long time ago that turns me this sentence from a Article of Hernán Wilkinson Blog:
"To which I want to go and I can hardly yet formally prove is that for me to have a 'good 'model involves having an isomorphism between model objects and the entities of the domain. This means that for each domain entity is a single object that represents "
I leave aside the question of isomorphism with reality , because the issue is best treated roughly especially more from the philosophy that since the computer . However, I believe that this should take the issue of environmental reality, that what you say is worth much if the domain is actually (or part thereof) as if it is not.
A definition of model I spent a long time is very much in tune with Hernan, "A conceptual system (abstract) achieved through representation of isomorphic or homomorphic a particular system, as intellectual construct designed to organize the experience is always partial and provisional, and practically any model is incomplete because it leaves out some properties of the original system, imperfect as the modeling technique introduced due to the support properties model. "François, Charles. Dictionary of General Systems Theory and Cybernetics .
The clarification that the model can be isomorphic or homomorphic is very important, because the isomorphism is symmetric, implying that for every object model , there is a single domain entity which is representation. And this would lead one side to the problem whether entities such as numbers, collections and abstract classes a part of the domain hierarchy. Some would say that the numbers are entities that are part of any domain. And to the extent that the domain is conceived in terms of hierarchy, then the abstract classes are implicit in this hierarchy and that the models in order (if the implementation uses inheritance), which is explicit.
so I would not assert categorically that numbers, sets and other abstractions are part of any domain. I take the example of musical notation, it is clearly a model of music "real." Is an isomorphic representation, as for (almost) every entity in the domain, ie, every note, every instrument, every joint, etc., there is an entity (a chart) in the model. If we had to make a software that models this model is the musical writing, we should look to mathematics as a few times, but you can read music without knowing add. Today I found a phrase from Leibniz in the review of the book Mathematics teacher that illustrates this well: "Music is an unconscious exercise in arithmetic, and that comes with it knows that handles numbers." The numbers are only necessary when for some reason you have to leave that state of "ignorance" for some purpose, how to understand certain laws, certain regularities. Software, for example, must translate the writing musical notes, you must know what is the twelfth root of two, a musician does not need this knowledge to know that two spots in two different positions on paper signify that you have to move a finger from one place to another.
Yes, you can always insist implicit relations. I prefer to think that rather than take something that is there, what we do is put something that is in us: to interpret.
In this case, both large abstractions, abstract classes, together with the emerging deployment needs only and all the paraphernalia of objects may have a program, would be covered in this "semantic gap" mentioned in the definition of systems theory. Are all "property due to the support model."
homomorphism In this case the model is rather a kind of utopia, wishful thinking, a desire for something more precise accuracy. But then it would not be expressed formally.
Another issue I think we have to consider is that many times, perhaps the majority, which is modeled is itself a model. If we take some of the classic examples of the kinds of programming and design of systems such as bank accounts, or book loan schemes, it is clear that these systems and are themselves a model. There is no single savings accounts relating to customers, there is a model designed for money to flow according to certain rules. The problem is that is naturalized, reified, (it is considered a given reality) rather than actually built a (model). It seems pretty clear that the "realities" to model has different degrees of abstraction, or if like, different distances from noumenal reality. These computer models would then "meta" levels with varying amounts of "meta." Sure, there will
ejemplos de modelos informátios que no modelan otros modelos. Tal vez sea el caso de los sistemas de simulación, que si bien pueden incluir partes que son metamodelos, el sistema total no lo es, porque lo que se aspira es que el modelo conceptual surga de algún modo de la simulación.
Revisando los usos de la palabra "modelo" encuentro que en la gran mayoría de los casos se trata de modelos teóricos; como dice la definición, sistemas abstractos que representan sistemas concretos. Excepto los " Toy models " o los modelos del aeromodelismo o similares, los únicos modelos que además de representar algo, hacen algo, son los modelos computacionales. El modelo atómico does what the atom, or make anything, just looking to represent. To understand, to teach. Instead the software is to model other models is that models of reality, models always doing something. Of course, this model can also be used to understand or to teach, but what is that never fails to .
I think this emphasis on making is the key to understanding how it differs from the paradigm of other objects, which reduced to mathematical or logical abstractions whatever you want to model, while objects are modeled from the tool itself, which is the do . For modeling a do is nothing to model behaviors . One way to insist that objects that are defined by their behavior, and boldly said in the previous article r. I have no theoretical tools to address these concepts in depth, I do not know if any. But I do know that in many other areas of emphasis in practice this is quite recent. I think for example in Roger Chartier, who has long studied the history of books, publishing, construction, not the traditional way, as the study of techniques and ideas, but as a practical study. Instead of the history of the work raises the History of Reading. This is worth at least as evidence that the object paradigm is in keeping with the spirit of the times.
Back then the problem of principle could say that a good model (object) is a model in which there is a homomorphism between the behavior of the entities of the domain and the behavior of objects. The difference in this definition is that now is not necessary correspondence between institutions but between behaviors. Two or more objects can do what a domain entity, object behaviors can subsume several entities of the domain. A good knife of Occam to avoid considering the abstractions that we created as part of a domain that did not know, o recíprocamente, evitar incoporar al diseño objetos que representan entidades que sólo tiene distintos nombres, pero que se comportan del mismo modo.
Todo esto dicho también en forma osada y sujeto a revisión.
0 comments:
Post a Comment