Content uploaded by Susana Fernandes
Author content
All content in this area was uploaded by Susana Fernandes on Jan 26, 2016
Content may be subject to copyright.
SYMCOMP 2013
Lisbon, 9-10 September 2013
c
ECCOMAS, Portugal
TWO-VARIABLE LINEAR PROGRAMMING: A
GRAPHICAL TOOL WITH MATHEMATICA
Jos´e C. Pereira1and Susana Fernandes2
1: Departamento de Engenharia Electr´onica e Inform´atica
Faculdade de Ciˆencias e Tecnologia
Universidade do Algarve
Campus de Gambelas, 8005-139 Faro, Portugal
e-mail: unidadeimaginaria@gmail.com
2: Departamento de Matem´atica
Faculdade de Ciˆencias e Tecnologia
Universidade do Algarve
Campus de Gambelas, 8005-139 Faro, Portugal
e-mail: sfer@ualg.pt
Keywords: linear programming for two variables, graphical visualization, dynamic, in-
teractive, teaching tool, Wolfram Mathematica, Computable Document Format
Abstract. This paper presents the GLP-Tool, an interactive tool for graphical linear pro-
gramming involving two variables. The GLP-Tool is designed to solve user-defined linear
programming problems with two variables, up to a didactical limit of five constraints (plus
the signal constraints). Implemented using the computer algebra system Mathematica,
this interactive tool allows the user to dynamically explore different objective functions and
constraint sets, and also perform post-optimal and sensitivity analysis. All the GLP-Tool
functionalities are represented graphically and updated in real time. These interactive,
dynamic, and graphical features make the GLP-Tool a powerful tool for teaching linear
programming both in undergraduate and high school courses. After completing its develop-
ment, we intend to make the GLP-Tool available at the Wolfram Demonstrations Project
website.
Jos´e C. Pereira and Susana Fernandes
1 INTRODUCTION
When introducing the subject of Linear Programming (LP) it is rather useful to present
the graphical method for solving a two-variable linear program as it provides valuable in-
sights about the general nature of multivariable linear programming models. The graph-
ical method illustrates numerous aspects of the more complex, algebra-based, solution
algorithm - the simplex method. Specifically, graphically solving a linear program pro-
vides students with intuitive visual aids to facilitate their understanding of concepts such
as the feasible region, basic feasible solutions, unbounded solutions, binding /nonbinding
constraints, degeneracy, slack, and so on.
Nonetheless, without a dynamic tool, it is not easy to show students what happens in a
(two-variable) LP problem as constraint boundary lines and objective-value lines move
around on a graphic. To provide students with an effective learning environment a tool
should show graphically and dynamically the construction of the feasible region of two-
variable linear programs, and should allow to interactively experiment with the feasible
solutions set and the objective function. This kind of tool is what is called an active
learning tool.
Active learning is generally defined as any instructional method that engages students
in the learning process [1]. The core elements of active learning are student activity
and engagement in the learning process within the classroom. In short, active learning
requires students to do meaningful learning activities and think about what they are doing.
Prince [2] summarizes some of the most relevant literature in the field of active learning,
which report strong evidences of the effectiveness of using active learning techniques.
In particular, the work of Kydd [3] reports the effectiveness of using an active learning
technical tool while teaching linear programming.
This paper presents the GLP-Tool, a good example of an active learning technical tool that
engages students and provides them with an effective learning environment. Implemented
using the computer algebra system Mathematica, this interactive tool allows the user to
dynamically explore and solve two-variable linear programming problems with different
objective functions and constraint sets, and also to perform post-optimal and sensitivity
analysis. All the GLP-Tool functionalities are represented analytically and graphically
and updated in real time.
The interactive, dynamic, analytical, and graphical features of the GLP-Tool make this
application a powerful tool for teaching LP both in undergraduate and high school courses.
It can be used both by teachers to graphically illustrate fundamental concepts and by
students to experiment with changes within a graphical representation of a linear program,
facilitating their understanding of numerous LP concepts.
When we set up to produce an innovative tool for graphical linear programming, the
GLP-Tool, we chose to implement it using the computer algebra system Mathematica
not only because it enables the production of sophisticated, dynamic, interactive visual
applications but also due to the previous experience of one of the authors in producing
Jos´e C. Pereira and Susana Fernandes
didactical tools with Mathematica [4].
