[coords] Question - Time domain coordinates and frames

[Arnold Rots]

We keep rehashing the same arguments - this is something we hashed out
several years ago.
In the spatial frames the origin is included in the frame class.
Then it makes sense to include the time origin (for a time offset axis)
also in the time frame.

The difference is simply that for time we have, in addition, three ways to
specify a time stamp that don't need an origin.

  - Arnold

-------------------------------------------------------------------------------------------------------------
Arnold H. Rots                                          Chandra X-ray
Science Center
Smithsonian Astrophysical Observatory                   tel:  +1 617 496
7701
60 Garden Street, MS 67                                      fax:  +1 617
495 7356
Cambridge, MA 02138
arots at cfa.harvard.edu
USA
http://hea-www.harvard.edu/~arots/
--------------------------------------------------------------------------------------------------------------


On Fri, Apr 13, 2018 at 5:03 PM, CresitelloDittmar, Mark <
mdittmar at cfa.harvard.edu> wrote:

> All,
>
> So, (again depending on what constitutes a Frame change in the time
> domain).  If we change the model a bit (see attached image).  We can remove
> the recursive problem, accommodate the annotation of particular 'real'
> values in data, and allow for the variations on JD/MJD that Paul alluded
> to.  It is NOT as rigid in the ISOTime.  With this model, origin changes in
> Frame would be more on the Calendar level.. and I agree with Markus that we
> could make a statement that we are supporting Gregorian, other systems can
> follow.
>
> The change basically puts the Frame relation to a specific moment in time,
> that being represented by an absolute TimeInstant:(JD/MJD/ISO) or a
> relative TimeOffset.  The TimeOffset, provides the zero point as
> TimeInstant + value for time since then.
>
> I'm also including some example serializations to show the mapping.  These
> use the current vo-dml/mapping syntax.
> The "_alt" versions use this diagram, the others are from the current
> model.
>
> Mark
>
>
>
> On Fri, Apr 13, 2018 at 9:04 AM, CresitelloDittmar, Mark <
> mdittmar at cfa.harvard.edu> wrote:
>
>> One more point on this:
>>
>> On Thu, Apr 12, 2018 at 5:18 PM, Arnold Rots <arots at cfa.harvard.edu>
>> wrote:
>>
>>> No, the timeOrigin is only REQUIRED in a TimeFrame if it is referenced
>>> by a TimeOffset.
>>> So, *both the TimeOffset and the timeOrigin can (and should!) refer to
>>> the same TimeFrame*.
>>> (they should, because specifying the origin in a different TimeFrame is
>>> asking for trouble)
>>>
>>>
>> What I'm taking away from this is:
>>   + TimeOffset and TimeOrigin are in the same Frame..
>>   + the TimeOffset has/requires a timeOrigin
>>   + the TimeOrigin should ignore the timeOrigin
>>
>> From this, I would have to conclude that the timeOrigin is a property of
>> the TimeOffset, not the Frame.
>>
>>
>>> It does mean that TimeOffsets that use different timeOrigins need to
>>> reference different TimeFrames.
>>> That is not a problem, since most TimeOffset-based time series will use
>>> the same timeOrigin.
>>>
>>
>> The question is more if they are *considered *to be in the same frame.
>> If I have 2 TimeOffsets,  T1 and T2, both in "TDB", "BARYCENTER"... are
>> they in the same frame?  even if they have different offsets?
>>
>> Mark
>>
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.ivoa.net/pipermail/dm/attachments/20180413/c53a6c50/attachment.html>