search

ivoa / dm-usecases

The is repo gathers all the material to be used in the DM workshop 2020
The Unlicense
1 stars 3 forks source link

TimeSeries example by Mireille Louys and François Bonnarel #29

Closed Bonnarel closed 3 years ago

Bonnarel commented 3 years ago

This example is based on a small model based on Mango Annotations is following ModelInstanceIntoVOT

mcdittmar commented 3 years ago

Is there a reason you are interested in using Mango to represent TimeSeries? The Cube model has long been expected to be the basis for TimeSeries, and while they have similar structure (which I like), there are important differences which make Mango appropriate for Catalog data, while Cube is better for Cube/TS/Spectra data products.

gilleslandais commented 3 years ago

I think also that Mango is a good place to serialize time series - If you have a time measure in output, you could choose the second axis among the measures listed in the Mango header (for instance a magnitude). We could so benefit the Mango semantic to specify the type/unit but also add other "rich" metadata like the filters description - metadata already serializable with mango - I could make a test in VizieR prototype -

lmichel commented 3 years ago

Mark, The goal of the multiband example is to show up how we can retrieve 3 light curves from a table where all photometric points are mixed up. For this, I need a TS model supporting multiple light curves. This is the purpose of mock_ts model.

mcdittmar commented 3 years ago
  • I could make a test in VizieR prototype - If you implement it in Vizier, I'd appreciate it if you implement both the Mango and Cube based TimeSeries so you could compare/contrast.
mcdittmar commented 3 years ago

Mark, The goal of the multiband example is to show up how we can retrieve 3 light curves from a table where all photometric points are mixed up. For this, I need a TS model supporting multiple light curves. This is the purpose of _mockts model.

I'll take a look at that example, I was thinking about that scenario too. I think Cube should be usable there, what incompatibility did you find?

lmichel commented 3 years ago

I'll take a look at that example, I was thinking about that scenario too. I think Cube should be usable there, what incompatibility did you find

Yes, Cube is definitely usable here. I just need a bit of time to implement and document it.

Bonnarel commented 3 years ago

Hi Mark, The main idea for us was to show some capabilities of Mango. And to test ModelInstanceIntoVOT up to work with the demonstrators. TimeSeries is a use case we know rather well due to previous work you know. That's why we built those examples rapidly. The toy model we built is not working for use cases more complex where the varying quantities are not a set of scalars for different axes. It will be easy to change that to "cube" later Cheers François Le 25/03/2021 à 23:43, Mark Cresitello-Dittmar a écrit :

Is there a reason you are interested in using Mango to represent TimeSeries? The Cube model has long been expected to be the basis for TimeSeries, and while they have similar structure (which I like), there are important differences which make Mango appropriate for Catalog data, while Cube is better for Cube/TS/Spectra data products.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/ivoa/dm-usecases/pull/29#issuecomment-807647835, or unsubscribe https://github.com/notifications/unsubscribe-auth/AMP5LTADMGFHOCPSTOS37PDTFO37VANCNFSM4Z2FA4DA.

loumir commented 3 years ago

Hi all, I am also interested to check the usage with CubeDM and try to sketch a template of classes , identified in the Cube DM, that represent light curves and TimeSeries of spectra for instance . I think to show how such models parts can be used is helpful to make the logical relations between metadata better understandable.