Hello to everyone,
I am trying to use the policy optimization tool but I do not have valid parameters to use (I belive that it is because I have the "not valid optimization instructions found" error).
I am not sure but I think that the only valid parameters are constants. The issue is that in my model I need that Vensim change some lookup functions in order to minimize the payoff.
The only characteristic of my lookup functions is that they should not exceed certain min and max values.
I try to imagine an example and this is the best I could think of: Suppose you have to minimize the total time of making a car. There are some steps to be done secuencially for finishing it. The time that takes each step to be completed depends on how many workers there are assigned in it. However, the constrain is that you have limited workers (supposing that everyone is capable of doing every process)
So I will have to find how many workers I would assign on each step in each moment --> lookup function y= f(x) where the variable x is the time and the variable y is the total number of workers in the activity y.
So based on my preview example, what I should do in Vensim is mimimize the total time varying the number of workers assigned in each step. That´s what I do not know how to do it.
If you have any suggestions or commentaries, they are very welcome.
Thanks guys,
Optimization issue
Optimization issue
Felipe Venegas
optimization
Hi
If you want to optimize a lookup, you will have to use a vector lookup and optimize the values of the vector.
What you describe with your example, is finding a rule that automatically adjusts the number of workers at each time step or after a longer time. I am not sure that a lookup is the best solution in that case because the policy will be always the same and not changing from the past results obtained. Obviously the number of workers must not only be based on the time and the total workers available. There are interesting examples in the Vensim documentation based on project management that you should study.
Joined a simplistic diagram without any lookup that has the work done at each time step exogenous (in blue) and that one must find some way to evaluate if one wants to set a goal of duration. One way is to estimate a stable productivity that calculates the work done and one can then optimize the goal duration. This is rather simplistic but illustrates that a lookup function might not be the solution.
The diagram was built quickly and may be bugged.
Regards.
JJ
[Edited on 17-12-2009 by LAUJJL]
If you want to optimize a lookup, you will have to use a vector lookup and optimize the values of the vector.
What you describe with your example, is finding a rule that automatically adjusts the number of workers at each time step or after a longer time. I am not sure that a lookup is the best solution in that case because the policy will be always the same and not changing from the past results obtained. Obviously the number of workers must not only be based on the time and the total workers available. There are interesting examples in the Vensim documentation based on project management that you should study.
Joined a simplistic diagram without any lookup that has the work done at each time step exogenous (in blue) and that one must find some way to evaluate if one wants to set a goal of duration. One way is to estimate a stable productivity that calculates the work done and one can then optimize the goal duration. This is rather simplistic but illustrates that a lookup function might not be the solution.
The diagram was built quickly and may be bugged.
Regards.
JJ
[Edited on 17-12-2009 by LAUJJL]
- Attachments
-
- rule optimization.mdl
- (4.97 KiB) Downloaded 267 times
JJ,
Thank you very much for your commentaries and for the example that you attached. My model is somehow similar to yours because in mine (a brief model is attached to this message) my goal is to minimize all the work that is still to be done in every step (in red).
In my initial scenario I suppose there is a specific number of workers needed at each step depending on the work to be done in each step (in blue)-->Done by a lookup function. What I want to evaluate if varying that lookup function that tells me how many workers should be assigned for working at each step so I can improve my payoff.
I will try to use the lookup vector to do that as you suggested me but I am not sure how to do that. I will revise the literature and manuals that specify how that function works. Thank you for this huge tip.
If anybody has any suggestions or commentaries, they are very welcome!!
PS: Can you please tell me which are those interesting examples of project management or where can I find them?
Thank you very much for your commentaries and for the example that you attached. My model is somehow similar to yours because in mine (a brief model is attached to this message) my goal is to minimize all the work that is still to be done in every step (in red).
In my initial scenario I suppose there is a specific number of workers needed at each step depending on the work to be done in each step (in blue)-->Done by a lookup function. What I want to evaluate if varying that lookup function that tells me how many workers should be assigned for working at each step so I can improve my payoff.
I will try to use the lookup vector to do that as you suggested me but I am not sure how to do that. I will revise the literature and manuals that specify how that function works. Thank you for this huge tip.
If anybody has any suggestions or commentaries, they are very welcome!!
PS: Can you please tell me which are those interesting examples of project management or where can I find them?
- Attachments
-
- rule optimization 2.mdl
- (10.87 KiB) Downloaded 254 times
Felipe Venegas