(3) The most useful models will be those that include parameters and state variables that can be readily determined by simple measurements or observations,
(4) Existing hypotheses/models are a useful resource, but they need to be evaluated
in view of objectives of the current model construction (Sinclair and Seligman
1996).
2.2.2 Quantitative Description of Hypotheses
Qualitative hypotheses need to be expressed as mathematical functions. Equations
need to be developed to express how each hypothesis can be expressed in the
model system. The interaction of hypotheses also needs to be taken into account,
which often introduces a whole new layer of assumptions. Commonly, it is
assumed that there is no interaction among hypotheses other than what has been
explicitly defined.
The quantification of the model can be the most challenging because it requires
a thorough knowledge of the system being modeled, and an understanding of the
most relevant relationships. Additional functions can always be added to the
model, but do these functions enhance the performance of the model for the stated
objective? Often, the most critical phase of the model construction will be
selection of the quantitative functions that sufficiently and efficiently describe the
model component being modeled. Assembling equations without understanding
and evaluating their relevance to the objectives overlooks a critical aspect of the
modeling (Soltani and Sinclair 2012).
2.2.3 Programming
Once the hypotheses have been defined and quantified, the model is finally constructed into computer code. That is, the knowledge and insight about the system
should have been captured and it should be a straightforward, even a trivial task, to
translate the hypotheses into computer code. However, considerable care is
required to accurately express the model in computer code. It is necessary to verify
the computer algorithms and the codes are correct from mathematical relationships
defined. Preparing computer programs usually requires debugging to eliminate the
errors that arise during programming.
The program code can usually be organized in individual sections as represented in a flow diagram. In more expensive models, it is often useful to structure
the code so that each section is placed into its own submodel. Placing parameter
values for the various functions in their own separate, initialization section can
facilitate the use of the model in simulations using different parameters. That is,
code the functions in the model using parameter names, and then define all
parameters at the beginning of the program. In this way, simulations of other
54
M. A. Vázquez-Cruz et al.
(4) Existing hypotheses/models are a useful resource, but they need to be evaluated
in view of objectives of the current model construction (Sinclair and Seligman
1996).
2.2.2 Quantitative Description of Hypotheses
Qualitative hypotheses need to be expressed as mathematical functions. Equations
need to be developed to express how each hypothesis can be expressed in the
model system. The interaction of hypotheses also needs to be taken into account,
which often introduces a whole new layer of assumptions. Commonly, it is
assumed that there is no interaction among hypotheses other than what has been
explicitly defined.
The quantification of the model can be the most challenging because it requires
a thorough knowledge of the system being modeled, and an understanding of the
most relevant relationships. Additional functions can always be added to the
model, but do these functions enhance the performance of the model for the stated
objective? Often, the most critical phase of the model construction will be
selection of the quantitative functions that sufficiently and efficiently describe the
model component being modeled. Assembling equations without understanding
and evaluating their relevance to the objectives overlooks a critical aspect of the
modeling (Soltani and Sinclair 2012).
2.2.3 Programming
Once the hypotheses have been defined and quantified, the model is finally constructed into computer code. That is, the knowledge and insight about the system
should have been captured and it should be a straightforward, even a trivial task, to
translate the hypotheses into computer code. However, considerable care is
required to accurately express the model in computer code. It is necessary to verify
the computer algorithms and the codes are correct from mathematical relationships
defined. Preparing computer programs usually requires debugging to eliminate the
errors that arise during programming.
The program code can usually be organized in individual sections as represented in a flow diagram. In more expensive models, it is often useful to structure
the code so that each section is placed into its own submodel. Placing parameter
values for the various functions in their own separate, initialization section can
facilitate the use of the model in simulations using different parameters. That is,
code the functions in the model using parameter names, and then define all
parameters at the beginning of the program. In this way, simulations of other
54
M. A. Vázquez-Cruz et al.