We intend to make the GLP-Tool available online to the general public, via the Wolfram
Demonstrations Project website. The corresponding source code will also be available at
this website.
We believe that this new GLP-Tool is an important contribution to the improvement of
the teaching-and-learning process of LP precisely by providing teachers and students alike
with an active learning technical tool to explore fundamental concepts of the subject.
The rest of this paper is organized as follows. In Section 2 we present an overview of
existing tools for graphical linear programming and also present the motivations for using
the computer algebra system Mathematica to produce a new innovative tool. In Section
3 we describe the GLP-Tool concept and explain how to use its main features. In Section
4 we provide several usage examples of how to introduce LP concepts with the didactical
GLP-Tool in a classroom environment. In Section 5 we make some final remarks about
our current and future work related with the GLP-Tool concept.
2 Web-based tools for graphical linear programming
There are several Java applets freely available online which graphically solve linear pro-
grams. Some of them graphically illustrate the steps of the algebra-based simplex method,
moving from one basic solution to the next until it reaches the optimal solution(s). Ex-
amples of these Java applets are the ‘Linear Programming and Pivoting in 2D’ [5], the
‘LP Explorer 1.0’ [6] or the ‘Graphical Simplex Algorithm (2D)’ [7]. While being very
useful when trying to visually explain/understand the simplex method we believe they are
not suited for introducing the subject of linear programming since they do not show very
important features in understanding LP as, for instance, the mapping of the objective
function value within the feasible solutions set.
Other Java applets exist that do allow the user to manipulate the constraints and/or the
objective function in order to see the effect on the feasible region and on the optimal
solution(s). Examples of theses applets are the ‘Exploring linear programming’ [8], the
‘Linear programming applet’ [9] or the ‘Animated linear programming applet’ [10]. Typ-
ically these applets allow the user to define a linear program with two variables with a
total number of constraints up to 4 or 5.
All Java applets freely available online that graphically solve linear programs are not vi-
sually sophisticated, when compared to a commercial application. Wolfram’s Mathemat-
ica is a powerful computer algebra system that enables the production of sophisticated,
dynamic, interactive visual applications. Wolfram’s Mathematica is used in scientific, en-
gineering, and mathematical fields and in other areas of technical computing. Creating
interactive visual models with Mathematica allows users to explore hard-to-understand
concepts, test theories, and quickly gain a deeper understanding of the materials being
introduced firsthand. The users can explore changes to text, functions, formulas, matri-
ces, graphics, tables, or even data. In what concerns the work presented in this paper,
Jos´e C. Pereira and Susana Fernandes
not only Mathematica algebraically solves LP programs using a single command as also
its graphics are completely integrated into its dynamic interactive language. Any visu-
alization can immediately be animated or made interactive using a single command and
developed into sophisticated, dynamic visual applications.
We have found some Mathematica demonstrations on the website of the Wolfram Demon-
strations Project which address the subject of linear programming, but none of them
works with an user-defined linear program. An intentionally very simple application is
the ‘Graph of Inequalities’ [11] which shows that the graph of a linear equation is a straight
line, the graph of a linear inequality is an half-plane and that the graph of a system of
linear inequalities is the intersection of the half-planes. There’s a Wolfram Mathematica
demonstration that graphically illustrates the steps of the (two-phase) simplex method on
random generated linear programs; the ‘Two-Phase Simplex Method’ [12]. While being
very useful when trying to visually explain/understand the simplex method we believe it
is not suited for introducing the subject of linear programming.
Some Mathematica demonstrations focus on illustrating the weak version of the funda-
mental theorem of linear programming which states that the optimal solution to a linear
program, if it exists, is attained at (at least) one vertex of the feasible solutions set.
Examples of these are the ‘Parametric Linear Programming’ [13], the ‘The Fundamental
Theorem of Linear Programming’ [14] or the ‘Graphical Linear Programming for Two
Variables’ [15]. While being very useful for illustrating the weak version of the fundamen-
tal theorem of linear programming, they do not enable the visualization of very important
concepts of LP. Other Mathematica notebooks exist that do allow the user to manipulate
the constraints (and the objective function) in order to see the changes on the feasible
region and on the optimal solution(s). Examples of these are the ‘Oil Mallee Farming
Optimization Problem’ [16] or the ‘A Simple, Standard Linear Programming Scenario’
[17].
