Ventana software support forum

It's probably not what you're looking for, but data sources for variables in calibrations are listed in the payoff report (runname.rep).

Here's your problem: when i use direction=0, the result is wrong (2,0,1;2,0,1). It is cleat that the correct answer should be 1,2,0;1,2,0 This is true if 1,2,0 represents the positions of each element in the sorted array. So, if the input is 3,2,4, the 0 in the third element would indicate that the ...

When I use 0 in the version that I posted, I get ... which looks OK. Am I missing something?

You have: price min to max[HOUR,TECH]=vector sort order(price x tech[HOUR,TECH],1) price order max to min[HOUR,TECH]=2-price min to max[HOUR,TECH] In the second operation, you're attempting to do pointer arithmetic on the sort order. But this is incorrect. The fact that an element i is in first posi...

The other possibility is to just put minus signs in all your integrations, and run the whole model backwards.

There is a trick you could do though - cumbersome in general, but feasible in a simple model. Create an auxiliary that calculates the value of the stocks ahead a unit of time (assuming Euler integration): future bad code = bad code + roc*TIME STEP Then use can use the "future" stocks to ca...

Vensim (and reality) is causal - that is, no variable can use information about the future.

Can you say more about the reason for this approach?

Also, you can use the table tool (or other analysis tools) to examine the inputs to the variable in question. Most likely you will find a division by zero, an exponential overflowing, or something like that. You can prevent division by zero with ZIDZ or XIDZ or an IF THEN ELSE statement, but general...

"RHS" is shorthand for "right hand side" of an equation.

What this probably means is that not all the inputs (causes) in your rate equation are defined, so that Vensim can't evaluate the units for the left hand side.

Another constraint is that members of subranges need to appear in a master range somewhere. For example: spread : hummus, peanut butter, jam sweets : honey, jam, molasses requires something like food : hummus, peanut butter, honey, jam, molasses However, for energy, it might make sense to keep the p...

You'll find a button in the bottom right corner of the subscript control panel that toggles between "simple" and "full" view of dimensions. That may do the trick. For example, if you have, cheese : cheddar, velveeta, brie, limburger, gruyere and add gross cheese : velveeta, limbu...

In terms of installation, they should coexist just fine. What problems do you see?

I think the context is compressible fluid flow in 3D space. You can't take a limit in the analytic sense. Generally, the way to handle a partial differential equation problem like this is to discretize space using array dimensions, just as we approximate continuous time with TIME STEP. There's an il...

For an aquifer, it would be more typical to model it continuously, e.g.

water in aquifer = INTEG( recharge rate - withdrawal rate, initial water in aquifer )

Another option (also not optimal) is to export the graph to a drawing program or powerpoint, ungroup the elements, and edit the line style there.

Typically you want to avoid this kind of discrete time formulation though. Much better to free yourself of the limits of Excel and rethink your model in continuous terms.

Depending on your problem space, there's a huge amount that can be accomplished with PLE/+, but logistics is the kind of area where you fairly quickly want arrays to handle detail.

Correct. Sensitivity analysis is expecting a list of constant values, not a time series. Glad it works!

This sounds like a 32/64 bit issue. On 32-bit Win systems, there is only Program Files. On 64 systems, 64 bit apps go in Program Files, and 32 bit apps are in Program Files (x86). If you remove the (x86) from the path to the vcvars32 call in mdl.bat, that should take care of things.

There won't be any particular hardware limitations, though more memory will help with huge models, particularly because double precision will be the default. All you'll really need is a 64bit operating system.

Bob Eberlein's objection to nonnegativity constraints in ithink, which I share, was always that they created invisible feedback structure. I think a good solution would be to permit the convenience of the constraint, but to make it visually explicit. In the case of the delay structure, the problem i...

vcvars32.bat is fairly incomprehensible to me, but it looks like it makes a registry query (starting from the call at line 3) to determine the location of common tools, and this is failing. I guess this must be some kind of installation error with Visual Studio, but since it works, I wouldn't worry ...

Summing inflows/outflows to get the net total is the way to go. If arrow spaghetti is an issue, you can hide the arrows, or use shadow variables or a separate sketch to make the summation away from the main stock/flow structure.

I'll look into the precision issue here.

A good test might be to open a command window and see whether you can run vcvars32.bat directly. There's no real benefit to using a 64 bit compiler at the moment, though there will be when Vensim 6.4 (64 bit beta) is released next month. The relevant difference at the moment is whether vcvars32 is l...