I think I confused the issue with so many details. The main issue is, if P1 has
invalid data and P2 has different invalid data, what happens to the validity of
the resulting field when P1 positions are replaced by P2 data component. And,
how does the Plot module treat this new field in terms of validity. Does any
manipulation of the invalid positons array need to be done so that neither P1
invalids or P2 invalids are plotted ? I would think so, and the results that I
see after the tinkering make sense to me. It's just the results without the
tinkering that I don't understand.
On Apr 26, 4:04pm, Chris Pelkie wrote:
> Subject: Re: [opendx-users] validity of points in x/y plots
> More thoughts.
>
> >Case #2:
> >So, I tried adding another step. I extract the invalid positions component
of
> >P1 and P2 and do a logical .or. to get a new invalid positions array
(assuming
> >"def positions"). I then replace the invalid positions component of the new
> >field object with this array. The plotted field object has a greater number
of
> >invalid points than either P1 or P2, so I initially assume that things are
> >working OK.
>
> I'm assuming you mean "dep" above (since the alternative is "ref").
Yes, I meant "dep" positions. Sorry typo
> >I compared plots using both methods, and the results are confusing. The
plots
> >look the same (same number of plotted points - It appears the fewer number)
> >regardless of use of the "combined" invalid positions array. Also, for
> >case #1,
> >some data that should be tagged as "invalid" still remains in the new field
> >(Describe tells me that valid positions range from some reasonable number to
> >e+38, obviously the out-of-range value). This is even though the high range
> >does not show up in the plot (x and y axis each range < 40)
> >
>
> Sorry sent the last note without rereading your report carefully.
> Invalidating does not change positions data, it just masks undesirable
> positions. How did you get e+38 in the first place? Is that the data value
> you are trying to invalidate?
Yes, I am trying to invalidate e+38. I see now the possible holes in my
description of this. Right, invalidating doesn't change position data - I was
referring to the validity of a field being characterized by Describe in terms
of the bounding box of valid points ("valid positions").
What I'm getting at is that if no tinkering of invalid positions array is done,
I don't get the description of validity that I'm looking for (locations of P1
invalids and P2 invalids both tagged "invalid"). Some of the original
invalidity (P1 or P2) appears to be lost when the replacement of positions is
done, since the "valid data" now ranges to e+38. Apparently, some instances of
e+38 have lost their invalidity tag. This might be expected following a Replace
of positions component, but you would think then that these e+38 values would
show up in the plot. But I don't see them in the plot.
> >Is Plot smart enough to throw out all the appropriate invalid points (in x
and
> >y) even though Describe appears to indicate that they remain in the field ?
Am
> >I doing this all wrong ?
>
> Invalids are only actually culled with Include with cull set to true. Most
> modules leave the invalid positions in the field but ignore them in some
> way appropriate to the module. Statistics skips them, Image doesn't render
> them, etc.
>
Right - I probably shouldn't have said "throw out". The wierd thing here is
that the instances of e+38 that Describe says are now "valid" seem to be
ignored (correct term) by Plot. Is Plot smarter than Describe here ?
> Chris Pelkie
> Vice President/Scientific Visualization Producer
> Conceptual Reality Presentations, Inc.
> 30 West Meadow Drive
> Ithaca, NY 14850
> [EMAIL PROTECTED]
>
>-- End of excerpt from Chris Pelkie