One word about the internet site WolframAlpha. WolframAlpha is a denominated com-
putational knowledge engine which can be used to solve a user-defined linear program.
When entering a linear program as - Max ‘objective function’ in ‘inequality 1’ and ‘inequal-
ity 2’ and .... - the engine outputs the optimal solutions(s), the corresponding optimal
value and it also graphs the feasible solutions set, highlighting the feasible solution(s)
where the optimal value is reached. WolframAlpha is a powerful engine but it is not an
interactive didactical tool. For instance, it does not allow the user to drag the objective
function line and see the way its value changes within the feasible region.
When looking for a graphical tool for introducing linear programming we found internet
freely available tools to be either unfit for this purpose ([5], [6], [7], [12]), misdirected ([13],
[14], [15]) or incomplete ([8], [11], [16], [17]). Only two of them ([9], [10]) were flexible
enough to enable the visualization of linear programs that either: have an unbounded
feasible region; have redundant constraints; have degenerated solutions; are unbounded;
are unfeasible. Unfortunately we encountered some malfunctions of the ‘Linear program-
ming applet’ [9] when defining totally different instances (namely the ones we will present
Jos´e C. Pereira and Susana Fernandes
further on this article). The‘Animated linear programming applet’ [10] is the only ap-
plication that actually computes the solution of the problem, but unfortunately it uses a
numerical system with finite precision which in some cases leads to errors in the output.
Another drawback is the fact that this application has a very poor graphical presentation.
Anyone can work online with all these applications, Java applets, Wolfram’s Mathematica
demonstrations or WolframAlpha. Nowadays the internet is globally reachable, nonethe-
less, in some situations (some portuguese schools, for instance) an application running
locally on a personal computer is necessary. Wolfram’s Mathematica demonstrations
have the advantage of being built into a single CDF file which runs as a stand-alone ap-
plication. Anyone can download it, and just ‘click’ his way through the installation of the
CDF player. For downloading a Java applet the user has to be able to read html or Java
script code, and also to be able to find all files needed to build the application.
With our new tool for graphical linear programming, the GLP-Tool, we aim at bringing
together the best of two worlds: the flexibility of Java applets and the sophistication and
easy packaging of Wolfram’s Mathematica demonstrations.
3 THE GLP-Tool CONCEPT
The GLP-Tool is a dynamic, interactive, and visual tool, implemented with Mathematica,
that allows to solve user-defined linear programming problems with two variables. In
particular, the user can explore different objective functions and constraint sets, obtain
graphical and numerical information on optimal solutions, and perform post-optimal and
sensitivity analysis.
The GLP-Tool explores the class of linear programming problems that can be written in
the general form
Max z =ax +by
s.t. aix+biy≤ti,i= 1,...,5
x, y ≥0
(1)
where the non-negativity constraints are optional.
The number of proper constraints is set to a didactical limit of five constraints, plus the
non-negativity constraints. This limitation was based on the fact that the GLP-Tool is
not a mere calculator but rather an active learning tool designed to engage students and
provide them with an effective Linear Programming learning environment. With that in
mind, we found that the five constraints limits was more than enough to illustrate all the
main LP concepts, without cluttering the visual interface of the GLP-Tool. In addition,
we could not find in the main literature any instances of graphical linear programming
problems that would exceed the five constraints limit.
All the functionalities of the GLP-Tool, along with all the displayed information, are
designed to conform with the way Linear Programming is used in research and presented
Jos´e C. Pereira and Susana Fernandes
in the classroom, both at the high school and undergraduate levels. All the GLP-Tool
features are represented graphically and updated in real time.
Figure 1 presents a usage example of the GLP-Tool. Visually, this active learning tool is
divided in two main panels.
Left Panel: This panel contains all the controls that allow the user to define the lin-
ear programming problem instances. In addition, it includes some options
related with the visual display of the feasible region and the objective func-
tion. The user can also choose to visualize the formalization of the problem
instance. More importantly, the feature Panel Mode, offers three different
ways to interact with the right panel, Explore,Max, and Min.
