Some Additional Techniques in Formulation

1. Index and Summation Notation

Many models can be written more succinctly with the use of indices and variables such as (𝑥𝑖 and 𝑦𝑖𝑗 ).
Moreover, if variables have indices, it is possible to use summation notation to write out objectives and
constraints. For example:

• 𝑎1 + 𝑎2 + 𝑎3 + 𝑎4 can be written as ∑4𝑖=1 𝑎𝑖

• 𝑥11 + 𝑥21 + 𝑥31 + 𝑥41 can be written as ∑4𝑖=1 𝑥𝑖1
• 𝑦11 + 𝑦12 + 𝑦13 + 𝑦14 + 𝑦21 + 𝑦22 + 𝑦23 + 𝑦24 + 𝑦31 + 𝑦32 + 𝑦33 + 𝑦34 = ∑4𝑗=1 ∑3𝑖=1 𝑦𝑖𝑗

Also, if you want to use given parameters such as the following unit production cost table:

Period 1 2 3 4 5 6
Unit Cost 4.20 5.10 6.00 5.40 4.80 4.20

you can denote the parameters using 𝑐𝑗 = unit cost for period 𝑗, 𝑗 = 1, 2, …, 6. If 𝑥𝑗 represents number of units
produced in period 𝑗, total production cost can then be written using summation notation: ∑𝑗 𝑐𝑗 𝑥𝑗 .

Resource capacities and demand requirements can also be dealt with in this manner.

2. Multiple Period Planning

In a multiple-period planning problem, it can be useful to keep track of the running level of a certain resource
at the beginning or end of each period (e.g. inventory, cash on hand, etc.). Hence, one can decide to define an
additional variable or expression (that can depend on other decision variables), just to serve as some sort of
“tracker”.

• For inventory problems, consider using 𝐼𝑖 = level of inventory at the beginning or end of period 𝑖
• For capital budgeting problems, consider using 𝐶𝑖 = cash/capital on hand at the beginning or end of
period 𝑖
• Then use relationships to define these variables, like (note: dependent on problem):
• 𝐼2 = 𝐼1 + production level in period 1 − demand in period 1
• 𝐶2 = 𝐶1 − cash outflow in period 1 + cash inflow in period 1
• Such relationships are called recurrence relations. These will appear again when you
encounter other optimization algorithms for multiple-period problems.

3. Unrestricted (URS) Variables

Unrestricted variables (URS) are those that can take positive, zero, or negative values. The strategy for this
situation is to express each URS as a difference of two nonnegative variables. For example, if 𝑥𝑗 is URS, we
can express 𝑥𝑗 = 𝑥𝑗+ − 𝑥𝑗− , where 𝑥𝑗+ and 𝑥𝑗− are both nonnegative. That is, whenever 𝑥𝑗 occurs in the model,
simply rewrite 𝑥𝑗 as the quantity (𝑥𝑗+ − 𝑥𝑗− ).

Remark: if 𝑥𝑖 is URS for multiple values of 𝑖, then create a pair of 𝑥𝑖+ and 𝑥𝑖− for each such 𝑖.
4. Surplus and Deficit
You will sometimes be interested in how much a certain quantity or function 𝑓(𝑥1 , 𝑥2 , … , 𝑥𝑛 ) is greater than,
or less than, a certain “target value” (e.g. is current inventory greater than 25 so holding cost is incurred, or
is it less than 25 so shortage cost will be incurred).
One possible strategy in this situation is to define the auxiliary variables 𝑠 (surplus) and 𝑑 (deficit), and then
form the constraint 𝑓(𝑥1 , 𝑥2 , … , 𝑥𝑛 ) − 𝑠 + 𝑑 = 25 (target value). We must, however, ensure that an
appropriate incentive or penalty is present in the objective function so that the correct pair of (𝑠, 𝑑) will be
chosen – the pair where at least one of 𝑠 or 𝑑 will be set to zero, and the nonzero variable will isolate the
surplus or deficit, respectively, of 𝑓(𝑥1 , 𝑥2 , … , 𝑥𝑛 ) from the target value.
Remarks: feel free to use subscripts 𝑠𝑖 and 𝑑𝑖 in case you need to use this strategy on multiple quantities

5. Minimax and Maximin

If we encounter the objective of maximizing 𝑍 = min{𝑥1 , 𝑥2 , … , 𝑥𝑛 }, use the following strategy:

• Step 1: Change objective to Maximize 𝑍 = 𝑦

• Step 2: Add the constraints 𝑦 ≤ 𝑥𝑖 for all 𝑖.
Notice that we are using the fact that the minimum in the set is less than or equal to everyone in the set,
including itself. Now analogously, if the objective is to minimize 𝑍 = max{𝑥1 , 𝑥2 , … , 𝑥𝑛 }, use the following:

• Step 1: Change objective to Minimize 𝑍 = 𝑦

• Step 2: Add the constraints 𝑦 ≥ 𝑥𝑖 for all 𝑖.

