To bring the power of your preferred solver
to your preferred programming environment

Q.1 Why use EZMod instead of AMPL or GAMS?
Q.2 Why use EZMod instead of calling directly the solver callable libraries (dll)?
Q.3 How does EZMod bring new perspectives?

• AMPL or GAMS are based on specialized programming languages and are intend for prototyping phase. New AMPL and GAMS users should then learn new programming languages in order to model and solve mathematical programming problems.

• On the other hand, the users of EZMod are welcome to continue to work with their preferred programming environment, be it Microsoft Visual Basic, VisualC++, Borland Delphi, C/C++ or C++ Builder, or any programming environment that allows calls to DLLs or accepts Active X components.

• EZMod is not a new Programming language. EZMod is a library of functions (DLL) or objects (Active X) that allows the user to specify the structure and the data of an optimization problem, to access the resources of the solvers (DLLs or executables) (currently CPLEX, XPRESS-MP and FRONTLINE SYSTEMS) and to recover various results and statistics.

• EZMod enables Operations Research and Management Science professionals as well as management consultants to combine with extreme ease the flexibility of visual programming environments and the power of optimization resources.

• EZMod completely frees the Operations Research and Management Science professionals, and management consultants from the stringent requirements of the solvers such as heavy matrix problem description, the order of problem statement, and the like. So the users of EZMod do not need to be experts or even familiar with the technical details of the solvers they intend to use. EZMod automatically takes care of all the interactions necessary with the solver to be used.

• EZMod offers the facility to easily and intuitively state optimization problems, to solve them and get the solutions. The statement of optimization problems can be made step by step in the most appealing order to the user, rather than being obliged to follow the rigid requirements of the solver in use.

• With EZMod, Operations Research and Management Science professionals need to work only with variable and constraint names that are meaningful to them, rather than with the matrix and vector indices required by most solver callable libraries.

• EZMod submits your problem, as you formulated, to one of the following solvers of your choice: CPLEX, XPRESS-MP and FRONTLINE SYSTEMS. This current list of supported solvers will be enlarged by adding other popular software.

• EZMod also permits to switch from one solver to another within the same application with extreme ease, if this need be.

• EZMod has a very important and useful added main feature compared to the current solvers in the market: it can be used to solve Quadratic Mixed-Integer Programming (QMIP) problems! With its most economic and popular linearization method, EZMod is equipped to solve large size QMIP problems efficiently.

• EZMod frees its user from the heavy matrix problem description.

• EZMod submits transparently your problem to one of the supported solvers. This allows a total flexibility in the implementation of the mathematical model.

• EZMod allows the user to describe mathematical problems step by step using problems name (or identifier), constants names (or identifier) and variables names (or identifier).

• By making the modeling and solution process solver independent.

• By making the programming toolkit programming environment independent.

• By simplifying the development of "Event-Based Models".  Such models are constructed dynamically as events, conditions and parameters arise during runtime.  For example, you can easily modify and extend your problem's specification and data in response to a button-click event without worrying about the impact of such modification on the problem formulation, for EZMod will make all the required adjustments before submitting it to the solver.

Example to illustrate the question Q.3:

Consider the following LP problem:
Let us name it "LPExample" and identify it by <PbId>

This model is subject to a condition that adds a set of variables to the objective function and to all the constraints.

We let you think about the way to design an application running this model and using the conventional tools.

We give you the solution with EZMod (Subroutine "Condition").

' Defining the Problem Identifier as Long and Public
Public PbId As Long

Private Sub GenerateModel(m As Integer, n As Integer, C, A, B)
  ' Defining indices
  Dim i, j As Integer

  ' Defining the returned status of the functions as Long
  '(available for almost functions)
  Dim status As Long

  ' Opening the problem LPExample
  PbId = ezmOpenPb("LPExample")

  ' Define the optimization criterion as Maximize with
  ' returning the status
  status = ezmOptCri(PbId, EZM_MAX)

  ' Describe the linear terms Yi of the objective function
  For i = 1 To n
    status = ezmCCoef(PbId, "Y" & i, EZM_SET, C(i))
  Next i

  ' Describe the constraints
  For j = 1 To k
    ' The sense of the constraints "Equal ..."
    status = ezmConType(PbId, "Con("&"j"&")", EZM_EQCON)

    ' The right-hand side
    status = ezmBCoef(PbId, "Con("&"j"&")", EZM_SET, B(i,j))

    ' For every Constraint The left-hand side (Yi, i = 1 ... m)
    For i = 1 To m
      status = ezmACoef(PbId, "Y" & i, "Con("&"j"&")", EZM_SET, A(i,j))
    Next i
  Next j

  ' Specify the type and bounds of the variables Yi
  For i = 1 To n
    ' as "continuous variables"
    status = ezmVarType(PbId, "Y" & i, EZM_CONVAR)

    ' Specify the lower bound of the variables Yi
    status = ezmLwCoef(PbId, "Y" & i, EZM_SET, Lw(i))

    ' Specify the upper bound of the variables Yi
    status = ezmUpCoef(PbId, "Y" & i, EZM_SET, Up(i))
  Next j

End Sub

' Under a specific condition, the model must involve a new set of variables
' Here is the subroutine "Condition"
Private Sub Condition(p As Integer, D)
  ' Defining the returned status of the functions as Long
  Dim status As Long

  ' Add the Xi to the objective function
  For i = 1 To p
    status = ezmCCoef(PbId, "X" & i, EZM_SET, D(i))
  Next i

  ' Add the Xi to the all the constraints
  For j = 1 To k
    ' For every Constraint add Xi, i = 1 ... p)
    For i = 1 To p
      status = ezmACoef(PbId, "X" & i, "Con("&"j"&")", EZM_SET, E(i,j))
    Next i
  Next j

  ' Specify the type and bounds of the variables Xi
  For i = 1 To p
    ' as "continuous variables"
    status = ezmVarType(PbId, "X" & i, EZM_CONVAR)

    ' Specify the lower bound of the variables Xi
    status = ezmLwCoef(PbId, "X" & i, EZM_SET, Lwx(i))

    ' Specify the upper bound of the variables Xi
    status = ezmUpCoef(PbId, "X" & i, EZM_SET, Upx(i))
  Next j

End Sub

And that's all!

' Here is the subroutine to solve the described LP problem
' It uses the same problem identifier (PbId) defined as Public

Private Sub Solve()
  ' Defining the returned status of the functions as Long
  Dim status As Long

  ' Test the condition
  If Cond = TRUE Then Condition(p, D)

  ' Solve the problem LPExample using CPLEX Solver
  status = ezmSolve(PbId, EZM_LP, EZM_CPLEX)

End Sub

' Here is the subroutine to get the solution of the LP problem
' It uses the same problem identifier (PbId) defined as Public

Private Sub Results()
  ' Defining the returned status of the functions as Long
  Dim status As Long

  ' Getting some results (The objective value, and X1 value)
  Text1.Text = CStr(ezmGetZ(PbId))
  Text2.Text = CStr(ezmGetX(PbId, "X1"))

  ' Get the solution with all the parameters and variables
  ' in a text file
  status = ezmWritePb(PbId, "C:\PbSolution.txt")

End Sub