Dr.

Muhammad Shafiq
Department of Industrial Engineering
University of Engineering and Technology
A specific type of mathematical programming in which optimal
solution of the original complex problems is found by solving a chain
of sub-problems
The optimal solution of one sub-problem is used as input to the next
sub-problem
When last sub-problem is solved, the optimal solution of entire
problem is achieved
Linkage between the stages is performed through recursive
computations
Forward or backward recursive equation will be developed
depending on problem nature
 Shortage path problem

12 7
2 5 7
4 15 3
1 6
7 10
1 4 8 9
3 3 7
2
3 15
6
4
Backward recursive

fi minimum total travel time from node i to node 9

tij travel time through arc (i, j )

fi  tij  f j i  9; j

fi  min tij  f j  i9

j

However, the shortage path from node i to node 9 should include

fi  min tij  f j  i9

j
f9  0

f8  t89  f9  10  0  10

f7  t79  f9  3  0  3

t68  f8  7  10
f 6  min    min    15
t69  f9  15  0 

f5  t57  f7  7  3  10

t45  f5  4  10 
   
t46  f6  3  15 
f 4  min    min    14
t47  f7  15  3 
t48  f8  7  10
 t34  f 4  3  14 
f3  min    min    17
t36  f 6  4  15

t24  f 4  6  14 
f 2  min    min    20
t25  f5  12  14

t12  f 2  1  20 
f1  min    min    19
t13  f3  2  17 

Shortage path from node 1 to node 9: 1-3-4-5-7-9

Total travel time=19

 Shortage path

7
2 5 7
4 3
6

4 8 10
1 9
3
2
3 15
6
 Forward Recursive

fi minimum total travel time from node 1 to node i

tij travel time through arc (i, j )

f j  tij  fi j  1; i

Hence, f j  min
i
tij  fi  j 1

However, the shortage path from node i to node 9 should include

f j  min tij  fi  j 1
i
f1  0

f 2  t12  f1  1  0  1

f3  t13  f1  2  0  2

t24  f 2  6  1 
f 4  min    min  5
t34  f3  3  2

t25  f 2  12  1
f5  min    min  9
t45  f 4  4  5 

t36  f3  4  2 
f 6  min    min  6
t46  f 4  3  5 
 t47  f 4  15  5
f 7  min    min    16
t57  f5  7  9 

t48  f 4  7  5 
f8  min    min    12
t68  f 6  7  6

t79  f7  3  16 
f9  min    min    19
t89  f8  10  12
 Principle of Optimality (Bellman)

An optimal policy has the property that whatever the initial state and initial decision are, the
remaining decisions must constitute an optimal policy with regard to the state resulting from the first
decision.

Stage n stage n+1

xn
sn sn 1

rn (sn , xn )
Transformation function

sn 1  tn (sn , xn )
Maximization problem (backward recursive technique)

If we denote fn (sn ) as maximum total revenue obtained when system moves from stage n to stage N
(the last stage), given the observed state at stage n is sn

f n  max rn (sn , xn )  f n 1 (tn (sn , xn ))

xn D ( sn )

D(sn ) is the set of all possible decisions of a given state sn at stage n.

Forward recursive

Similarly when forward recursive technique is employed, if we denote fn (sn ) as the maximum total
revenue obtained when system moves from stage 1 to stage n

f n 1 (sn 1 )  max rn (sn , xn )  f n (sn )

xn D ( sn1 )

D(sn 1 ) is the set of all possible decisions to transfer from state sn at stage n to state sn 1 at stage n+1
 Shortage path from node 1 to node 10

2 7 5 1
4 8
6 4
2 3
3 6
4 2
1 3 4 6 3 10
4 9 4
3 1 3
3
4 5
7
Stage 1 Stage 2 Stage 4 Stage 5
Stage 3
 Backward recursive

Stage 4

s4 x4 r4 ( s4 , x4 )  f5 (t4 ( s4 , x4 )) f 4 ( s4 ) x4*
10

Node 8 3 3 10

Node 9 4 4 10
When the state is node 8:

r4 (s4 , x4 ) = r4 (8,10) =3 f5 (t4 (s4 , x4 )) = f5 (10) =0

f 4 (s4 )  f 4 (8)  3

When the state is node 9:

r4 (s4 , x4 ) = r4 (9,10) =4 f5 (t4 (s4 , x4 )) = f5 (10) =0

f 4 (s4 )  f 4 (9)  4
Stage 3

s3 x3 r3 (s3 , x3 )  f 4 (t3 (s3 , x3 )) f3 ( s3 ) x3*

8 9

Node 5 1+3=4 4+4=8 4 8

Node 6 6+3=9 3+4=7 7 9

Node 7 3+3=6 3+4=7 6 8

Stage 2

x2
s2 r2 (s2 , x2 )  f3 (t2 (s2 , x2 )) f 2 ( s2 ) x2*

5 6 7

Node 2 7+4=11 7+4=11 6+6=12 11 5 or 6

Node 3 3+4=7 2+7=9 4+6=10 7 5

Node 4 4+4=8 1+7=8 5+6=11 8 5 or 6

Stage 1

x1
s1 r1 (s1 , x1 )  f 2 (t1 (s1 , x1 )) f1 ( s1 ) x1*

2 3 4

Node 1 2+11=13 4+7=11 3+8=11 11 3 or 4

 1-3-5-8-10
 1-4-5-8-10
 1-4-6-9-10
 Five medical teams will be dispatched to 3 regions to help
improve medical care. The performance is measured by the
expected additional person-years of life. The estimated
performance measures are given in table below.
 Xn : (n =1, 2, 3) are the number of teams to allocate to stage (country) n
 Sn :number of medical teams still available for allocation to remaining countries (n, . . . , 3)
 pi(xi) : performance measure from allocating xi medical teams to country n
 fn(sn, xn) :total maximum performance obtained when Sn teams are allocated
 Resource allocation problem to produce N products
 Backward
 Forward