Right Panel: In this panel it is presented all the graphical information concerning the
linear programming problem instance. The feasible region is depicted along
with the constraint boundary lines. The objective function line is graphed
for a fixed value of z. The feasible solutions corresponding to this value of z
are also presented, when they exist. In addition, the problem formalization
and the optimal solution information are displayed in two insets, when the
corresponding options are chosen.
The GLP-Tool has a very intuitive interface that allows even the most inexperienced user,
with no previous knowledge in educational software, to use all the GLP-Tool features in
an efficient and autonomous way, right from the start.
3.1 How to use the GLP-Tool
As mentioned before, all the graphical and analytical information presented in the left and
right panels is displayed in real time, thanks to the symbolic and numerical computation
capabilities of Mathematica. This renders the GLP-Tool as an eminently dynamic tool
designed for active learning.
The user can set the values of all the coefficients of a linear problem using the respective
sliders. The values can be changed one by one in a discrete or continuous way. In
particular, this kind of control allows the automatic continuous change of the coefficient
values by clicking the play button ( I) depicted, for example, bellow the a-slider in
Figure 1. When choosing this option, the user will immediately see an animation of the
corresponding graphical information in the right panel. For instance, one may rotate
or translate one of the constraint boundary lines and observe how the feasible region
changes accordingly. In addition, the corresponding problem formalization can be seen
continuously changing, when the respective light green inset is displayed in the right panel
(see Figure 1). Note that, this inset is not a static display but rather a dynamic object
in which all the contained information is updated in real time.
The problem formalization can also be seen in the graphic itself. When the mouse pointer
is over each constraint boundary line a small yellow tag is displayed, identifying the line
Jos´e C. Pereira and Susana Fernandes
Figure 1: A general usage example of the GLP-Tool. The chosen panel mode is Max.
with the corresponding constraint (see Figure 1). The same is true for the objective
function line when graphed for the optimal z.
The Panel Mode is one of the main features of the GLP-Tool. This functionality includes
two optimize modes Max and Min that allow the user to obtain the optimal solution(s)
of the respective optimization problems. The optimal information is then displayed both
graphically and numerically in the right panel (see Figure 1). Note that, these modes
are able to compute all kinds of optimal sets, wether they may be empty, with a single
element or with multiple solutions, both bounded or unbounded.
A good usage example of the GPL-Tool is to choose one of the two optimize modes, with a
fixed feasible region and click the play button for the objective function coefficients aand
b. The corresponding line will start rotating and will eventually “visit” several vertices of
the feasible region. The user can then observe the optimal solution information change
accordingly. This visualization is, for instance, a vey good way to introduce the concept
of basic feasible solutions. In addition it also promotes a better understanding of the role
of the objective function coefficients in post-optimal and sensitivity analysis.
The Panel Mode includes a third mode, Explore. As its name indicates, this option allows
the exploration of the set of feasible solutions for any given set of constraints. While in
this mode, the user can click-and-drag the objective function line across the visible region
of the graphic in the right panel, with the mouse pointer. When this line intersects the
feasible region, the feasible solutions will be highlighted in red (see Figure 2). At the same
Jos´e C. Pereira and Susana Fernandes
Figure 2: Using the GLP-Tool to explore the set of feasible solutions. The chosen panel
mode is Explore.
time, the corresponding value of zis updated within the problem formalization display.
Another useful feature of the GLP-Tool is the ability to resize the Plot Window (see
Figures 1 and 2). This functionality is crucial to the visualization of any problem instance,
no matter how odd are the coefficient values. Up to our knowledge there is no other
graphical linear programming application with such feature, which seriously reduces their
working scope.
We note that the usage descriptions contained in this section are far from being com-
prehensive. All the GLP-Tool features can be combined to produce a large variety of
different useful interactions. As the user becomes more acquainted with this tool, he/she
will naturally find those interactions that fit best his/her purpose.
It is also through this kind of dynamic interaction that the GLP-Tool becomes a fully
active learning technical tool that engages students and teachers and provides them with
an effective teaching/learning environment.
