ).

36
Proposition 3.1. For fixed L ∈ [Le , Ls ], the Hessian matrix for JET C(Q, β̃, L) is positive
definite at point (Q∗ , β̃ ∗ ).
Proof: Refer Appendix B.

Now we consider the constraint 0 < β̃ ≤ β˜0 . From equation 3.15, we obtain that β̃ ∗ is
positive as c1 , α, h, π0 , D, θ, L and ψ(k) are positive. For a given L ∈ [Le , Ls ], if β̃ ∗ < β̃,
then (Q∗ , β̃ ∗ ) is an interior optimal solution. However if β̃ ∗ ≥ β̃, then the decision maker
may decide against offering the backorder price discount, as the decision maker builds
decisions in order to get profits. In this case, the decision maker may decide against offering
the price discount, that is β̃ ∗ = β̃ .
Now, the SLC (as expressed in equation 3.10) is taken into consideration and if

( ασ ) Lψ(k) ≤ Qi∗ for Q = Qi∗ , then Qi∗ is the local minimum of JET C(Q, β̃, L) and SLC is

inactive. Otherwise optimal value of Q should be at least equal to Qs = ( ασ ) Lψ(k) which
is greater than Qi∗ so that specified level of service can be achieved at minimum joint
expected total cost and Qs is the local minimum of JET C(Q, β̃, L). Therefore, for given
L ∈ [Le , Ls ], the optimal Q = Qo is given by max{Qi∗ , Qs }.
We cannot obtain the explicit general solution for Q and β̃ by solving equations 3.14 and
3.15, because the evaluation of each of the expressions needs knowledge of the value of the
other. However, we can obtain the optimal value of Q and β̃ by using an iterative procedure.
It can be easily proved the convergence of the procedure by adopting a similar graphical
technique used in Hadley and Whitin [84]. The following algorithm 3.4.4 is developed to
find the optimal values for the order quantity, lost sales and lead time. The flowchart of the
algorithm is also illustrated in figures 3.1, 3.2 and 3.3.

3.4.4 Algorithm

Step 1 For each L ∈ [Le , Ls ], determine the JET C(Qoi , β̃i , L) using the sub-algorithm.

Step 2 Find Min JET C(Qoi , β̃i , L) for every L ∈ [Le , Ls ].

Step 3 Let JET C(Q∗ , β̃ ∗ , L) = MinJET C(Qoi , β̃i∗ , L) for every L ∈ [Le , Ls ], then

37
JET C(Q∗ , β̃ ∗ , L∗ ) is the minimum joint expected total cost of the proposed model and
(Q∗ , β̃ ∗ , L∗ ) is the optimum solution.

Sub-Algorithm

Step 1 For given L ∈ [Le , Ls ], repeat steps (1.1)-(1.3) until no change occurs in the values of
Qi and β̃i . Denote the solution by (Qi∗ , β˜i∗ )

Step 1.1 Start with β˜i1 = β˜0 .

Step 1.2 Substituting β˜i1 into equation 3.14 evaluates Q1 .

Step 1.3 Utilizing Q1 determines β˜i2 from equation 3.15.

Step 2 Compare β˜i∗ with β˜0 .

Step 2.1 If β˜i∗ < β˜0 , then go to step (3).

Step 2.2 If β˜i∗ ≥ β˜0 , then for this given L ∈ [Le , Ls ], let β˜i∗ = β˜0 , and the corresponding
Qi∗ can be obtained by substituting β˜0 into equation 3.14 then go to step (3).

Step 3 Set Qoi = max{Qi∗ , (σ /α) Lψ(k)}, for every L ∈ [Le , Ls ] then determine the new β˜∗∗
by putting Q = Qoi in equation 3.15 and perform steps (3.1) and (3.2).

Step 3.1 If β˜i∗∗ < β˜0 , then go to step (4).

Step 3.2 If β˜i∗∗ > β˜0 , then for this given L ∈ [Le , Ls ], let β˜i∗∗ = β˜0 , then go to step (4).

Step 4 Compute the corresponding JET C(Qoi , β̃i , L) by putting Q = Qoi and β̃ = β˜i∗∗ in
equation 3.11.

