# Plans A and D — publication and independent review packet

Prepared 20 October 2026, release 0.5.0. This is a reviewable packet, not a submission
receipt. No new repository, DOI, ORCID, registration, endorsement, partner agreement
or independent review has been created by this update. No outreach has been sent.

## Identity and release

Creator: Christopher Hill, independent research. Confirm an existing ORCID or create
one through Christopher's authenticated account; never adopt a same-name identifier.
The current release is 0.5.0, not the proposed 1.0.0. The research ZIP excludes private
mail settings, website configuration and analytics. Check LICENSE-SCOPE.txt and the
PanTHERIA provenance before deposit. Its README, pinned dependencies, raw/derived data,
source code, citation metadata and output manifests make the proposed deposit concrete.

Publish the exact research release in the project-controlled repository/deposit account.
Record the actual URL and version DOI only after publication succeeds. Cite that version
DOI for results depending on these bytes; a concept DOI is appropriate for the evolving
project. A reserved DOI is not proof of a public deposit. Neither a DOI nor a timestamp
certifies scientific validity, guarantees legal protection or proves priority over all
earlier work. Keep existing website disclosures and license attribution accurate.

The PanTHERIA analysis and already-inspected memristor intake cannot retroactively become
preregistered. A future untouched-data study can be registered after pilot decisions are
final, with pilot use disclosed. Verify the selected registration service and submission
availability at the time of use; no generalist preprint route is assumed available.

## Venue routing

CODECHECK: request an independent rerun using this bundle, its output manifest and exact
commands. `codecheck.yml` is author-supplied metadata only; it contains no certificate,
reviewer, badge or invented record. A code rerun is distinct from checking scientific
validity. See https://codecheck.org.uk/guide/community-workflow-author and
https://codecheck.org.uk/spec/config/1.0/ .

ReScience C: develop a substantial independently written reproduction of a specific
published result, documenting successes and failures. The present scalar recurrence
check is not a completed paper replication. The venue excludes replication of your own
or close collaborators' research. An editor can reject a submission; acceptance depends
on review, not a predicted timetable. Disclose communications and let the editor assess
reviewer independence. See https://rescience.github.io/write/ .

arXiv: the January 21, 2026 policy no longer treats an institutional email alone as
sufficient for new authors. The first path combines that email with existing accepted
authorship in the relevant endorsement domain; otherwise seek an eligible personal
endorser. Existing endorsement is retained. Endorsement is not peer review or guaranteed
acceptance. See https://blog.arxiv.org/2026/01/21/attention-authors-updated-endorsement-policy/ .

Zenodo version guidance: https://zenodo.org/help/versioning .
No grant application, journal submission or promotional claim has been produced on an
assumption of eligibility. Read each program's current authorship/AI rules before drafting
or submitting. A calibration pilot can precede registration if its use is transparently
disclosed; waiting for an acceptance promise is not a scientific prerequisite.

## Replication target and limitations

`protocol_tools.py` implements exact scalar on/off exponential relaxation. For a=1-x,
alpha=exp(-tp/tau_on), beta=exp(-tr/tau_off), a1=alpha and
a_(n+1)=alpha*(1-beta)+alpha*beta*a_n. Its fixed point is
alpha*(1-beta)/(1-alpha*beta). An adaptive ODE integration agrees within 1e-9 in the
included boundary checks. These are normalized state calculations with illustrative
parameters, not measured current or a reproduction of a numbered published figure.

For Rivera-Sierra/Bisquert figure reproduction, freeze the target figure, units and
observation law, then implement the full series-resistance/capacitance/internal-voltage
system where required. Audit the source's stated time units against the plotted dynamics;
do not silently change them. The existing measured intake remains a sourced intake,
not a fitted validation of the oscillator. Shared datasets from the same group or device
are not automatically independent replications.

Scalar relaxation does not make the full oscillator equivalent to CALM. Exponential
fading is not rate-independent relay/return-point memory. A fixed point alone is not
learning: specify an objective, update rule and held-out task if claiming learning.

## Potential contacts, not established partners

Nathan Keim, Penn State: potential critique of material-memory protocol and rival
mechanisms. Official directory: https://science.psu.edu/physics/people/nxk913 .
PROFUND, led by Juan Bisquert: original-source clarification and a precise replication
target; author clarification is distinct from independent review. Group-linked record:
https://zenodo.org/communities/profund-bisquert/records .
These listings indicate relevant public research, not willingness to collaborate.
The independent reviewer should disclose conflicts; earned coauthorship should never
be suppressed merely to qualify for a particular venue.

## Draft review request — not sent

Subject: Request for a bounded methods review of NetPhysics release 0.5.0

Dear prospective reviewer,

I am Christopher Hill, an independent researcher developing NetPhysics. I would value
a critical review of one bounded part of the attached research packet: the mammal
scaling benchmark, the proposed material-memory controls, or a computational rerun.

The public-data benchmark includes 573 mammal species. A fixed 10 kg hinge does not
improve the primary family-held-out prediction; all splits and sensitivities are included.
The acoustic and elastomer work is still at protocol stage. I am not claiming a new
physical law, peer review, or your endorsement.

If this is within your interests, would you be willing to suggest a specific flaw to
address or an appropriate independent reviewer? The packet includes code, source data,
licenses and the exact output manifest. I would disclose any resulting collaboration
and agree the scope before further work.

Thank you for considering it,
Christopher Hill
NetPhysics

Before sending: select a specific recipient and one relevant question; confirm the
attachment/link and sender contact details. Do not send a bulk generic collaboration
claim. Log any actual response and permission before naming an institutional partner.
