Interesting topic. Follow.

Dear Boris

You say 'The use of optimization models is most effective in determining the General parameters of the physical phenomena'

I am afraid that I disagree.

MCDM models do not determine the general parameters. What they do, is that considering the existing parameters of a problem or scenario, they try to find

the best solution, and THIS DOES NOT HAPPEN with present MCDM models, simply because not all existing parameters of the problem ar inputted, mainly because MCDM methods don't contemplate that possibility. They work with only a limited kind of parameters (I would instead use the word 'characteristics'), and IGNORE the others.

You are absolutely correct when saying that these models GREATLY SIMPLIFY reality, precisely because they do not consider all parameters, and/or by assuming assumptions that may be very reasonable, but also highly subjective.

Sometimes, simulation is very useful for generating data, and really I don't see how the MCDM may be the stepping stone for simulation.

True, you can simulate criteria weights, performance factors, resources etc., but the simulation is a tool, not a method.

I have never seen a simulated MCDM model, although perhaps it may exist.

Dear Nolberto Munier ,

Thank you for the detailed, clear and short answer.

I want to draw your attention (very briefly, because I really do not like the long and confusing discussions in the RG).

It can be shown [https://www.researchgate.net/publication/341208074_Reviewers_1_Sie_Long_Kek_Universiti_Tun_Hussein_Onn_Malaysia] that the concept of comparative uncertainty [L. Brillouin, Science and Information Theory, Academic Press, New York, 1953.] can be used as an universal criterion for the optimality of a model of a physical phenomenon, a process for measuring a physical constant (a "quantity" in accordance with SI requirements), or a technological process. The magnitude of the comparative uncertainty depends on the qualitative composition of the base quantities from SI and the total number of variables taken into account.

BR

Dear Boris

Thank you for your documented answer.

I perused your paper, and although I am familiar with the concepts and formulas expressed there, I am not match for you with your ample knowledge, and in addition, it is not a field of my expertise.

It appears that we are in two different universes, you refer to quasi philosophical concepts, when I work with such concepts as homologated and documented prices, volumes, with established environmental limits and indicators which are the bread and butter of MCDM.

Of course, there are also many qualitative and uncertain concepts, but there exist also mathematical tools to deal with them.

You speak in your paper about ‘threshold mismatch between the model and the phenomenon’, and you are right, because this is the way to appraise the validity of a method, for instance ‘The Gravity Model’ or the deviation of light according Einstein’ Theory of Relativity. In these two examples you can determine that difference, because in both you can compare your results with something that you know or that exists.

However, in MCDM you can’t, simply because you do not have a yardstick, something to compare with. Think that if we knew it, then we wouldn’t need any MCDM model. In here, we have complete uncertainty, the obtained result could be close to reality or far away, or something in between. This is the reason which makes MCDM very difficult.

Dear Nolberto Munier ,

In the question and my answer, I touched ONLY the topic of choosing the optimal model of the PHYSICAL phenomenon, in which variables from SI or other measurement systems can be used. This fully applies to theoretical and experimental physics and engineering, and not to quasi-philosophical concepts.

I am familiar with [L. N. Bramley, K. Kopp, K. B. Korb, Cutting through a Tangled Network: An Information-Theoretical Perspective of the Information War, Air Power Australia Analysis, 2 (2012). http://www.ausairpower.net/APA-2012-02.html.], which proves that any decision-making mechanism is inherently limited by the behavior of collecting and processing information from the system of which it is a part. However, these authors propose a strategy for choosing a not optimal, but rational approach.

You are a specialist in the Multi Criteria Decision-Making (MCDM) discipline. I wish you success in this work and in your quest to reduce "total uncertainty" and achieve a result close to reality.

BR

Boris

Thank you very much Boris

Dear Boris

Rereading your last comment I realized about a very important comment: '.....which proves that any decision-making mechanism is inherently limited by the behavior of collecting and processing information from the system of which it is a part. However, these authors propose a strategy for choosing a not optimal, but rational approach'

What these authors assert is something that I, without knowing this publication, have also said and insisted in my comments in RG and papers. However, I believe that that assertion is too general and is missing something.

Of course, collecting and analyzing information from the system is inherent limited by the behaviour of the DM. This information, the facts, the characteristics of a scenario are there, for everybody to be seen, although perhaps not at first sight and without an investigation. If the DM does not use it, is his/her problem and decision. I think that most DM do their most in trying to understand a scenario and try to model it the best they can.