3.5 Numerical example


In order to illustrate the preceding solution procedure, we consider an inventory system
with the following data: D=600 units/year, h=$ 25/unit/year, π=$ 25/unit/year, π0 =$
25/unit/year, σ =7 units/week, σ02 =30, σ12 =0.2, α ∗ =1.25, β=0.3, β˜0 =(1 − β)=0.7, θ=0.1,

38
Start

Set β˜i1 = β˜0


calculate Q1 using Eq. (3.14)

Calculate β˜ic and then Qc for each


L ∈ [Le , Ls ] using Eqn. (3.14),(3.15)

Qc = Q1 , No
β˜ic = β˜i1
Yes

Set (Q1 , β˜i1 )



= (Qi∗ , β̃i )

β˜i∗ ≥ β˜0

set β̃i = β˜0 calculate


β˜i∗ < β˜0 corresponding Qi∗
using Eq. (3.14)

Figure 3.1: Flowchart of algorithm

39
Process A
Start

Set Q = Qoi , where



Qoi = maxQi∗ , (σ /α) Lψ(k)
evaluate β˜∗∗ using Qoi in Eqn. (3.14)

β˜i∗∗ < β˜0 β˜i∗∗ ≥ β˜0

Set β˜i∗∗ = β˜0

Set Q = Qoi and β̃ = β˜i∗∗


calculate JET C(Qoi , β̃i , L)
using Eqn. (3.11)

Figure 3.2: Process A of the flowchart of the algorithm

40
Process B
Start

Find JET C(Q∗ , β˜∗ , L) =


Min JET C(Qoi , β˜∗ , L)

Set JET C(Q∗ , β˜∗ , L∗ ) is the


optimal solution

Stop

Figure 3.3: Process B of the flowchart of the algorithm

δ=115, η=1, α=0.9, γ=0.025, c1 =5. We solve the cases for which the safety factor k= 0.40,
0.45, 0.50, 0.55 and 0.60. For the given five values of k, we use the proposed algorithm to
find the optimal solution of the model which is given in table 3.1 and the optimal solutions
are summarized in table 3.2. Moreover, we list the optimal results of the no investment
policy in the same table to exhibit the goods of investing in ordering cost and lost sales
reduction. The graphical representation of joint expected total cost, JET C(Q, β̃, L) with
respect to Q and L for various k values is given in figure 3.4.

3.6 Sensitivity analysis


We now study the effects of parameters h and c1 on the optimal replenishment size Q∗ ,
lost sales rate β̃ ∗ , lead time L and joint expected total cost JET C(Q∗ , β˜∗ , L) to further illustrate

41
Table 3.1: The optimal solutions for descriptive example and comparisons

Ordering cost and lost sales reduction Fixed ordering cost and lost sales
k (proposed model) (A = A0 = 200 & βe = βe0 = 0.7) Savings
Q ∗ L A∗ βe∗ JET C(.) Q∗ L∗ JET C(.) (%)
80 1 $21 0.0013 $1954 132 1 $3339 41.48
98 2 $90 0.0011 $2416 132 2 $3389 28.71
0.40 112 3 $131 0.0010 $2774 135 3 $3481 20.31
129 4 $160 0.0010 $3032 137 4 $3582 15.35
144 5 $182 0.00098 $3247 140 5 $3647 11.72
158 6 $200 0.00097 $3436 158 6 $3793 09.41
79 1 $21 0.0014 $1936 130 1 $3320 41.68
97 2 $90 0.0012 $2434 131 2 $3362 27.60
0.45 108 3 $131 0.0011 $2753 133 3 $3449 20.18
120 4 $160 0.0010 $2997 136 4 $3545 15.46
134 5 $182 0.0010 $3202 138 5 $3638 11.98
147 6 $200 0.0009 $3380 147 6 $3731 09.41
79 1 $21 0.0015 $1921 129 1 $3302 41.82
96 2 $90 0.0013 $2418 130 2 $3336 37.96
0.50 107 3 $131 0.0012 $2737 131 3 $3418 19.92
115 4 $160 0.0011 $2977 134 4 $3511 15.21
124 5 $182 0.0010 $3169 136 5 $3601 11.99
136 6 $200 0.0009 $3336 138 6 $3684 09.45
78 1 $21 0.0016 $1905 128 1 $3286 42.03
95 2 $90 0.0014 $2403 129 2 $3312 27.45
0.55 106 3 $131 0.0012 $2722 130 3 $3389 19.68
114 4 $160 0.0011 $2963 132 4 $3479 14.83
120 5 $182 0.0011 $3152 134 5 $3566 11.61
125 6 $200 0.0010 $3307 136 6 $3646 09.30
77 1 $21 0.0018 $1891 128 1 $3270 42.17
94 2 $90 0.0015 $2389 127 2 $3290 27.39
0.60 104 3 $131 0.0013 $2708 128 3 $3363 19.48
113 4 $160 0.0012 $2950 130 4 $3449 14.47
119 5 $182 0.0011 $3140 132 5 $3533 11.12
124 6 $200 0.0011 $3295 134 6 $3611 09.59

the model and to gain some managerial insights. The sensitivity analysis of h and c1 is
performed for the different values of k = 0.40, 0.45, 0.50, 0.55, 0.60. The computational
results are shown in figures 3.5 to 3.12. We obtain the following managerial phenomena:

