Originally posted by: zsg10071086

• Hi, everyone. I meet a problem in my research about OPL studio. I want to do rolling horizon which is also can be regarded as multi-stage problem. In each stage I need to output the results, and part of the results need to be input in the next stage to do optimization. I also want to put the input data and output data in the same .dat file.  That is, after each stage, the results of a decision variable ( It is a large sum of number) is output in the .dat file to cover (instead of) the input data of the same decision variable.  any one know how to fulfil this?

  Thanks in advance 

Originally posted by: davidoff

This is indeed possible through flow control (script using a main function)

You can begin to look at some examples given in OPL , such as mulprod project for basic changes (look at mulprod_main.mod),

Then more advanced scripts are illustrated in the cutstock project that implements column generation

David

Originally posted by: zsg10071086

Dear David,

Thank you for the attention. I have read the mulprod example. In that example, only one value is needed to be changed in each iteration. But in my case, in the previous iteration, I output a matrix for a decision variable (a large number of value), and these output need to be used as input in the following iteration. Till now, what I plan to do is: output the results for each iteration in a .txt file separately, and then copy the data of the previous stage from the .txt file to a .dat file to run the current iteration. But the current data need to cover the data of the previous stage. So till now, I have not solve this. I wonder whether there are some examples exactly done something like this. Any help is welcome.

Originally posted by: davidoff

one matrix for each decision variable ?

Or you mean the matrix related to a set of decision variables values ?

Originally posted by: zsg10071086

I mean the variable is like this: Xe, where according to different e X has a large sum of number. The decision variable is one-dimensional array. Thanks.