Re: [visad] using ErrorEstimate

Hi Tom,

I guess I'm missing the boat on how VisAD deals with errors. So how would I use something like ErrorEstimate(double[] value, double error, Unit u) to represent errors in my FlatField? I'm having trouble imagining what "values" and "error" would be. Do these "values" have anything to do with my "values" in the range of my FlatField? Do you have a use case that clarifies its use?

It seems that you can make a Real with an error: Real(double value, double error). I assume this means value +/- error as in standard deviation, or is it variance? And if I added two Reals, A and B, the error would come out as sqrt(errA^2 + errB^2)?

If that's the case, it seems like I should be able to do something equivalent with FlatField-s where I have an array of values and an array of errors (e.g. sigmas) that correspond to those values.

Thanks,
Doug

Tom Rink wrote:
Hi Doug,

I don't think VisAD has a way of dealing with Errors that are
a function of the domain, maybe you can compute the mean/
variance directly from your data in this case.  You might be
able to construct a FieldImpl over the times with individual
FlatFields each one holding a separate Error, but this seams
kinda messy.

Tom

Doug Lindholm wrote:
Hi Tom,

So how do I get my errors into an ErrorEstimate? The constructors have args like: (double[] value, double error, Unit u).

If I have:
double[] times;
double[] values;
double[] errors;

with a FlatField that has times -> values
how would I construct the ErrorEstimate to pass to FlatField's setRangeErrors?

Thanks,
Doug

Tom Rink wrote:
Hi Doug,

Doug Lindholm wrote:
Hi,

I'm trying to figure out how to use ErrorEstimate-s. I have simple time
series data as FlatField-s (time -> value) to do some math on and I'd like to propagate some errors. The reality is that I have an array of values and an array (of the same size) of variances.

Do I need to construct an ErrorEstimate[] to pass to setRangeErrors? If so, how? I don't see a constructor that makes sense. I simply want to say, "here are the variances for each of my values."
Yes, though there's an ErrorEstimate for each RealType in the
FlatField range, the number of estimates must equal the FlatField's
TupleDimension. The ErrorEstimate class defines the propagation
method for various operations. FlatField.getRangeErrors will return
the result. The ErrorEstimates can also be specified in 'setSamples'.

Tom
Thanks,
Doug
_______________________________________________
visad mailing list
visad@xxxxxxxxxxxxxxxx
For list information, to unsubscribe, visit: http://www.unidata.ucar.edu/mailing_lists/