42
Table 3.2: Summary of the optimal solutions and comparisons

Ordering cost and lost sales reduction Fixed ordering cost and lost sales
k (proposed model) (A = A0 = 200 & βe = βe0 = 0.7) Savings
Q L A∗
∗ βe∗ EAC(.) Q∗ L∗ EAC(.) (%)
0.40 80 1 $21 0.0013 $1954 132 1 $3339 41.48
0.45 79 1 $21 0.0014 $1936 130 1 $3320 41.68
0.50 79 1 $21 0.0015 $1921 129 1 $3302 41.82
0.55 78 1 $21 0.0016 $1905 128 1 $3286 42.03
0.60 77 1 $21 0.0018 $1891 128 1 $3270 42.17

k=0.40
k=0.45
Joint Expected Total Cost (JETC)

3500 k=0.50
k=0.55
3000 k=0.60

2500

2000

1500
160

140

120

100
6
5
80 4
Order quantity (Q) 3
60 2
1 Lead time(L)

Figure 3.4: Graphical representation of the optimal solution

(i) From tables 3.3 to 3.6, for k=0.40, 0.45, 0.50, 0.55 and 0.55, it is interesting to note
that the order quantity Q, lead time L and joint expected total cost JET C(Q∗ , β˜∗ , L∗ )
increases and the lost sales β̃ decreases when the investment cost function c1 increases.
This result is expected because this fact occurs in the real life. Since, if the investment
cost function increases then naturally the total cost function increases.

(ii) From tables 3.7 to 3.10, for k=0.40, 0.45, 0.50, 0.55 and 0.55, it is interesting to note
that the order quantity Q, lead time L and joint expected total cost JET C(Q∗ , β˜∗ , L∗ )
increases and the lost sales β̃ decreases when the holding cost h increases. This result is

43
expected because higher holding cost may amplify total cost in the real marketing and
the high level safety factor may hold large amount of quantity in inventory, therefore,
the expected cost automatically increases.

k=0.40
Joint Expected Total Cost (JETC)

k=0.45
3500 k=0.50
k=0.55
3000 k=0.60

2500

2000

1500
160

140

120

100
6
80 5
4
Order quantity (Q) 3
60 2
1 Lead time(L)

Figure 3.5: Graphical representation of the optimal solution when c1 =10.

3.7 Conclusion
In this chapter, a continuous review inventory model is formulated for a single product
involving uncertain received quantity. Further, we have considered the lost sales and
ordering cost investment functional forms as well as negative exponential crashing cost
function with a variable lead time to accommodate more practical features of the real-time
inventory management. The purpose of this chapter was to afford a strategic
decision-making in the supply chain to achieve the target SLC of the buyer at minimum
joint expected total cost. An iterative procedure is presented to find the optimal values of
the order quantity, lost sales and lead time. A numerical example has been given to validate
the efficiency of the proposed model.

44
Table 3.3: Effects of c1 = 10 on optimal solution

Ordering cost and lost sales reduction Fixed ordering cost and lost sales
k (proposed model) (A = A0 = 200 & βe = βe0 = 0.7) Savings
Q ∗ L A∗ βe∗ JET C(.) Q∗ L∗ JET C(.) (%)
79 1 $21 0.0027 $1957 130 1 $3339 41.38
96 2 $90 0.0022 $2455 130 2 $3389 27.56
0.40 107 3 $131 0.0021 $2777 131 3 $3481 20.22
115 4 $160 0.0020 $3035 134 4 $3582 15.27
124 5 $182 0.0020 $3250 136 5 $3678 11.64
136 6 $200 0.0019 $3439 138 6 $3793 09.33
79 1 $21 0.0028 $1939 130 1 $3320 41.60
97 2 $90 0.0024 $2437 131 2 $3362 27.51
0.45 108 3 $131 0.0022 $2756 133 3 $3449 20.09
120 4 $160 0.0020 $2999 136 4 $3545 15.40
134 5 $182 0.0020 $3205 138 5 $3638 11.90
147 6 $200 0.0020 $3383 147 6 $3731 09.33
79 1 $21 0.0031 $1924 129 1 $3302 41.73
97 2 $90 0.0026 $2421 130 2 $3336 27.43
0.50 108 3 $131 0.0023 $2740 131 3 $3418 19.84
120 4 $160 0.0021 $2980 134 4 $3511 15.12
134 5 $182 0.0020 $3172 136 5 $3601 11.91
147 6 $200 0.0020 $3339 138 6 $3684 09.39
78 1 $21 0.0033 $1908 128 1 $3286 41.94
95 2 $90 0.0028 $2406 129 2 $3312 27.36
0.55 106 3 $131 0.0025 $2725 130 3 $3389 19.59
114 4 $160 0.0023 $2965 132 4 $3479 14.77
120 5 $182 0.0021 $3155 134 5 $3566 11.53
125 6 $200 0.0020 $3310 136 6 $3646 09.22
77 1 $21 0.0035 $1894 128 1 $3270 42.07
94 2 $90 0.0030 $2392 127 2 $3290 27.30
0.60 104 3 $131 0.0026 $2711 128 3 $3363 19.39
113 4 $160 0.0025 $2953 130 4 $3449 14.38
119 5 $182 0.0023 $3143 132 5 $3533 11.04
124 6 $200 0.0022 $3298 134 6 $3611 08.68