4 TEACHING GRAPHICAL LINEAR PROGRAMMING USING THE
GLP-Tool
In this section we present a didactical use of the GLP-Tool.
We assume that the formulation of linear programs was already discussed and now we
Jos´e C. Pereira and Susana Fernandes
want to introduce the graphical method for solving a two-variable linear program.
Consider the following LP problem instance:
min Z =x+y
s.t. 3x+ 2y>12
3x−3y>2
x65
x, y >0
(2)
The GLP-Tool starts by default with a feasible region defined only by the non-negativity
constraints. The tool shows the construction of the feasible region as new constraints are
added.
Working with instance (2), after introducing all the constraints the teacher should point
out that the constraint x>0 is actually redundant (the green dashed vertical line does
not bound the feasible region in Figure 2). The students can easily verify this fact by
checking and unchecking the corresponding non-negativity option in the left panel and
observe that the feasible region in the right panel remains unchanged.
Next, it is necessary to set the values of the coefficients of the objective function. After-
wards, the students can use the Explore mode to observe the change in the value of zas
the objective function line is dragged across the right panel (see Figure 2). The teacher
can question which feasible solution(s) achieve(s) the optimal value of the problem. By
using the Min and Max modes it is possible to actually compute the optimal solutions (4,0)
and (5,13
3) for the minimum value z= 4 and the maximum value z=28
3, respectively.
This information is displayed in the light yellow inset in the right panel while the optimal
solution is highlighted in the graphic (see Figure 1)
While in the Max mode, we can also introduce the concept of binding and nonbinding
constraints. The students can observe that any change in the independent terms of the
inequalities 3x−3y>2 and x65 lead to a different optimal solution. Therefore,
these inequalities are binding ones. As for the other constraints, some variation of the
independent terms is allowed before the optimal solution changes (how much variation?)
and so, they are considered nonbinding. In particular, the case of the x>0 constraint
shows that all redundant constraints are also nonbinding.
The teacher should have shown up to this point how it is possible to change the coefficients
values either manually, by dragging the slider or by direct input, or automatically, by
clicking the play button.
Returning to the initial problem instance (2), it is possible to introduce post-optimal
analysis by questioning the students if the optimal solution would remain the same when
the constraint x65 changes to x66. And what about when the constraint 3x+2y>12
changes to 3x+ 2y>13? After observing the effects of such changes, the teacher should
encourage the students to try and modify other independent terms, one at a time. Figure
3 presents one possible example of this kind of interaction with the GLP-Tool.
Jos´e C. Pereira and Susana Fernandes
(a) LP instance with t1= 12. (b) LP instance with t1= 13. (c) LP instance with t1= 19.
Figure 3: Three similar LP problem instances only with different t1values inputed directly.
This kind of interaction with the GLP-Tool is very useful to introduce many concepts of
post-optimal and sensitivity analysis.
Similarly, sensitivity analysis can be introduced by questioning the students within which
interval may the independent term of the inequality 3x+ 2y>12 vary without changing
the optimal solution. This concept can then be generalized to include all other coefficients.
As before, the students should modify the values of several coefficients, of binding and
nonbinding constraints, and try to determine within which limits does the optimal solution
remains unchanged.
Notice that, when changing the coefficients/independent terms of the inequalities some
times we will get a feasible region with only one solution while other times we will get
an empty feasible region. In this case, an inset is displayed in the right panel with the
information that the problem is unfeasible, as depicted in Figure 4.
Figure 4: Example of an unfeasible LP problem. In this case, the GLP-Tool simply
displays that information in an inset in the right panel.
Considering again the initial problem instance (2), the students can now perform the
sensitivity analysis of the coefficients of the objective function. An interesting usage
Jos´e C. Pereira and Susana Fernandes
(a) LP instance with one maxi-
mal solution.
(b) LP instance with multiple
maximal solutions.
(c) LP instance with another
single maximal solution.
