Problems with expressions in Monte Carlo

I split this into a separate thread, because it makes no sense appending to a thread which was over a year old which just happens to have the word "monte carlo" in it.

I suspect your expressions are referring to result names which are not the "aliased" names. There's good coverage of this in the documentation. In order for me to point you at it, and ensure this is the problem, can you:

a) let me know what subversion of IC you're using (Help->About in the CIW)
b) post the expressions you have which cause the trouble

Regards,

Andrew.

The subversion is IC6.1.4.485

The expression is:
ocnxlOutputExpr( "fallTime(clip(v(\"/BL'\" ?result \"tran-tran\") 3.65e-07 3.95e-07) ymax(clip(v(\"/BL'\" ?result \"tran-tran\") 3.65e-07 3.95e-07)) nil 0.6 nil 10 90 nil \"time\")" ?name "precharge" ?plot t ?save t)

I believe it is using the same plot and not subsequent ones, because when I change the nominal parameters I get a different result but no variation.

Thanks for your help,

Karo

Change both occurrences of ?result \"tran-tran\" to ?result 'tran or ?result \"tran\" .

Search in cdnshelp for "Data Name Aliasing" (put quotes around this phrase in the search field), and  you'll find a section in the Monte Carlo chapter (chapter 7) of the ADE XL documentation (<ICinstDir>/doc/adexl/adexl.pdf if you want to look at the PDF directly).

Regards,

Andrew.

Thanks a lot for your help. That seems to work. I was wondering why does this make a difference?

I haven't exactly figured out Data Name Aliasing. I'll let you know if that also works.

Thanks,
Karo

Karo,

The reason is that when you run Monte Carlo, the results go into a different database name with a name specific to the fact that it's within a monte carlo analysis, and which iteration it is on. The "aliased" name always points to the "current" point in this sweep. A name such as "tran-tran" is actually referring to a specific database, which in this case is the nominal run. To avoid you having to know which specific database to calculate with (which changes at each monte carlo point), the idea is that expressions should refer to the aliased result name (e.g. 'tran, 'ac, etc). Being able to refer to "tran-tran" allows you to have an expression which (say) finds the difference between the current iteration and the nominal.

This is what the data name aliasing explains - essentially the solution I've given is using the aliased name.

Regards,

Andrew.

Karo