45
Table 3.4: Effects of c1 = 15 on optimal solution

Ordering cost and lost sales reduction Fixed ordering cost and lost sales
k (proposed model) (A = A0 = 200 & βe = βe0 = 0.7) Savings
Q ∗ L A∗ βe∗ JET C(.) Q∗ L∗ JET C(.) (%)
80 1 $21 0.0040 $1959 132 1 $3339 41.33
98 2 $90 0.0034 $2458 132 2 $3389 27.47
0.40 112 3 $131 0.0031 $2780 135 3 $3481 20.14
129 4 $160 0.0030 $3038 137 4 $3582 15.19
144 5 $182 0.0030 $3253 140 5 $3647 10.80
158 6 $200 0.0029 $3442 158 6 $3793 09.25
79 1 $21 0.0043 $1941 130 1 $3320 41.57
97 2 $90 0.0036 $2440 131 2 $3362 27.42
0.45 108 3 $131 0.0032 $2759 133 3 $3449 20.05
120 4 $160 0.0031 $3003 136 4 $3545 15.29
134 5 $182 0.0030 $3208 138 5 $3638 11.82
147 6 $200 0.0030 $3386 147 6 $3731 09.25
79 1 $21 0.0046 $1927 129 1 $3302 41.64
96 2 $90 0.0039 $2424 130 2 $3336 27.34
0.50 107 3 $131 0.0035 $2743 131 3 $3418 19.75
115 4 $160 0.0032 $2983 134 4 $3511 17.60
124 5 $182 0.0030 $3175 136 5 $3601 11.83
136 6 $200 0.0030 $3342 138 6 $3684 09.28
78 1 $21 0.0049 $1911 128 1 $3286 41.84
95 2 $90 0.0041 $2408 129 2 $3312 27.29
0.55 106 3 $131 0.0037 $2727 130 3 $3389 19.53
114 4 $160 0.0034 $2968 132 4 $3479 14.69
120 5 $182 0.0032 $3158 134 5 $3566 11.44
125 6 $200 0.0030 $3313 136 6 $3646 09.13
77 1 $21 0.0053 $1896 128 1 $3270 42.02
94 2 $90 0.0044 $2394 127 2 $3290 27.23
0.60 104 3 $131 0.0040 $2714 128 3 $3363 19.30
113 4 $160 0.0037 $2955 130 4 $3449 14.32
119 5 $182 0.0034 $3145 132 5 $3533 10.98
124 6 $200 0.0033 $3301 134 6 $3611 08.58

