cancel
Showing results for 
Search instead for 
Did you mean: 

TP/VS Run Time

Former Member
0 Kudos

How do you calculate the default run time for TP/VS optimizer?

Is the logic for deciding the run time based on transaction data volume (number of orders / deliveries) or it also considers the master data volume for T-Lanes, Incompatibilities, Material Location combinations, Location Master, Resource Master etc ?

How do you calculate the ideal run time?

Should a higher run time always give better result? Can it give worse result than the lower run time in certain cases? If yes, why?

Sunil Madhogarhia

Accepted Solutions (1)

Accepted Solutions (1)

Former Member
0 Kudos

Hi Sunil,

The maximum runtime depends on the Variables that are been handled by the optimizer.The more complex your model is the more number of variables would be there to optimze.

The ideal time for runtime can be arrived by experience where you gradually keep on increasing the time, and observe the log. If after somtime you find that the iterations are not improving the result of the optimizer much you can stop and that is the optimum run time.

Thanks,

Harsh

Answers (1)

Answers (1)

Former Member
0 Kudos

Hi,

Its not calculated and is generally given a default value to stop in case the optimiser does not arrive at a solution an goes into an infinite loop. The time limit actually ends the optimiser and gives the solution that is has calculated last.

I have to explain how a optimiser works in detail. The optimiser runs by starting at a base solution say 0, 0, 0 ) and calculates costs it changes the solution from (0,0,0) to (a,b,c) if the costs are less at ( a,b,c) in the same way it checks at all possible solutions. Sometimes it may happen it may run into loop and it runs enlessly even though the solution is not getting any better and hence needs to be stopped.

The higher runtime may not result in better solution but it should be sufficiently large so that its not interrupted when not in loop, i.e should be greater that the time it generally finishes.

Regards,

Sireesh

Former Member
0 Kudos

Thanks Sireesh. Can you further clarify how to fix the ideal run time for a specific optimizer profile.

Former Member
0 Kudos

Hi Sunil,

Dont mention anything as maximum run time... System will automatically terminate after finding optimul solution. And if its fall in a loop, then maintaining maximum run time also does not make any sense. Maximum run time will be ignored. If system does not find optimul solution the its ignore maximum run time. Its betteryou maintain maximum no. of improvement.

Thanks,

Satyajit

Former Member
0 Kudos

Hi Satyajit,

Pls clarify :

1)"If system does not find optimal solution, it will ignore max run time" Does this mean that it will keep running even after the run time maintained in the VS optimizer profile? I think there is nothing called optimal solution from the optimizer's point of view, there definitely is best possible solution that it could find within the max run time given to it.

2)How do you maintain maximum no. of improvement? I mean, where in the optimizer profile do we maintain this?

Regards...Sunil

Former Member
0 Kudos

Hi Sunil,

You are right. In TP/VS, the optimizer will run as per the run time specified or the number of iterations specified. It will not terminate when the optimal solution is found. The optimizer uses EVS (Evolutionary Search) method to arrive at the solution. You can decide the optimum run time only by trial and error over a period of time. There is no specific guideline on this. You can observe the cost graph for this purpose. After certain amount of time, the cost graph will not show further improvement and this may be the ideal run time.

Regards

Sandeep Patil

Former Member
0 Kudos

Sunil,

What I stated earlier based on snp optimizer. I was not aware of different rule for TP/VS optimizer. I thought the basic thing would be same. In snp optimizer if you put max run time 10 mins, and after that time period optimizer doesnt find any suitable solution then will not be terminated. It will try to find out atleast one fesible soilution.

And no. of improvement plays a role in discreate optimizer process. It may act as 'heuristic is a first solution' then its keeps trying to improve result. If you put max no. of improvement '10', then it will re-calculate the solution untill its reach 10th no. of improvement of result (most optimal).

Thanks,

Satyajit