ekalmX — Research¶

Background research, prior art audits, and theoretical foundations that informed the design.

Structure¶

research/
├── README.md                  # This file
├── research_enskf.md          # Audit of existing kalman_filter/ codebase (~5,000 lines)
├── ens_kalman_process.md      # EKP theory: math, numerical requirements, process zoo, BLR connection
└── operational_da.md          # Streaming / alert-service patterns; cold-start budget; commitments

Reading Order¶

research_enskf.md — inventory of existing code being migrated (9 discrete filters, 9 continuous-time filters, patch decomposition, parameter estimation). Includes gap analysis and mapping to ekalmX architecture.
ens_kalman_process.md — deep-dive on Ensemble Kalman Processes (EKI, EKS, UKI, ETKI, GNKI). Covers mathematical formulations, numerical requirements (ensemble size, scheduling, constraints), process zoo comparison tables, JAX API design, example applications, connection to the Bayesian Learning Rule, and gaussx integration.
operational_da.md — operational tail of the research-to-production arc: alert services, reanalysis, online retraining. Cold-start budget, streaming patterns, what filterax commits to vs. punts to deployment.

How These Relate to the Design Docs¶

Research file	Informs	Key content
`research_enskf.md`	`api/components.md` (filters, smoothers), `boundaries.md` (gap analysis)	What exists today, what needs formalization
`ens_kalman_process.md`	`api/components.md` (processes, schedulers), `architecture.md` (gaussx integration), `boundaries.md` (BLR decision tree)	Math behind each process, when to use which
`operational_da.md`	`features/state_persistence.md`, `integrations/plumax.md`, `features/multi_instrument.md`	Why the new ADRs D11–D16 exist; latency / streaming concerns