feat(psychometric): restore Driver p.16 TIPREDVARstd v/v=1 on main

[seonghobae]

Adapter-lineage status: blocked on canonical owner

This Draft supersedes dirty predecessor #272 for the TEPP-specific TIPREDVARstd naming/admission lineage, but it is not a valid production landing vehicle in its current form because it still implements reusable scalar covariance-standardisation arithmetic inside legacy psychometric_core.

Canonical owner work has converged upstream. The narrower concurrent fast-mlsirm #1723 is closed unmerged after its unique scalar exact-unit evidence was preserved in the broader surviving fast-mlsirm PR #1722, current candidate contract fast_mlsirm.covariance_standardization@1.0.0. #1722 covers scalar and matrix covariance-to-correlation standardisation and currently remains an open, unreleased upstream candidate; its PR head is not a TEPP production dependency.

Required TEPP conversion before landing

After the surviving owner contract is merged and released/versioned:

1. replace the local scalar-standardisation arithmetic with a version-pinned fast-mlsirm ACL/adapter;
2. retain only TEPP-owned model/time policy: the TIPREDVARstd name, event-time structural admission and ctsem-specific refusal/traceability semantics;
3. demonstrate parity against the preserved feat(psychometric): restore Driver p.16 TIPREDVARstd v/v=1 on main #272/feat(psychometric): restore Driver p.16 TIPREDVARstd v/v=1 on main #315 cases, including exact binary64 1.0 at v = 3;
4. remove the duplicate local production kernel and wrong-owner error variants;
5. fold the resulting adapter into the Longitudinal Modeling/Measurement bounded-context landing vehicle rather than reviving a standalone one-map product boundary;
6. reacquire exact-head Rust, coverage, documentation, security and qualifying review evidence.

Until those conditions are met, keep Draft and do not merge this branch even if hosted checks happen to pass. If #1722 is superseded before release, re-read the surviving canonical-owner contract rather than pinning this PR description to a remembered PR number.

Scientific lineage

Driver, C. C., Oud, J. H. L., & Voelkle, M. C. (2017). Continuous time structural equation modeling with R package ctsem. Journal of Statistical Software, 77(5), 1–35. https://doi.org/10.18637/jss.v077.i05

The named ctsem reporting semantics remain TEPP evidence; the reusable static numerical identity does not.

[coderabbitai]

Important

Draft PR not reviewed

Draft PRs are not automatically reviewed by default.

• Trigger a manual review

To automatically review draft PRs, update your CodeRabbit configuration:

reviews:
  auto_review:
    drafts: true

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}