46
Table 3.5: Effects of c1 = 20 on optimal solution

Ordering cost and lost sales reduction Fixed ordering cost and lost sales
k (proposed model) (A = A0 = 200 & βe = βe0 = 0.7) Savings
Q ∗ L A∗ βe∗ JET C(.) Q∗ L∗ JET C(.) (%)
80 1 $21 0.0053 $1963 132 1 $3339 41.21
98 2 $90 0.0045 $2460 132 2 $3389 27.41
0.40 112 3 $131 0.0041 $2782 135 3 $3481 20.08
129 4 $160 0.0040 $3041 137 4 $3582 15.10
144 5 $182 0.0039 $3256 140 5 $3647 10.72
158 6 $200 0.0039 $3444 158 6 $3793 09.20
79 1 $21 0.0057 $1944 130 1 $3320 41.45
97 2 $90 0.0048 $2442 131 2 $3362 27.36
0.45 108 3 $131 0.0043 $2761 133 3 $3449 19.95
120 4 $160 0.0041 $3006 136 4 $3545 15.20
134 5 $182 0.0040 $3210 138 5 $3638 11.76
147 6 $200 0.0039 $3388 147 6 $3731 09.19
79 1 $21 0.0061 $1928 129 1 $3302 41.61
96 2 $90 0.0051 $2426 130 2 $3336 27.28
0.50 107 3 $131 0.0046 $2745 131 3 $3418 19.69
115 4 $160 0.0042 $2985 134 4 $3511 14.98
124 5 $182 0.0041 $3177 136 5 $3601 11.77
136 6 $200 0.0040 $3345 138 6 $3684 09.20
78 1 $21 0.0066 $1912 128 1 $3286 41.81
95 2 $90 0.0055 $2411 129 2 $3312 27.20
0.55 106 3 $131 0.0050 $2730 130 3 $3389 19.45
114 4 $160 0.0046 $2971 132 4 $3479 14.60
120 5 $182 0.0043 $3160 134 5 $3566 11.39
125 6 $200 0.0043 $3334 136 6 $3646 08.56
77 1 $21 0.0070 $1898 128 1 $3270 41.96
94 2 $90 0.0059 $2397 127 2 $3290 27.14
0.60 104 3 $131 0.0053 $2717 128 3 $3363 19.21
113 4 $160 0.0049 $2958 130 4 $3449 14.24
119 5 $182 0.0046 $3148 132 5 $3533 10.90
124 6 $200 0.0043 $3305 134 6 $3611 08.47

47
Table 3.6: Effects of c1 = 25 on optimal solution

Ordering cost and lost sales reduction Fixed ordering cost and lost sales
k (proposed model) (A = A0 = 200 & βe = βe0 = 0.7) Savings
Q ∗ L A∗ βe∗ JET C(.) Q∗ L∗ JET C(.) (%)
80 1 $21 0.0066 $1965 132 1 $3339 41.15
98 2 $90 0.0056 $2462 132 2 $3389 27.35
0.40 112 3 $131 0.0051 $2785 135 3 $3481 19.99
129 4 $160 0.0050 $3043 137 4 $3582 15.05
144 5 $182 0.0049 $3258 140 5 $3647 10.67
158 6 $200 0.0048 $3447 158 6 $3793 09.12
79 1 $21 0.0072 $1947 130 1 $3320 41.36
97 2 $90 0.0060 $2445 131 2 $3362 27.28
0.45 108 3 $131 0.0054 $2764 133 3 $3449 19.86
120 4 $160 0.0051 $3008 136 4 $3545 15.15
134 5 $182 0.0050 $3213 138 5 $3638 11.68
147 6 $200 0.0049 $3391 147 6 $3731 09.11
79 1 $21 0.0076 $1931 129 1 $3302 41.52
96 2 $90 0.0064 $2428 130 2 $3336 27.22
0.50 107 3 $131 0.0058 $2748 131 3 $3418 19.60
115 4 $160 0.0053 $2988 134 4 $3511 14.90
124 5 $182 0.0051 $3180 136 5 $3601 11.69
136 6 $200 0.0050 $3347 138 6 $3684 09.15
78 1 $21 0.0082 $1916 128 1 $3286 41.69
95 2 $90 0.0069 $2413 129 2 $3312 27.14
0.55 106 3 $131 0.0062 $2732 130 3 $3389 19.39
114 4 $160 0.0057 $2973 132 4 $3479 14.54
120 5 $182 0.0053 $3163 134 5 $3566 11.30
125 6 $200 0.0030 $3318 136 6 $3646 08.90
77 1 $21 0.0088 $1900 128 1 $3270 41.90
94 2 $90 0.0074 $2399 127 2 $3290 27.08
0.60 104 3 $131 0.0067 $2719 128 3 $3363 19.15
113 4 $160 0.0061 $2961 130 4 $3449 14.15
119 5 $182 0.0057 $3150 132 5 $3533 10.84
124 6 $200 0.0054 $3306 134 6 $3611 08.45

