Qirui Hu

Shanghai University of Finance and Economics

Unclaimed · AI-structured profileOpenAlex

Mathematical statistician working on asymptotic inference for Riemannian optimization under local differential privacy

papers here

reviewed by Evolve

first-author

corresponding

—

works · OpenAlex

—

h-index · OpenAlex

Bibliometric figures come from OpenAlex (2026-10-13) and cover the author's entire public record, not just the papers listed here. Reproducibility signal comes from Evolve's own reviews.

Research profile

Output—

Output metrics such as publication counts and venue mix are not available in this snapshot, which captures only two 2026 arXiv preprints; the axis is therefore left unscored rather than estimated.

Impact—

Impact signals — citations, citations per paper, and downstream usage — are absent from the input data; with two fresh 2026 preprints and no citation record, this axis remains honestly unscored.

Focus0/100

The radar reports a focus score of 0, while the underlying field distribution splits evenly (0.5/0.5) between 'mathematical statistics / differential privacy / Riemannian optimization' and 'mathematical statistics (nonparametric inference on Riemannian manifolds)'. Read directly, the cataloged work spans just two tightly related labels within one discipline — both papers do inference for manifold-valued targets — so the even split reflects balanced two-branch work rather than scattered topics.

Collaboration36/100

The collaboration score is 36, with 5 distinct coauthors recorded in the snapshot. Both cataloged papers list Hu as last author (not corresponding), and the recurring partner names are Xiaotian Chang and Yangdi Jiang, suggesting a small, stable working group rather than a wide network; the modest score matches that compact footprint.

Activity—

Activity measures such as yearly publication rhythm and recent output velocity cannot be computed from a snapshot whose papers all fall in a single year (2026); this axis is left unscored.

Reproducibility55/100

Reproducibility is the highest scored axis at 55, computed as the mean (55) over 2 Evolve-reviewed papers, each rated medium readiness. Evolve's verdicts flag concrete gaps — per-design safeguard tuning for coverage in the local-DP paper, and fragile small-sample coverage with no baseline comparison in the Gumbel-tube paper — consistent with theory-first work whose computational pipeline is not yet fully validated.

Research summary

Qirui Hu is a mathematical statistician at Shanghai University of Finance and Economics whose cataloged work centers on asymptotic inference for manifold-valued statistical objects. The Evolve catalog holds two 2026 arXiv preprints, on both of which Hu is the last-listed author and not the corresponding author. The first, 'Riemannian Simultaneous Inference for Tangent Vector Field Regression,' develops Gumbel-calibrated simultaneous confidence tubes for kernel regression of tangent vector fields on Riemannian manifolds, built on parallel transport; Evolve's review calls the theory rigorous but notes fragile small-sample coverage and no baseline comparison. The second, 'Locally Private Inference for Riemannian Stochastic Optimization,' claims a central limit theorem and transcript-based Wald inference for manifold-valued minimizers under local differential privacy, combining Polyak-Ruppert averaging, symmetric-pair regression, and Hessian estimation; the review judges it rigorous while flagging that coverage leans on per-design safeguard tuning. Field attribution splits evenly, 0.5 and 0.5, between mathematical statistics with differential privacy and Riemannian optimization, and nonparametric inference on Riemannian manifolds. Hu has five distinct coauthors in this snapshot, with Xiaotian Chang and Yangdi Jiang recurring, and a collaboration score of 36. Reproducibility is the strongest scored axis at 55, with both papers at medium readiness. Output, impact, and activity metrics are not yet available and remain unscored in this early-stage profile.

Collaboration pattern

The snapshot shows a compact collaboration footprint: 5 distinct coauthors and a collaboration score of 36. Xiaotian Chang and Yangdi Jiang are the recurring names in the coauthor list, appearing repeatedly across catalog entries. On both 2026 arXiv preprints Hu is the last-listed author but not the corresponding author, a pattern consistent with a senior or supervisory role within a small, stable group rather than a broad international network.

Research keywords

local differential privacyRiemannian stochastic optimizationPolyak-Ruppert averagingasymptotic inferencesymmetric-pair regressionHessian estimationRiemannian manifoldstangent vector fieldskernel regressionparallel transportsimultaneous confidence bandsGumbel limit

Highlights

Gumbel-calibrated simultaneous bands for tangent vector field regression
The 2026 preprint 'Riemannian Simultaneous Inference for Tangent Vector Field Regression' builds simultaneous confidence tubes for kernel regression of tangent vector fields on Riemannian manifolds, calibrated through a Gumbel limit and relying on parallel transport. Evolve's review calls the theory rigorous but notes fragile small-sample coverage and no baseline comparison.
Wald inference for Riemannian stochastic optimization under local differential privacy
In 'Locally Private Inference for Riemannian Stochastic Optimization' (2026), the work claims a CLT and transcript-based Wald inference for manifold-valued minimizers under local differential privacy, combining Polyak-Ruppert averaging, symmetric-pair regression, and Hessian estimation. The review judges it rigorous, while coverage leans on per-design safeguard tuning.
A consistent inference-first signature on manifolds
Both cataloged papers target uncertainty quantification for manifold-valued objects — simultaneous bands for tangent fields and Wald tests for optimization minimizers — with Hu last author on each and both rated medium reproducibility readiness, marking a coherent methodological agenda at this early stage.

Timeline

2026Preprint 'Riemannian Simultaneous Inference for Tangent Vector Field Regression' appears on arXiv with Hu as last author; Evolve later rates it medium reproducibility readiness.
2026Preprint 'Locally Private Inference for Riemannian Stochastic Optimization' appears on arXiv with Hu as last author; Evolve later rates it medium reproducibility readiness.
2026Both preprints enter the Evolve catalog, forming the current snapshot: 5 distinct coauthors, an even 0.5/0.5 field split, and a reproducibility mean of 55.

Papers in the Evolve catalog

Frequent co-authors

Xiaotian Chang×2 Yangdi Jiang×2 Xiaotian Chang×1 Yangdi Jiang×1 Qirui Hu×1

About this profile

This profile is AI-structured by Evolve.Pub from public bibliographic data (OpenAlex) and Evolve's own paper reviews. It is not written or endorsed by the researcher unless the profile is claimed. Radar dimensions are computed from real bibliometrics with fixed formulas — raw figures are shown alongside; missing dimensions mean insufficient data, never zero. Errors? Claim the profile to correct or hide it.

Bibliometric dimensions are computed from a small OpenAlex-based snapshot covering only two 2026 arXiv preprints; output, impact, and activity axes lack sufficient data and are not scored.
No ORCID or country data is available; 'Qirui Hu' is a common name, so this profile is name-resolved from public bibliographic data and could merge or split same-name researchers.
The reproducibility score (mean 55) rests on only 2 reviewed papers, both rated medium readiness; it does not assess work outside the catalog.
No citation counts, h-index, venue history, or career facts are provided in the input; none are inferred in this profile.