DM: Simulation Data Model RFC and other comments Tuesday, 1130-1300, A

Preview:

Citation preview

DM:Simulation Data Model

RFC and other comments

Tuesday, 1130-1300, A

Overview

• Recap SimDM• RFC Comments

– AI:action item

• Implementations– Note (LePetit etal) – SimDB@VO-URP

• Further work– SimDB (this afternoon)– SimTAP (yesterday)

Particle catalogues FOF group catalogues

Density field grids

Subhalo merger tree forest

Synthetic galaxy catalogues

Mock galaxy catalogues(“light cones”)

Mock images

GADGET3

initial conditionsFOF group

finder

SUBFIND

L-GalaxiesMoMaF

SkyMaker

“Gridder”

Galaxy spectra

BC03/M05 SSPs

+

Other example: PDR

Which docs in RFC?

• see header of RFC page

• May need better cross-linking in docs themselves

• AI: Find out how to combine docs in final PR package.

Use of term Protocol• SimDM:simdb/protocol/Protocol:

describes how an experiment can/will/must be defined.– Compare ”Method by which an observation is made” ()

• Defines parameters, results,some semantics

• Compare to SIAP, SSA etc.– SAME!

• So in fact very apt name to use!

• AI: describe possible confusion in document and suggested ways to avoid it.

Service not well defined

• A resource that allows access to SimDM/Resource-s– to results of 1..* Experiments– to results of experiments for a

given SimDM/Protocol– to resources in a SimDM/Project

• Vaguely defined on purpose– service will define details itself

• Can be a SimDAL service– when defined for a finite set of SimDM/Protocols,

could have SimTAP as queryData part.• AI: improve description (CR, GL)

How to use the model?• SimDAL+SimTAP

– SimDB?– SimRegistryExt?

• Too complex for simple models?• SimDM targets discovery of services

– Also results of “simple models” have been created with protocols• Sometimes required to assist description of complex models

– examples above• AI: Try to describe simple models in SimDM• NB

Spec contains no statement about exclusivity of SimDM– special cases may deserve (if not need) special attention– “micro-simulations”, “cosmological simulations”

SimDB => SimDM• in utypes and names of XML files, replace

SimDB with SimDM• AI: Do it, how about root package (simdb)?

Algorithm• What if different protocols use same Algorithm.

No normalised usage of common algorithms• REPLY:

Algorithm uses semantic label to refer to externally assumed common definition. No redundancy.

• AI: none

New comments(discussion yesterday with VOParis)

• ProductType: Should it be on Protocol iso on Product– Query: “Give me all SimDM/Protocol-s that

can produce spectra.”

• Do we need a kind of InputDataType on Protocol?– Required information if one wants to run a

SimDM/Protocol

• AI: put discussion on wiki

Semantics

• We use SKOS concepts

• Semantic vocabularies for SimDM– See Franck in semantics session

• AI: Check we have properly written vocabularies containing all SKOS concepts used in the model.

Implementations

• What is an implementation?1. serialisation of real(istic) instaces in XML

schema

2. (Web) application using the model

1. Described in Note (link)– see B. Ooghe Thursday morning DM)

2. SimDB@VO-URP (local link)

Registry

• Which SimDM/Resource-s qualify as Registry resources.

• SimDM → Registry extension for SimDAL services.

• AI: Discuss what must be done, this afternoon theory session.

Recommended