48
Table 3.7: Effects of h= 30 on optimal solution

Ordering cost and lost sales reduction Fixed ordering cost and lost sales
k (proposed model) (A = A0 = 200 & βe = βe0 = 0.7) Savings
Q ∗ L A∗ βe∗ JET C(.) Q∗ L∗ JET C(.) (%)
75 1 $21 0.0012 $2152 121 1 $3670 41.36
90 2 $90 0.0010 $2698 121 2 $3728 27.63
0.40 112 3 $131 0.0010 $3082 124 3 $3833 19.59
129 4 $160 0.00097 $3387 129 4 $3949 14.23
144 5 $182 0.00095 $3643 144 5 $4090 10.93
158 6 $200 0.00093 $3870 158 6 $4242 08.77
74 1 $21 0.0013 $2133 120 1 $3649 41.55
89 2 $90 0.0011 $2680 120 2 $3699 27.55
0.45 104 3 $131 0.0010 $3041 122 3 $3798 19.93
120 4 $160 0.00098 $3333 124 4 $3907 14.69
134 5 $182 0.00096 $3576 134 5 $4023 11.11
147 6 $200 0.00095 $3791 147 6 $4156 08.78
73 1 $21 0.0014 $2116 119 1 $3631 41.72
88 2 $90 0.0012 $2664 119 2 $3673 27.47
0.50 98 3 $131 0.0010 $3015 121 3 $3766 19.94
111 4 $160 0.0010 $3291 123 4 $3871 14.98
124 5 $182 0.00098 $3522 125 5 $3972 11.33
136 6 $200 0.00096 $3724 127 6 $4067 08.43
72 1 $21 0.0015 $2100 118 1 $3613 41.88
87 2 $90 0.0012 $2648 118 2 $3647 27.39
0.55 97 3 $131 0.0011 $3000 119 3 $3736 19.70
105 4 $160 0.0010 $3267 121 4 $3837 14.86
114 5 $182 0.00099 $3483 123 5 $3935 11.49
125 6 $200 0.00098 $3671 125 6 $4027 08.84
71 1 $21 0.0016 $2086 118 1 $3597 42.01
86 2 $90 0.0013 $2635 117 2 $3624 27.29
0.60 96 3 $131 0.0012 $2988 118 3 $3708 19.42
104 4 $160 0.0011 $3255 120 4 $3805 14.45
109 5 $182 0.0010 $3465 121 5 $3901 11.18
116 6 $200 0.00099 $3641 123 6 $3991 08.77

49
Table 3.8: Effects of h= 35 on optimal solution

Ordering cost and lost sales reduction Fixed ordering cost and lost sales
k (proposed model) (A = A0 = 200 & βe = βe0 = 0.7) Savings
Q ∗ L A∗ βe∗ JET C(.) Q∗ L∗ JET C(.) (%)
70 1 $21 0.0011 $2336 113 1 $3976 41.25
91 2 $90 0.0010 $2950 113 2 $4043 27.03
0.40 112 3 $131 0.00097 $3390 115 3 $4159 18.49
129 4 $160 0.00094 $3742 129 4 $4315 14.23
144 5 $182 0.00092 $4039 144 5 $4499 10.93
158 6 $200 0.00090 $4304 158 6 $4690 08.77
69 1 $21 0.0012 $2317 112 1 $3954 41.40
85 2 $90 0.0010 $2913 112 2 $4013 27.41
0.45 104 3 $131 0.00098 $3339 114 3 $4123 19.02
120 4 $160 0.00096 $3669 120 4 $4251 13.69
134 5 $182 0.00094 $3951 134 5 $4410 10.41
147 6 $200 0.00092 $4201 147 6 $4579 08.26
68 1 $21 0.0013 $2298 111 1 $3935 41.60
83 2 $90 0.0011 $2894 111 2 $3985 27.38
0.50 95 3 $131 0.00099 $3283 112 3 $4090 19.73
111 4 $160 0.00097 $3608 114 4 $4206 14.21
124 5 $182 0.00087 $3810 124 5 $4336 12.13
136 6 $200 0.00094 $4111 136 6 $4483 08.30
68 1 $21 0.0014 $2283 110 1 $3916 41.70
82 2 $90 0.0012 $2878 110 2 $3958 27.29
0.55 90 3 $131 0.0010 $3261 111 3 $4058 19.64
102 4 $160 0.00098 $3561 113 4 $4170 14.60
114 5 $182 0.00096 $3814 115 5 $4279 10.87
125 6 $200 0.00095 $4035 125 6 $4402 08.34
71 1 $21 0.0015 $2268 110 1 $3900 41.85
86 2 $90 0.0012 $2864 109 2 $3935 27.22
0.60 96 3 $131 0.0011 $3248 110 3 $4029 19.38
104 4 $160 0.0010 $3539 111 4 $4138 14.48
109 5 $182 0.00099 $3778 113 5 $4244 10.98
116 6 $200 0.00095 $3986 115 6 $4344 08.24

