Skip to content

[HACK] IA Modeling code comparison #48

Description

@aniruddh-herle

Summary

Given the same data vector as input, different modeling codes should produce the same parameters as output. It would be a very useful consistency check if every available IA modeling code is used to fit the same measurement, and the posteriors are consistent with each other.

Leads & Contacts

Proposed Timing & Format

  • Preferred day(s): No preference
  • Duration: Maybe 90 mins, might need more time
  • Modality: hybrid
  • Session type: working session

Team / Participants (initial list)

  • open to anyone interested in validating their code

Goals & Deliverables

  • A plot of model vs data for wgg and wg+ with residuals with models from all participating codes
  • A plot of parameter posteriors for all available model combinations, linear/ non-linear bias + NLA/ TATT

Approach / Tasks

  • I share my pre-computed measurements of wgg and wg+ with participants
  • Everyone attempts to model the data with their own codes
  • We compare outputs

Data / Dependencies

  • I will share wgg and wg+ calculated using IACorr on z=0 snapshot of the FLAMINGO 2.8 Gpc run

Resources Needed

  • Room to work together

Outputs & Sharing

  • Will make folder in the echoIA github repo
  • Final slides with the deliverables marked out above

Related Items

  • None

Checklist

  • [✔] Lead(s) can attend at proposed time(s)
  • [✔] License/permissions for data/software are clear
  • [✔] Agree to the Code of Conduct

Metadata

Metadata

Assignees

Labels

No labels
No labels

Type

No type
No fields configured for issues without a type.

Projects

No projects

Milestone

No milestone

Relationships

None yet

Development

No branches or pull requests

Issue actions