Hi
I join zipped three simplistic models.
The first essai.mdl works correctly if one runs the reality checks.
The second essai1.mdl does not work, althouth it is nearly equivalent to the first one.
I understand more or less why. Vensim can interpret the input one equation after the other with essai and not with essai1. Is there some way to avoid writing all the equations 1,2,3 etc... .
Of course there is a solution with the essai2. But one is obliged to subscript the other variables which I want to avoid.
The idea behind this is to subscript some reality chek equations using different parameter values eventually generated randomly.
I do nof know if I am excessively suspicious, but when one runs a reality check, it makes the check only once, on the parameter values of the model, and making the check on a set of parameter values may increase the credibility of the checking. With the solution used with the essai2, one is obliged to add a subscript to most of the variables which makes the model a bit complex for less simple models.
Best regards.
JJ
[Edited on 18-8-2010 by LAUJJL]
problem with the use of reality checks
problem with the use of reality checks
- Attachments
-
- essai.zip
- (3.04 KiB) Downloaded 177 times
-
bob@vensim.com
- Senior Member
- Posts: 1107
- Joined: Wed Mar 12, 2003 2:46 pm
Hi JJ,
Test Input equations need to satisfy the same rules as regular equations. You have
z = z sub[sub]
for the input z test input. This has a subscript on the right that does not appear on the left. The error reporting is not as clear as it might be, but there is no way that equation can be computed.
Subscripting all variables is one approach to doing what you want to do. Another might be using the time dimension to randomize the inputs - this works in your example, but may not be useful for more dynamic equations.
Test Input equations need to satisfy the same rules as regular equations. You have
z = z sub[sub]
for the input z test input. This has a subscript on the right that does not appear on the left. The error reporting is not as clear as it might be, but there is no way that equation can be computed.
Subscripting all variables is one approach to doing what you want to do. Another might be using the time dimension to randomize the inputs - this works in your example, but may not be useful for more dynamic equations.
Probel with the use of reality checks
Hi Bob
Thank you for your answer.
I had already understood the impossibility for Vensim to work out the equation z = s sub|sub]. But as one can replace it with a list of independant equations, I hoped that there was a solution to avodi the tedious writing of multiple different equations. I will follow your advice and build with each version model a separate version model with an additionnal subscript dedicated to randomized parameter checking.
Best regards.
JJ
Thank you for your answer.
I had already understood the impossibility for Vensim to work out the equation z = s sub|sub]. But as one can replace it with a list of independant equations, I hoped that there was a solution to avodi the tedious writing of multiple different equations. I will follow your advice and build with each version model a separate version model with an additionnal subscript dedicated to randomized parameter checking.
Best regards.
JJ