Ventana software support forum

But what is the purpose of these operations?

Please do - I'm curious to see how this works out.

OK ... here's the issue. At present you have a lookup containing the data, used to generate real_report. The model samples the overwintering population on day 30 to generate "report". The payoff compares the two and computes a sum of squares. So far so good. Problem #1: the data in real_report isn't...

You can definitely anneal prior to generating the sample - you just want to be sure that the temperature is down to 1 before the end of the burnin period.

I don't see your attachments.

I think the first question is probably whether your payoff is properly weighted. With Normal distributions, it should have an expected value roughly equal to the number of points.

Can you share your setup - if not the model and data, at least the .vpd and .voc files and a payoff report (.rep)?

Right. That could make the syntax of the GRAPH function: GRAPH( inputVar, xmin, xmax, [vector of points] ). I think the Min/Max/Zoom items may be just for display. The big question is probably whether the 3rd argument should be xincrement or xmax. I agree that embedding constants in equations is bad...

Is this basically a lookup table?

That might be one of the cases where you actually do need to exit & relaunch to get it to work. If not, it's a bug.

If you have some time to experiment, I'd be interested in some timing comparisons between your Ryzen machine and my Intel-based workstation.

It's probably a parentheses nesting problem. It's better if you make the sin wave a separate variable, then clip it in another equation. Also, the parameters should be made into variables so that units balance. For example: sinwave = amplitude*SIN( (Time-PhaseShift)/period*2*3.1415926 ) clippedSinwa...

Can the existing setup be preserved via addition of _VDFX to the compiler preprocessor definition list, or are there other things that need to change?

One option would be to do a diff on the new shipping venext.c example vs. your venext.c and see what differs.

You might look up some of the other models that address testing:
https://metasd.com/2020/03/model-covid-19-us/
https://forio.com/app/jeroen_struben/co ... ction.html
https://www.medrxiv.org/content/10.1101 ... 1.full.pdf

Right - you need to modify your .dll project to use the new copy of vensim.h etc. (typically in users\public\comp).

This means you have an external function library loaded that needs to be recompiled with the new version's headers.

If you're not using it, you can delete the reference at Tools>Options>Startup.

Will look when I can. Your description sounds right though. Just watch out for missing minus signs - easy mistake to make.

Sorry - requires some thought and I've been ultra busy. 1. This is not automated yet, so you have to create an equation to express your prior. Here's an example. Suppose I have a policy effect parameter. If you don't know what its value could be, just that it's >0, you could use the improper log-uni...

Another trick that I used ages ago, before GET VDF existed, was to create a separate model containing just a pair of variables that transform the base value into a data variable with a different name. Something like: Main model: price of cheese = {some simulated stuff} data price of cheese {data var...

Great! It's been a while since I've done this, but there's always a chicken-egg problem when the Base run that you're looking up comparison values from doesn't exist yet. I think I've solved this with some kind of IF THEN ELSE( base data = :NA:, 0, base data) logic, but I don't remember the details....

I think you want GET VDF DATA.

I agree - this worked until very recently. We'll have a fix soon.

If you think about a run with 4 time steps, the random draws might be Time 1 2 3 4 Value 3 8 2 6 If you halve the time step, you get Time 0.5 1 1.5 2 2.5 3 3.5 4 Value 3 8 2 6 1 7 7 9 The value 2 now occurs at time 1.5 instead of time 2, etc. In a real model, it's more complex than that, because you...

Time step will have a big effect on any randomness - if you draw twice as many values, you get a completely different realization of the random streams. To calibrate a model with random inputs, you really need to look at the performance over an ensemble of simulations, rather than a single run. This...

The best thing to do is create some custom graphs that compare the model to the data, and then do some hand calibration to see what's going on.

If shrinking the time step is changing the payoff, that's probably a sign that it's also changing the dynamics, and that in turn is changing the fit.

The choice of time step really shouldn't have anything to do with the data interval.

I think this might be a bug in the payoff reporting that we've fixed in development already. If so, the .out file written should still contain your final parameters, so you can rerun the endpoint solution.