In trying to close LIT-101, which was about a reference to "limited classification" in the Level 2 object catalog, Zeljko Ivezic and I rediscovered a gap in DM's requirements flowdown. We discussed this by teleconference today.

The SRD (LPM-17) contains a "will" statement about DM providing "pre-defined filters optimized for traditionally popular transients, such as supernovae and micro lensed sources".

This was flowed down nearly verbatim to the LSR (LSE-29) as LSR-REQ-0026, "Predefined Transient Filters":

Requirement: Pre-defined filters optimized for traditionally popular transients shall be made available. It shall be possible for the project to add new pre-defined filters as the survey progresses.
Discussion: The list of pre-defined filters, by way of example, should include ones for supernovae and microlensed sources.

This requirement was never flowed down to the OSS (LSE-30), the DMSR (LSE-61), or the DPDD (LSE-163), in the space of system-level controlled documents.

The requirement should either be formally disclaimed, which would require a variance against the SRD and a change request against the LSR, or the proper flowdown should be performed.

The latter would be in two parts: a CCB-level change request for the OSS, DMSR, and DPDD, as well as, within DM, the addition of substantive language to LDM-151 towards fulfilling this requirement.

As the OSS and DMSR are currently completely silent on this, it would be acceptable simply to flow down the LSR requirement verbatim to both of these (as if a <copy> relationship in SysML terms). However, as noted in LIT-101, the DPDD currently contains language which appears to be in conflict with LSR-REQ-0026, specifically:

we do not plan to provide any classification (eg., “is the light curve consistent with an RR Lyra?”, or “a Type Ia SN?”).

This would have to be edited to clarify that while we will not attempt to produce exclusive classifications - that is, assignment of objects to unique categories, but we will provide pre-defined filters, with potentially highly overlapping selections, that provide good completeness but perhaps only very modest purity for a small number of object types of common interest.

It is important to retain the notion that this would be done using only LSST data.

Strictly speaking, as a "pre-defined filter" will not be thought of by most of our readers as a "data product", it might not need to be mentioned in the DPDD, but because the existing DPDD language suggests a strong conflict, it would be very good to clarify it.

It will also be important to harmonize what we do about this requirement with what we say about the "mini-broker" in our requirements flowdown, as this is also not currently very clear.