QMikeAI

QMikeAI research library

Papers, limits
and open code.

The research behind the questions we ask: how reservoirs learn, what they can remember, and how their assumptions can be examined.

Empirical reservoir studies

Learning from a finite sample.

Daniel Mo Houshmand · 2025

Sample Efficiency Crisis in Quantum Reservoir Computing: Scaling Analysis on 156-Qubit IBM Hardware and Rigetti Simulation

An empirical study of the training data needed to turn quantum reservoir measurements into useful predictions.

Question
How does a larger quantum reservoir change the amount of data needed for a reliable readout?
Method
Reservoir features and regularised readouts are studied across hardware and simulated settings, with classical comparisons.
Evidence
A versioned preprint, research code and a later hardware-provenance correction are public. The Zenodo deposit predates that correction; use the pinned repository record for corrected hardware attribution.
Boundary
These results do not establish weather-forecast skill or a general quantum advantage. A registered preprint is not evidence of peer review.

Source records

The Zenodo deposit has a registered version. The separate SSRN record has no verified manuscript version; consult the corrected provenance for hardware attribution.

  1. Original deposit
    PreprintZenodo · v1.0.0DOI 10.5281/zenodo.17910992Registry metadata (JSON)
  2. Separate SSRN record
    Posted contentSSRN · version unverifiedDOI 10.2139/ssrn.5914465Registry metadata (JSON)

Theoretical limits

What the dynamics can remember.

Daniel Mo Houshmand · 2026

The Fundamental Tension in Topological Quantum Reservoir Computing: Why Unitarity Opposes the Echo State Property

A theoretical study of the tension between reversible quantum evolution and the fading of a reservoir’s initial state.

Question
Can a closed, unitary Fibonacci-anyon reservoir forget its initial state in the sense required by the echo state property?
Method
Mathematical analysis of distance preservation under unitary evolution, accompanied by numerical reservoir studies.
Evidence
The original no-go preprint and revised tension preprint share one Zenodo version lineage. They are versions of the same research record, not independent studies.
Boundary
The argument depends on the stated dynamics and echo-state assumptions. It is not a blanket prohibition on dissipative or measurement-assisted quantum reservoirs.

Source records

The Zenodo preprints form one version lineage, shown newest first. SSRN is a separate record with no verified manuscript version.

  1. Revised preprint
    PreprintZenodo · v3.0.0DOI 10.5281/zenodo.18371699Registry metadata (JSON)
  2. Original no-go preprint

    Fundamental Limitations of Topological Quantum Reservoir Computing: A No-Go Theorem for Fibonacci Anyonic Systems

    PreprintZenodo · v1.0.0DOI 10.5281/zenodo.17889778Registry metadata (JSON)
  3. Separate SSRN record
    Posted contentSSRN · version unverifiedDOI 10.2139/ssrn.5914523Registry metadata (JSON)

Research software

Making simulation assumptions inspectable.

Daniel Mo Houshmand · 2026

rigetti-fake-backends: FakeNovera and FakeCepheus

Independently developed simulation backends for exploring device-inspired noise assumptions.

Question
How can researchers inspect and reproduce the assumptions behind a hardware-inspired simulation?
Method
Public Python software exposes Novera- and Cepheus-inspired noise models for simulation workflows.
Evidence
The public repository provides code and citation metadata. This entry describes the public software record.
Boundary
These are independent simulation tools, not an official Rigetti provider or evidence of a hardware run. No verified publication DOI or peer-reviewed paper is claimed here.

Source records

The version below follows the public software citation record.

  1. Public software record
    SoftwareGitHub · v0.1.0