Open ajshajib opened 1 year ago
General Ideas:
Develop standard formats for passing lens data around (I know some work has been done along these lines by some folks in TDCOSMO)
Develop tools for orchestrating lens analyses at scale.
Develop standard plotting routines for visualizing results at scale.
Some specific things I can work on:
Ideas I have heard about that others should chime in and clarify/correct/riff on are:
I believe Sydney @smericks and Tian @astroskylee want to spend some quality time together planning their respective and complementary catalog forecasting projects, presumably aiming for maximum efficiency and synergy :-)
@padma18-vb and I are thinking to pair-code an AGNLensPop
class in sim-pipeline
to inherit from (?) and then build off @nkhadka21 and @sibirrer 's splendid GGLensPop
exemplar code. Its initial manifestation would use OM10, and teach us the sim-pipeline
API. Once we have that in place we will be able to have an informed discussion (later!) with @oguri about helping integrate his code into sim-pipeline
. We're assuming Masamune is unavailable this week but are looking forward to catching up with him properly in late August/September. Padma and Narayan had some good conversations this week - a good way for her to get started on Friday might be to spend time chatting and planning with @sibirrer , and then maybe I can be useful joining in the coding later (barking git commands over her shoulder etc)
@PatrickRWells I vote we focus on your 1st and 3rd bullets (collaborate on embedding Round Trip lenses into realistic sightlines and then look at using DC2/DP0.2 to do that). We may or may not hear back from Ji Won about re-using her DC2 sightlines by tomorrow but we can plan on getting help from her on that at some point. I'd love to spend time planning with you - we could perhaps do that while @smericks and @astroskylee are doing their planning chat, and @padma18-vb is getting set up.
Phil H @P-1884 - what's your thinking about how to spend Sprint Day? Simply joining in with any of the above is of course an easy and ever-present option!
Sounds great @drphilmarshall ! Let's make a first attempt on an AGN Source class and propagate it through a lens population class. Perhaps I can already improve a bit the interface in the current sim-pipeline to better enable it (or at least make some markings where to add it)
@drphilmarshall My first task will be project planning with Tian and Sydney (either today or at the sprint tomorrow). I was then hoping to get an end-to-end simple version of paltas going (had some trouble first time round) making use of the available help. I may listen into Padma and Simons debrief on sim-pipeline as that will be very relevant to know too for Project 2!
One small item for me is that I would like to consolidate some redundant-looking Slack channels. So, it would be good to have a quick chat with @drphilmarshall (and any others interested) about the origin stories and purposes of the SL channels, and how to consolidate those best.
For the AGN hack session, I have made a new branch in the sim-pipeline adding a skeleton of the quasar class and better more modular inter-compatibility: https://github.com/LSST-strong-lensing/sim-pipeline/tree/agn_sources, @padma18-vb
@PatrickRWells and I have tracked down @jiwoncpark 's code at https://github.com/jiwoncpark/node-to-joy/tree/master and will spend time inspecting and discussing it tomorrow, along with her paper at https://arxiv.org/abs/2211.07807 . We're thinking we'll need to make new sightlines for each sim-pipeline
lens (rather than re-using Ji Won's data).
Am I right in thinking that ALL of our planned sprint projects are pipeline / Round Trip related? If so we could use #sl-pipeline as the Slack channel for chatter, if @Nikki1510 doesn't object to the temporary burst of noise :-) We could all agree to be diligent about confining discussion to threads, to help with that.
@PatrickRWells and I have tracked down @jiwoncpark 's code at https://github.com/jiwoncpark/node-to-joy/tree/master and will spend time inspecting and discussing it tomorrow, along with her paper at https://arxiv.org/abs/2211.07807 . We're thinking we'll need to make new sightlines for each
sim-pipeline
lens (rather than re-using Ji Won's data).
An undergraduate student of mine is working on a structure-enhanced ray-tracing based on analytical code instead of DC2, using GLASS for the large scale structure, and then the halo mass function for small scales (yet not fully correlated) to at least predict the distribution of shear and convergences at small scales. See PR here: https://github.com/sibirrer/sim-pipeline/pull/7 by trivialTZ @trivialTZ, this could complement a more detailed implementation using N-body simulation products
Here's an attempt to schedule our time today: feel free to jump in and modify as needed. I'm mostly guessing based on the posts above but you might have better ideas for how we should spend our time. This is likely most useful for @martin-millon @ajshajib and me for figuring out how best to support the various projects, but should be useful for everyone to see what will be going on. Feel free to add more projects in more columns! I'll paste this to #sl-pipeline as well.
Strong Lensing Topical Team
Sprint to make progress on various fronts involving simulation pipeline, forecasting, analysis pipeline etc.
Contacts: Anowar Shajib, Nikki Arendse, Suhail Dhawan Time: All day Main communication channel: #desc-sl GitHub repo: https://github.com/LSST-strong-lensing/sim-pipeline In-person/Virtual/Hybrid: Hybrid Zoom room (if applicable):
Goals and deliverables
Identify issues/barriers in the current projects and make progress using the collective expertise of the participants
Products from the sprints
Resources and skills needed
Enthusiasm in strong lensing, other than that, any technical skills are greatly welcome
Detailed description
The SLTT will meet during one of the unconference sessions to discuss ideas for the sprint