50
Table 3.9: Effects of h= 40 on optimal solution

Ordering cost and lost sales reduction Fixed ordering cost and lost sales
k (proposed model) (A = A0 = 200 & βe = βe0 = 0.7) Savings
Q ∗ L A∗ βe∗ JET C(.) Q∗ L∗ JET C(.) (%)
66 1 $21 0.0011 $2509 106 1 $4262 41.13
91 2 $90 0.00098 $3201 106 2 $4338 26.21
0.40 112 3 $131 0.00095 $3698 112 3 $4475 17.36
129 4 $160 0.00091 $4096 129 4 $4681 14.28
144 5 $182 0.00089 $4435 144 5 $4907 09.62
158 6 $200 0.00087 $4738 158 6 $5138 07.79
66 1 $21 0.0011 $2490 105 1 $4240 41.27
85 2 $90 0.0010 $3151 105 2 $4307 26.84
0.45 104 3 $131 0.00096 $3630 107 3 $4428 18.02
120 4 $160 0.00093 $4004 120 4 $4597 12.90
134 5 $182 0.00091 $4326 134 5 $4796 09.80
147 6 $200 0.00089 $4612 147 6 $5003 07.82
65 1 $21 0.0012 $2473 105 1 $4220 41.40
78 2 $90 0.0010 $3108 104 2 $4278 27.35
0.50 96 3 $131 0.00098 $3561 106 3 $4394 18.96
111 4 $160 0.00095 $3924 111 4 $4529 13.36
124 5 $182 0.00093 $4229 124 5 $4700 10.02
136 6 $200 0.00091 $4498 136 6 $4882 07.87
64 1 $21 0.0013 $2456 104 1 $4201 41.54
77 2 $90 0.0011 $3094 103 2 $4251 27.22
0.55 89 3 $131 0.00099 $3517 104 3 $4361 19.35
102 4 $160 0.00096 $3859 106 4 $4484 13.94
114 5 $182 0.00094 $4146 114 5 $4621 10.28
125 6 $200 0.00092 $4397 125 6 $4777 07.96
64 1 $21 0.0014 $2441 103 1 $4184 41.66
77 2 $90 0.0012 $3082 102 2 $4226 27.07
0.60 85 3 $131 0.0010 $3494 103 3 $4332 19.34
94 4 $160 0.00097 $3816 105 4 $4451 14.27
106 5 $182 0.00096 $4094 106 5 $4568 10.38
116 6 $200 0.00094 $4332 116 6 $4701 07.85

51
Table 3.10: Effects of h= 45 on optimal solution

Ordering cost and lost sales reduction Fixed ordering cost and lost sales
k (proposed model) (A = A0 = 200 & βe = βe0 = 0.7) Savings
Q ∗ L A∗ βe∗ JET C(.) Q∗ L∗ JET C(.) (%)
65 1 $21 0.0010 $2683 100 1 $4523 40.68
91 2 $90 0.00096 $3452 101 2 $4618 25.25
0.40 112 3 $131 0.00092 $4006 112 3 $4792 16.40
129 4 $160 0.00090 $4256 129 4 $5047 15.67
144 5 $182 0.00086 $4831 144 5 $5316 09.12
158 6 $200 0.00084 $5173 158 6 $5586 07.39
63 1 $21 0.0011 $2655 100 1 $4510 41.13
85 2 $90 0.00098 $3389 100 2 $4585 26.09
0.45 104 3 $131 0.00094 $3914 103 3 $4717 17.02
120 4 $160 0.00091 $4340 120 4 $4943 12.20
134 5 $182 0.00088 $4700 134 5 $5183 09.32
147 6 $200 0.00086 $5203 147 6 $5426 07.43
62 1 $21 0.0012 $2637 100 1 $4489 41.26
78 2 $90 0.00099 $3331 99 2 $4556 26.89
0.50 96 3 $131 0.00095 $3834 100 3 $4682 18.11
111 4 $160 0.00092 $4240 111 4 $4855 12.67
124 5 $182 0.00090 $4582 124 5 $5064 09.52
136 6 $200 0.00088 $4885 136 6 $5281 07.50
62 1 $21 0.0012 $2283 98 1 $3916 41.36
74 2 $90 0.0010 $2878 98 2 $3958 27.12
0.55 89 3 $131 0.00097 $3261 99 3 $4058 18.78
102 4 $160 0.00094 $3561 102 4 $4170 13.18
114 5 $182 0.00091 $3814 114 5 $4279 09.77
125 6 $200 0.00089 $4035 125 6 $4402 07.55
61 1 $21 0.0013 $2607 98 1 $4452 41.44
73 2 $90 0.0011 $3287 97 2 $4503 27.00
0.60 82 3 $131 0.00098 $3733 98 3 $4619 19.18
94 4 $160 0.00095 $4097 99 4 $4749 17.14
106 5 $182 0.00094 $4410 106 5 $4893 09.87
116 6 $200 0.00092 $4677 116 6 $5057 07.51

52
k=0.40

Joint Expected Total Cost (JETC)


3500 k=0.45
k=0.50
k=0.55
3000 k=0.60

2500

2000

1500
160

140

120

100

80 6
5
Order quantity (Q) 4
3
60 2
1 Lead time(L)

Figure 3.6: Graphical representation of the optimal solution when c1 =15.

k=0.40
Joint Expected Total Cost (JETC)

3500 k=0.45
k=0.50
k=0.55
3000
k=0.60

2500

2000

1500
160

140

120

100

80 6
Order quantity (Q) 5
4
60 3
2
1 Lead time(L)

Figure 3.7: Graphical representation of the optimal solution when c1 =20.