However, it also could be that the DM is aware of these different issues, worries about them, but has his hands tied because there is no MCDM method that can handle them. I think that practitioners will agree that most MCDM methods can't model and out-of- the - ordinary situation.

For instance, how many methods can model alternatives preferences, inclusive/and exclusive alternatives, binary variables, initial decision matrices with formulas, decimal or integer values, multiple scenarios, conditioning between alternatives and criteria, for instance establishing that each alternative must comply at least with a certain number of criteria, etc.?

Probably none.

THIS IS THE PROBLEM. Practitioners are not to be blamed by getting dubious results because not all facts have been considered. They do what they can with the available existent MCDM tools. When you read that according to some authors AHP can solve complex problems, it is hard not to scorn at that. Why?

Simply, because AHP with its hierarchical structure was not designed to solve that kind of problems. Can you imagine deciding to build a bridge or a tunnel based on a hierarchical structure?

Something similar happens with other methods not necessarily related with a hierarcbical structure

Dear Nolberto Munier ,

You raised important decision-making issues.

I am a practitioner (I invented Pumpable ice technology - see WIKI - as applied to cooling seafood and energy storage systems). Therefore, "I pull the blanket to my side."

If I understand correctly, your important comment: “..... understand the script and try to model it as best as possible” contains a lot of shades.

1. At present, a lot of researchers present data that are difficult to challenge or confirm. As a refrigeration engineer, I checked 1,500 articles in the International Journal of Refrigeration in recent years [B. Menin, “Uncertainty assessment of refrigeration equipment using an information approach,” Journal of Applied Mathematics and Physics, vol. 8, no. 1, pp. 23–37, 2020. Available: https://www.scirp.org/journal/Paperabs.aspx?PaperID=97483]. Only 30% (!) of the articles had a comparison of experimental data, theory and the model uncertainty achieved simultaneously. Most of the work cannot be put into practice. "THIS IS THE PROBLEM." Researchers need to be blamed for dubious results; it's their choice.

2. A similar situation in other disciplines: in psychological science, metallography, physics, life sciences, economics.

I only touched on topics that concern me.

I outlined a gloomy picture, although I always hope for the best.

Sincerely yours

Boris

Dear Boris

Well, I don't know about refrigeration because it is not my area.

However, what I do believe, is that are practitioners that know they are using methods that are inappropriate for a problem, they know it, and they don't care; the aim is being published and if possible cited. However, again, we can't blame them because as I said, no MCDM method is capable at present, to model and solve even certain simple characteristics of a scenario.

We should start by taking conscience of this fact and try to develop models, which, even incomplete, even with drawbacks, allow for at least a decent representation of reality. Just as a analogy with your area, it would be the same if nowadays refrigeration equipment were using ammonia, CO2 of Freon 12.

For that reason, I am pessimistic about the future.

Just think, from the 60s, with the first developments in MCDM, and six decades later, what improvement has been done? None. Remember that fuzzy does not count, since it aims at improving data reliability, not procedure.

Methods such as AHP, that was developed with a purpose, with a hierarchical structure, and which was perfectly applicable to the military and to lineal organizations, continue being used, when those types of organizations, when that hierarchical structure do not longer exist.

When I ask a practitioner about that, no one was able to give me a coherent reason. And of course, AHP is not the only one, especially nowadays that many methods use weights computed by AHP.

The best model of the world is the world itself, and as we are unable to really understand that model, we need ( a variety of ) simpler models, based on theory , measurements and experience . Modeling is an art, that only improves by practicing . There is not one perfect model, there is only hard work and modesty.

Relevant limitations as well as the corresponding functions of goals should be covered.

And, when construction a model, you have to determine, what purpose the model is meant for. Would you like to predict, for next month, next day, would you like to understand that part of the world, you are modeling, would you like to optimize something , are there important restriction involved, and so on.

And, try to use relevant theory, so, for instance, when you would like to model a pendulum, then you have to use newton's laws, when you have to model a spacecraft, use the insights of Einstein, when you would like to understand part of the economy , consult experts in economics, try to ensue a combined effort, so, try to learn from the disciplines you are not familiar with, Always be humble, because " the world" will always surprise you!

Dear J.W.

Wise words!

