I have a *very long* MC run which is taking far more than expected. It simulated some 1000 points already and I would like to examine the results. Unfortunately (for a reason that escapes me) the most important parameter is an expression working for normal sweeps but not for MC. Therefore I cannot see what I would like right now.
I will probably be able to 'trick' the simulator after the sim has finished. And now is the question:
--> If I stop the simulation, will ADEXL create a 'top level' PSF from which to extract the istogram of the variable (expression) I need?
I cannot see this PSF database in the root of the simulation right now.
I would like to avoid wasting so many days of simulation so if anybody knows either:
- the answer to the above question, or
- a way of accessing ADEXL results structure in order to create histograms
I would really be grateful :-)
I've just tried this in IC18.104.22.1680.16.2, and it worked fine; I was able to plot the histograms on my incomplete monte carlo run. My expectation was that this would work - I just wanted to confirm it before I replied.
Of course, don't blame me if it doesn't work for you ;-)
thanks for your super-prompt reply :)
However, I was not clear enough: in my outputs setup, there are signals and expression which are 'working'. For these, I can see the mean/sigma current values and even plot the partial results (histogram) if I want.
However, next to those, I also have expressions which do not evaluate correctly in the output pane. ADEXL is giving some error when simulating MC, however the exact same expression work seamlessly if you run a 'simple' single point simulation, and after MC sim has finished you can work your way around this issue by re-building the expression from its components. At this point, you can have your mean/sigma and histogram for those expressions which do not readily evaluate while simulating.
Right now, we are 'in the middle' of a MC sim. At this point:
- there is a PSF database at top level, but it is empty!
- there are PSF databases in each sim folder (each MC run), but although there are data files (pss.td and so on..), those PSF are not recognized as databases from the Results Browser, so we cannot open them to display results.
I am using IC6.1.5-64b-500.11, btw.
...On a different note, maybe you can shed some light on the type of error we are getting, or its possible root cause.
Here attached is:
A quick answer because I'm tied up with meetings today so can't really focus on this (so it might be best to go to customer support).
Andrew Beckett said:Hi Michele,A quick answer because I'm tied up with meetings today so can't really focus on this (so it might be best to go to customer support).The PSF database at the top level should be empty, because it is populated based on what ran - it's actually a set of references to the lower level run results.I would expect you to be able to open a single point run in the results browser, assuming it has finished (doing RMB->Results Browser over the individual run output).Whether you can access the individual run results will depend on whether you have saved family data or not (one of the monte carlo options are).
indeed you are right. We could not access the curves because we (on purpose) decided not to save data families. It gets very heavy very quickly with MC.
Andrew Beckett said:I suspect your problem is caused by the fact that you have dependent expressions using the "ADE L" way of having just the output name referenced rather than using calcVal("outputName"). There are some issues with doing that in ADE XL. If you used calcVal(), I would expect it to work...
...Yeps...We lost that simulation, anyways. Once that specific function is returning an error, of course there is no data saved.
And you are right again :) : I developed the functions using ADE L/ Calculator. I changed any reference to a previously defined function with CalcVal(function, testbench) and solved the issue.
By the way, are these differences documented somewhere easily accessible?
MicheleA said:By the way, are these differences documented somewhere easily accessible?
Unfortunately it should really be in the Known Problems and Solutions - it ought to work without the calcVal, but doesn't. I recall a colleague hitting a similar problem a while back, but they don't seem to have written it up as a solution or it ended up in the KP&S either. Sorry...