Figure 5: Three similar LP problem instances only with different objective function coef-
ficients. As the values of coefficients aand bchange, so does the optimal solution. This
example illustrates the weak version of the fundamental theorem of LP. It can also be
used to show why and when do multiple optimal solutions occur.
example in this analysis is to modify the value of the coefficients aand busing the
corresponding play button, while on the Max mode. As the values of the coefficients
change, the objective function line can be seen rotating in the right panel, moving from
one vertex to another adjacent vertex of the feasible region. Also, as depicted in Figures
5(a), 5(b), and 5(c), the information concerning the current maximal solution is displayed
and updated in the corresponding inset. The teacher should note that this example
clearly illustrates the weak version of the fundamental theorem of LP, which states that
the optimal solution to a linear program, if it exists, is attained at at least one vertex of
the feasible region.
In addition, the same example can also be used to introduce the concept of multiple
optimal solutions. After observing the motion of the objective function line and looking
at Figures 5(a), 5(b), and 5(c), the students should be asked questions such as When do
multiple optimal solutions occur? How many are there? What do they have in common?
What makes the values of coefficients aand bin Figure 5(b) special?
Let us now consider the following LP problem instance
max Z =−x+y
s.t. 3x+ 2y>12
3x−3y>2
x, y >0
(3)
As depicted in Figure 6, LP problem (3) has an unfeasible region. However, the teacher
should note that the maximization problem is in fact bounded but, in turn, the set of
maximal solutions is unlimited and corresponds to a ray of the boundary of the feasible
region (see Figure 6(a)). On the other hand, using the Explore mode the students can
Jos´e C. Pereira and Susana Fernandes
(a) LP instance with ilimited multiple maximal
solutions.
(b) Unbounded minimization LP problem in-
stance.
Figure 6: One LP problem instance with an unbounded feasible region. Nonetheless, the
maximization problem is bounded and the set of maximal solutions corresponds to a ray
of the boundary of the feasible region. The minimization problem is unbounded.
verify that there is no lower limit for the optimal value. Shifting to the Min mode they
can then concluded that indeed, the minimization problem is unbounded and no minimal
solution can be attained (see Figure 6(b)).
When introduced for the first time to the notion of unbounded feasible solutions sets,
students tend to believe that this always implies that the LP problem itself is unbounded
as in Figure 6(b) or that, at the very least, the set of optimal solutions must be unlimited
(or maybe just with infinite multiple elements?) as in Figure 6(a). In any case, to help
clarify the relations between these various notions, it is useful to consider the objective
function min z =x+yfor the LP instance (3). In this case, as depicted in Figure 7, the
LP problem has one single optimal solution (x, y) = (4,0), corresponding to the minimal
value z= 4. The teacher should promote a discussion about the relations between this
case and the two previous cases, depicted in Figure 6.
5 FINAL REMARKS
This paper presents an innovative active learning tool for introducing graphical linear
programming, the GLP-Tool. This dynamic, interactive, visual tool was implemented
using the computer algebra system Mathematica. When available in the CDF format, the
GLP-Tool can be used for free as a standalone application by anyone with access to a
computer.
•When introducing the subject of Linear Programming it is rather useful to present
the graphical method for solving a two-variable linear program as this method pro-
Jos´e C. Pereira and Susana Fernandes
Figure 7: Example of an LP problem with an unbounded feasible solution but with only
one optimal solution.
vides valuable insights about the general nature of multivariable linear programming
models. Nonetheless, without a dynamic tool, it is not easy to show students what
happens in a LP problem as constraint boundary lines and objective-value lines
move around on a graphic.
•The GLP-Tool is a visual, interactive, dynamic tool where all the analytical and
graphical information is update in real time. These features render the GLP-Tool
as a fully active learning technical tool that engages students and teachers and
provides them with an effective teaching/learning environment.
•All the functionalities of the GLP-Tool, along with all the displayed information
are designed to conform with the way Linear Programming is presented in the
classrooms both at the high school and undergraduate levels.
•Implemented with Mathematica the GLP-Tool brings together the best of two
worlds: the flexibility of Java visual applets and the sophistication and easy pack-
aging of Wolfram’s Mathematica demonstrations.
•The GLP-Tool has a very intuitive interface that allows even the most inexperienced
user, with no previous knowledge in educational software, to use all its features in
an efficient and autonomous way, right from the start.
Jos´e C. Pereira and Susana Fernandes
•In our future work, we plan to continue improving the GLP, including new fea-
tures such as the graphical visualization of the steps of the simplex method or the
computation of the sensitivity analysis intervals for each coefficient of a problem
instance.
