Ventana software support forum

I'm dealing with a particularly infuriating instance confirming your hypothesis about computers at this very moment. Question is, what to do about this behavior in cases where it's a problem. I hate to create a decimal type, due to the potential performance issues. One option would be to create some...

Table functions (called Lookups in Vensim) are available: http://www.vensim.com/documentation/usr08.htm

You should be able to launch 2 copies of the program. Higher versions of Vensim permit opening multiple models within a single application.

Here's one quick thing to try: if your mdl.bat uses the compiler flag /O2, try removing it. This reduces the code optimization, making compiling faster, sometimes but not always at the expense of some speed. We'd be interested to see the model via email though.

Hi JJ - Just to be sure I understand - are you comparing the speed of 1000 iterations of essai_vitesse to a single run of essai_vitesse2? - Tom

The reason you don't get zero is that the binary representation of a decimal is not always exact. A double is a binary floating-point representation, as in http://msdn.microsoft.com/en-us/library/e02ya398.aspx You'd get the same answer in a programming language - Java for example: double x = 1.0 - 0...

What file could not be opened? The Vensim file format has not changed, and version 6 is fully backward compatible with 5.x. If the model is the issue, you could try using .mdl format, though there shouldn't be any problem with .vmf. The problem with the 5.7 mdl.bat is likely from the link step, LINK...

The optimizer is only statistically likely to find the optimum if you use multiple start for long enough. This is true of all global optimization approaches that can't exploit problem-specific information. If your payoff surface is rough (many local optima), other approaches might be better, includi...

Actually I see that you're using XP so I tried that and I can replicate the crash. It works fine on Win7 and Mac. Your model generates floating point errors in its base run. Resolving those (try a much smaller time step for one thing), plus adding parameter bounds, will solve the problem, though it ...

I assume that you mean a Custom Graph, in which case I believe that the answer is no.

If you mean the model diagram, the comment and units appear when you hover over a variable with the cursor.

This works fine for me on Win7 6.1c - I assume you're on a Mac though? I'll move it over and give that a try. One thing that would help quite a bit here is to put upper and lower bounds on all your variables. Even if you can't determine narrow limits a priori, searching from 0 to 1 billion is a lot ...

This may be a matter of 32 vs 64 bit system file locations. You probably need to check the paths for the calls to the compiler & includes in your mdl.bat.

An upgrade to the current version of Vensim would almost surely solve your problem. 5.7 is six years old and predates Win7.

We need to know a little more - can you post the model and your control files? Upgrading to 6.1 could help, though I don't recall any optimization specific bug fixes.

If you write out the integral, it's stock = { flow*dt + initial stock So the flow has to have the same units as the stock, with a time unit in the denominator. Then the flow's time cancels with dt (TIME STEP). Lookups can have units, though typical practice is to make things dimensionless, which sup...

I think we need to know a little more about what graphs you're displaying, and how, to answer this.

5.0a is indeed ancient.

The LP_SOLVE functionality is in an external function library (see http://vensim.com/workbench/ ) so it could potentially be adapted to 5.0, but I doubt it makes sense to do that.

First of all, I applaud your efforts to replicate a model from the lit. I do this quite a bit myself, and find that I learn a lot more about a model by hand coding it. If you look closely at the error message, it indicates that "change in cust.orient." is undefined - you have two instances...

Is this what you're after? http://vensim.com/lookups-in-vensim/

Also, check Chapter 8 of the User Guide. http://www.vensim.com/documentation/usr08.htm

There are some drawbacks to using real data - it can be a lot of work, and it can add features to your model that distract from a simple understanding of its behavior. However, the benefits often outweigh the drawbacks. Using real data may help you to discover problems in the model or the data, and ...

As the admin said, there's no real rule. Payoff sensitivity is useful for testing whether you've found the max/min, and how broad it is, but not for determining the iterations to get there. Whether you care to continue optimizing depends on your purpose. If you're doing policy optimization for a fir...

Two other notes: In DSS, rather than creating a constant and hiding it, you could create a :MACRO: to contain the constant. In this particular case, feps isn't really needed - you could use the ZIDZ function in place of the (x+feps) construction. If you want some behavior other than ZIDZ, you could ...

I don't have it in front of me (traveling) but I think you're correct - my mistake.

A time step lag in state updates is a pretty fundamental property of ODE systems, not to mention real physical systems. What are you trying to model that requires cascading instantaneous updates?

Not 100% sure I understand the question, but I think it's not practical to manipulate queue contents at the item or even time step slat level. Also, any input to an ALLOC has to be an array - the function can't look inside a queue, unless you first somehow select the queue elements into an array. Ev...

More options: Unfortunately, with the runup to the SD conference, I'm super busy so it's tough to keep up, but I'll try to respond to comments.

I'm not familiar with WITNESS, but if it's a discrete event simulator it might be more suitable for some problems. Generally, the dividing line is between tactical/operational control problems, where you really care about distributions and individual elements, and more strategic problems that blend ...