[NeXus-tech] axes for multidimensional data

[Pete Jemian]

Here's an example plot of a data set.
It was collected with the EPICS sscan record using a two-layer next of 
1-D scans.  The outer scan was the delay generator, the inner scan was a 
positioner.  Note the drift in the top rows of the positioner scan.
I'll see if I can translate this into an HDF5 file as an example data 
set for us.

On 2/22/2011 4:58 AM, Tobias Richter wrote:
> Hi Mark,
>
> On Tue, 2011-02-22 at 08:46 +0100, Mark Koennecke wrote:
>> I am a little confused: up to now we had discussed this problem in the
>> context of a scan in 2D or more. Not as
>> a general thing.
>
> As I wrote the @drive idea is an aside that might be useful, but I'd
> want it to be left out for now. Otherwise the problem is general only in
> that it applies to scans in nD with n>1.
>
>> Then in the context of scans in 2D the problem has come up at PSI too. I
>> will add another two ideas towards a solution
>> to the mess. The first is to use an additional data attribute which
>> describes what has been meant:
>>
>> data:NXdata
>>       data[nx*ny]
>>           @signal=1
>>           @scan_intent=nx:ny
>>        x[nx*ny]
>>           @axis=1
>>           @primary=1
>>       y[nx*ny]
>>           @axis=1
>
> This (and Ben's suggestion) suffers from not passing the slicing test.
> For our users the main selling point of hdf5 is efficient storage of
> multi dimensional data. Here the view is almost the more dimensions the
> better - leave all the off-by-one errors in dealing with the arrays to
> the well tested file backend. This is important for user written
> analysis scripts.
>
> But even if we will not find an agreement here, as long as it is legal
> to write data the way we prefer here, there should be a solution that
> solves our axis problem as well.
>
> Regards,
> Tobias
>