REFERENCES
[1] Bonwell, C. C., Eison, J. A., ‘Active learning: Creating excitement in the class-
room’, ASHEERIC Higher Education Report No. 1, George Washington University,
Washington, DC, 1991.
[2] Prince, M., ‘Does active learning work? A review of the research’, Journal of Engi-
neering Education, ASEE, Vol. 93(3) , pp. 1-9, 2004.
[3] Kydd, C., ‘The Effectiveness of Using a Web-Based Applet to Teach Concepts of Lin-
ear Programming: An Experiment in Active Learning’, Transactions on Education,
INFORMS, Vol. 12(2) , pp. 78-88, 2012.
[4] Concei¸c˜ao, A. C., Pereira, J. C., Silva C. M., Sim˜ao, C. R. ‘Mathematica in the Class-
Room: New Tools for Exploring Precalculus and Differential Calculus’, CSEI2012 -
1st National Conference on Symbolic Computation in Education and Research, Lis-
bon, Portugal, 2012. http://hdl.handle.net/10400.1/1105
[5] Shepard, B., ‘Linear Programming and Pivoting in 2D’, The Computational Geom-
etry Lab at McGill, 2010. http://cgm.cs.mcgill.ca/ beezer/cs601/main.htm Accessed
May 20, 2013.
[6] Hall, J., Baird, M., ‘LP Explorer 1.0’, University of Edinburgh, 2002.
http://www.maths.ed.ac.uk/LP-Explorer/ Accessed May 22, 2013.
[7] Zhang, Y., ‘Graphical Simplex Algorithmv(2D)’,UCMERCED, 2010.
https://eng.ucmerced.edu/people/yzhang/ projects/clientsideLP Accessed May
21, 2013.
[8] Green, L., ‘Exploring linear programming’, Lake Tahoe Community College, 2010.
http://www.ltcconline.net/greenl/java/IntermedCollegeAlgebra/LinearProgramming/
LinearProgramming.html Accessed May 21, 2013.
[9] Kydd, C., ‘Linear programming applet’, University of Delaware, 2010.
http://www.udel.edu/present/tools/lpapplet/lpapplet.html Accessed May 14, 2013.
[10] Wright, D., ‘Animated linear programming applet’, St. Edward’s
University Computer Sciences Advanced Computing Lab, 2010.
http://www.cs.stedwards.edu/∼wright/linprog/AnimaLP.html Accessed May
22, 2013.
Jos´e C. Pereira and Susana Fernandes
[11] Pegg, Jr. E., ‘Graph of Inequalities’, Wolfram Demonstrations Project.
http://demonstrations.wolfram.com/GraphOfInequalities/ Accessed May 22, 2013.
[12] Mukherjee, S., ‘Two-Phase Simplex Method’, Wolfram Demonstrations Project.
http://demonstrations.wolfram.com/TwoPhaseSimplexMethod/ Accessed May 22,
2013.
[13] Bunduchi E., Mandric I., ‘Parametric Linear Programming’, Wolfram Demonstra-
tions Project. http://demonstrations.wolfram.com/ParametricLinearProgramming/
Accessed May 22, 2013.
[14] Boucher, C., ‘The Fundamental Theorem of Lin-
ear Programming’, Wolfram Demonstrations Project.
http://demonstrations.wolfram.com/TheFundamentalTheoremOfLinearProgramming/
Accessed May 22, 2013.
[15] Carducci, O. M., ‘Graphical Linear Programming
for Two Variables’, Wolfram Demonstrations Project.
http://demonstrations.wolfram.com/GraphicalLinearProgrammingForTwoVariables/
Accessed May 22, 2013.
[16] Kragt, M., Jiang, Z., ‘Oil Mallee Farming Opti-
mization Problem’, Wolfram Demonstrations Project.
http://demonstrations.wolfram.com/OilMalleeFarmingOptimizationProblem/
Accessed May 22, 2013.
[17] Boucher, C., ‘A Simple, Standard Linear Pro-
gramming Scenario’, Wolfram Demonstrations Project.
http://demonstrations.wolfram.com/ASimpleStandardLinearProgrammingScenario/
Accessed May 22, 2013.