6. Dealing with Absolute Value

Absolute value expressions of the form |𝑓(𝑥1 , 𝑥2 , … , 𝑥𝑛 )| are useful for problems that involve minimizing
errors or distances. They can technically be written as they are. However, the models obtained would not be
readily suitable for optimization algorithms. Hence, some adjustments must be made.

• Case 1: If |𝑓(𝑥1 , … , 𝑥𝑛 )| occurs in a constraint

o |𝑓(𝑥1 , … , 𝑥𝑛 )| ≤ 𝑎 and 𝑎 > 0: write two constraints 𝑓(𝑥1 , … , 𝑥𝑛 ) ≥ −𝑎 and 𝑓(𝑥1 , … , 𝑥𝑛 ) ≤ 𝑎
o |𝑓(𝑥1 , … , 𝑥𝑛 )| ≥ 𝑎 > 0: for now, write two separate constraints 𝑓(𝑥1 , … , 𝑥𝑛 ) ≥ 𝑎 and
𝑓(𝑥1 , … , 𝑥𝑛 ) ≤ −𝑎. This time around, we need to deactivate exactly one of these two
constraints. The standard procedure of deactivation will be covered in the next module.
• Case 2: If |𝑓(𝑥1 , … , 𝑥𝑛 )| occurs in the objective function
o If all 𝑥𝑖 are nonnegative and all coefficients are positive, you can simply remove the absolute
value bars outside the function.
o Otherwise, we can view |𝑓(𝑥1 , … , 𝑥𝑛 )| as a measure of deviation from 0 (or another particular
target value). We can then use the strategies outlined in items 3 and 4 above.
Remark: You may find that the surplus and deficit strategy would not work effectively on some problems
with the objective of maximizing 𝑍 = |𝑓(𝑥1 , … , 𝑥𝑛 )|, where |𝑓(𝑥1 , … , 𝑥𝑛 )| is viewed as a distance or error.
To check your understanding of this module, try out the following simple exercises.
1. A baker has 30 oz. of flour and 5 packages of yeast. Baking a loaf of bread requires 5 oz. of flour and 1
package of yeast. Each loaf of bread can be sold for 30¢. The baker may purchase additional flour at 4¢/oz.,
or sell leftover flour at the same price. Formulate an LP to help the baker maximize profits.

2. Finco Investment Corporation must determine investment strategy for the firm during the next three
years. Currently, $100000 is available for investment. Investments A, B, C, D, and E are available. The cash
flow associated with investing $1 in each investment is given in the following table:

Cash Flow ($) at Time

0 (currently) 1 2 3
A –1 +0.50 +1 0
B 0 –1 +0.50 +1
C –1 +1.2 0 0
D –1 0 0 +1.9
E 0 0 –1 +1.5

For example, $1 is invested in investment B requires a $1 cash outflow at time 1, returns 50¢ at time 2, and
returns $1 at time 3. To ensure that the company’s portfolio is diversified, Finco requires that at most $75000
be placed at any single investment. In addition to investments A to E, Finco can earn interest at 8% per year
by keeping uninvested cash in money market funds.

Furthermore, returns from investments may be immediately reinvested. For example, the positive cash flow
received from investment C at time 1 may immediately be reinvested in investment B. Finco cannot borrow
funds, so that cash available for investment at any time is limited to cash on hand. Formulate an LP that will
maximize cash on hand at the end of time 3.

3. Consider the following LP problem:

Maximize 𝑍 = |2𝑥1 − 3𝑥2 |, subject to the following constraints:

• 4𝑥1 + 𝑥2 ≤ 4
• 2𝑥1 − 𝑥2 ≤ 0.5
• 𝑥1 , 𝑥2 ≥ 0

Reformulate the objective appropriately in order to deal with the absolute value function. Afterwards, solve
the problem using any preferred algorithm. Check that your solution (values of decision variables and
optimal 𝑍 value) are consistent with your formulation.

4. Steelco’s main plant currently has a steel manufacturing area and shipping area. The locations of these
areas can be modeled on a coordinate plane, with coordinates (700, 600) and (1000, 0), respectively, relative
to some origin point (distances are in feet). The company must determine where to locate a casting facility
and an assembly and storage facility to minimize the daily cost of moving material through the plant. The
number of trips to be made each day are shown in the following table:
 Cost (¢) per 100 feet
From To Daily # of Trips
Traveled
Casting Assembly and storage 40 10
Steel manufacturing Casting 8 10
Steel manufacturing Assembly and storage 8 10
Shipping Assembly and storage 2 20

Assuming that all travel is in only an east-west or north-south direction, formulate an LP to determine where
to place the casting facility and the assembly and storage facility in order to minimize daily transportation
costs.