53
k=0.40

Joint Expected Total Cost (JETC)


k=0.45
3500 k=0.50
k=0.55
3000 k=0.60

2500

2000

1500
160

140

120

100
6
80 5
4
Order quantity (Q) 3
60 2
1 Lead time(L)

Figure 3.8: Graphical representation of the optimal solution when c1 =25.

k=0.40
Joint Expected Total Cost (JETC)

k=0.45
4000 k=0.50
k=0.55
3500 k=0.60

3000

2500

2000
160

140

120

100
6
80 5
Order quantity (Q) 4
3
60 2
1 Lead time(L)

Figure 3.9: Graphical representation of the optimal solution when h=30.

54
k=0.40

Joint Expected Total Cost (JETC)


k=0.45
4500
k=0.50
4000 k=0.55
k=0.60
3500

3000

2500

2000
160

140

120

100
6
80 5
Order quantity (Q) 4
3
60 2
1 Lead time(L)

Figure 3.10: Graphical representation of the optimal solution when h=35.

k=0.40
k=0.45
Joint Expected Total Cost (JETC)

k=0.50
5000
k=0.55
k=0.60
4000

3000

2000
160

140

120

100
6
5
80 4
Order quantity (Q) 3
60 2
1 Lead time(L)

Figure 3.11: Graphical representation of the optimal solution when h=40.

55
k=0.40

Joint Expected Total Cost (JETC)


k=0.45
6000 k=0.50
k=0.55
5000 k=0.60

4000

3000

2000
160

140

120

100
6
80 5
Order quantity (Q) 4
3
60 2
1 Lead time(L)

Figure 3.12: Graphical representation of the optimal solution when h=45.

3.8 Appendix A
Derivation of E(Y 2 ):
We know that V (Y ) = E(Y 2 ) − (E(Y ))2
σ02 + σ12 Q2 = E(Y 2 ) − (E(Y ))2
σ02 + σ12 Q2 = E(Y 2 ) − α 2 Q2
E(Y 2 ) = σ02 + (σ12 + α 2 )Q2

3.9 Appendix B.
We want to prove the Hessian Matrix of JET C(Q, β̃, L) at point (Q∗ , β̃ ∗ ) for fixed L ∈
[Le , Ls ] is positive definite. We first obtain the Hessian matrix H as follows:
∂2 JET C(Q,β̃,L) ∂2 JET C(Q,β̃,L)
 
∂Q2
 
∂Q∂β̃
H = 
 
 ∂2 JET C(Q,β̃,L) ∂2 JET C(Q,β̃,L) 

∂β̃∂Q ∂β̃ 2

56
where

∂2 JET C(Q, β̃, L) 2(u + vlnL)D hσ02 D(π + π0 β̃)σ Lψ(k) 2Dδe−γL
= + + +
∂Q2 αQ3 αQ3 αQ3 αQ3

∂2 JET C(Q, β̃, L) Dπ0 σ Lψ(k)
=− .
∂Q∂β̃ αQ3

∂2 JET C(Q, β̃, L) Dπ0 σ Lψ(k)
=−
∂β̃∂Q αQ3
∂2 JET C(Q, β̃, L) θc
2
= − 21 .
∂β̃ β̃

Then we proceed by evaluating the principal minor determinant of H.


The first principal minor of H is

2(u + vlnL)D hσ02 D(π + π0 β̃)σ Lψ(k) 2Dδe−ηL
|H11 | = + + + >0
αQ3 αQ3 αQ3 αQ3

The second principle minor of H is:


√ √ !2
σ02 h (π + π0 β̃)σ Lψ(k)
" #
2D −γL θc1 −Dπ0 σ Lψ(k)
|H22 | = (u + vL) + + + δe −
αQ3 2D 2 β̃ 2 αQ2
√ √ !2
σ02 h (π + π0 β̃)σ Lψ(k)
" #
2D −γL θc 1 −Dπ 0 σ Lψ(k)
(u + vL) + + + δe >
αQ3 2D 2 β̃ 2 αQ2

Therefore |H22 | > 0.


We see that all the principal minors of the Hessian Matrix are positive. Hence, the Hessian
Matrix H is positive definite at (Q∗ , β̃ ∗ )

57

You might also like