Mother Nature has found along millions of years the best model and the best solution for everything.

It is true that practicing is fundamental, but it is not always possible to model a scenario.

In my opinion, sometimes, it is not a problem of expertise but the availability of tools to build a model, and that is something missing in practically all MCDM methods. Of course, we are not Nature, and can't pretend to model something in a 100 %, but the problem is that we don't even intent to improve our methods. We believe that they are OK, and they are not. At present we are only able to model poor representations of reality, and improving data.

Ironically, the very large advantage that we have is that we don't know which the best solution is, and then it is possible for us to say that we succeed and found the best, and in some cases we even say that we found the optimal solution.

Anyway, who is going to refute it?

If we could model a scenario taking into account ALL of its characteristics, we could assume, if we use a RATIONAL MCDM model, and without personal interference with the data, that most probable we got a good result. No more that that!

Dear J.W.

You again hit the nail!

I agree 100% with all your statements, and I believe that people should take into account, that whatever method they use, that it is not perfect, it is only an approximation. Instead if blindly following instructions from the method developer, we have the obligation to ask ourselves if a certain procedure is right, if is reasonable, or if it is free assumption from the model builder, which does not look right. Don't be just a follower, use your brains, and be guided by your common sense.

We can help a lot if we criticize a method, pointing to its drawbacks and trying to improve it, even with suggestions to its creator.

We should not use methods because a friend used it in a similar problem, or because we have the software, or because it is easy to learn. Normally complicated problems can't be solved by ready to use formulations

We need to study the scenario, knowing exactly its characteristics, getting data, and only then to decide which method to use, independently if we like it or not.

Dear Kosta

I couldn't say it better!

The problem is that if there is a limitation in a scenario and a method can't model it, why to bother? Just ignore it, it is easier!

If a goal can't be properly addressed, don't worry, nobody will notice!

If a practitioner believes that he/she reached the best solution, well, it may be a personal satisfaction, but a very bad conclusion.

Now, if a practitioner follows these absurd procedures, if he/she does not want to use his/her capacity for researching, for reasoning, well, probably he/she is in the wrong business.

If somebody writes a paper expressing her/his own ideas, he must explain and justify them. MCDM is not a novel, there is not a plot, where the writer can write whatever she/he likes.

If he/she is not sure about his/her ideas it is better to keep silent, or make sure that they ready to support them.

Dr. Boris Michailovich Menin, I have an example where the criteria are given with symmetry arguments, relying on the thermodynamics of the free energy of a physical system, the phase transition theory of Acads. L. Landau * and E. Lifshitz (1937):

• Ferro/Antiferromagnetics,
• Ferroelectrics.
• Superconductors.
• Superfluids.
• Liquid crystals.

Please check the Wikipedia commons article, and also the original paper.

https://en.wikipedia.org/wiki/Landau_theory

There is a link to the original article:

* L. Landau Zh. Eksp. Teor. Fiz. 7, pp. 19–32 (1937)

Dear Pedro L. Contreras E. ,

They say that at the seminar Landau was only one smart - himself.

With Zeldovich, everything was the other way around - everyone could express the most "stupid" ideas.

From my hump (:-)) Brillouin seems to have presented the best criterion for assessing the plausibility of a model describing a physical phenomenon in the micro- and macro-world: comparative uncertainty.

Until now, it has not received sufficient attention. ALL use the relative uncertainty that subjectivity is inherent in.

The proximity of the calculated comparative uncertainty to the optimal one confirms the validity of the model.

Unfortunately, my arguments do not convince the CODATA experts of the need to apply comparative uncertainty. CE LA VI.

Thank you, Dr. Boris Michailovich Menin.

Thought teachers during USSR times.

Thank you for pointing out, Brillouin works. I will check your work:

Article Information Measure Approach for Calculating Model Uncertain...

Perhaps of some interest for your query:

https://www.researchgate.net/post/how_to_explain_the_difference_between_simulation_and_experiment_result

You may want to check my post elsewhere at this forum ― as a case study related to transient mass balance analysis:

https://www.researchgate.net/post/Hello-everyone-can-you-please-help-me-answer-this-questionWhat-is-the-steady-state-concentration-of-the-toluene-in-the-individuals-vascular-system

Dear Carlos Araújo Queiroz

Thank you for your link.

I would like to suggest the another:

Article Can Information Theory Help to Formulate an Optimal Model of...