Numerical analyst at the Chinese Academy of Sciences building rigorous foundations for scientific machine learning
3
papers here
1
reviewed by Evolve
3
first-author
0
corresponding
—
works · OpenAlex
19
h-index · OpenAlex
Bibliometric figures come from OpenAlex (2026-10-17) 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—
The output axis tracks publication volume and its trajectory over time. The underlying per-work record is incomplete in the current snapshot, so Evolve leaves it unscored; the per-year counts shown in the bibliometric data are indicative only.
Impact—
Impact measures citation uptake — how widely this work is cited and built upon. The citation-linked evidence needed for scoring is not yet available in this snapshot, so the axis is left unscored rather than estimated.
Focus100/100
Focus scores 100/100, the maximum concentration. Topic tagging places 69 of the tagged works in advanced numerical methods in computational mathematics, with advanced mathematical modeling in engineering (39) and advanced numerical analysis techniques (19) as adjacent clusters, plus differential equations and numerical methods (15) and model reduction and neural networks (13). The reviewed catalog sits entirely in numerical analysis and mathematical machine learning, confirming a tightly bounded agenda.
Collaboration28/100
The collaboration score is 28/100, reflecting a compact network: 3 distinct co-authors in the recent window, with Han Wang and Hao Yu recurring among frequent co-authors ('Hao Yu' appears twice in the source data and may denote two people). All three catalog papers are first-authored without a corresponding role, consistent with a small, stable, analysis-driven team rather than a large distributed collaboration.
Activity86/100
Activity scores 86/100. The counts-by-year series runs unbroken from 1996 through 2026, and the last two years are the series maximum: 7 works each in 2025 and 2026, against 18 works in the most recent three-year window (2024: 4, 2023: 4). This is a sustained and currently accelerating publication rhythm rather than a burst-and-pause pattern.
Reproducibility20/100
Reproducibility scores 20/100, computed from a single assessed paper (reproPapers = 1, mean 20). Across the Evolve catalog, reproducibility readiness is rated low for the random feature and GNN completeness preprints and medium for the Barron regularity paper, whose H/He/Li cusp checks offer a numerical handle. As theory-first arXiv work, most results are theorems rather than replicable pipelines; the score is indicative given the one-paper base.
Research summary
Pingbing Ming, based at the Chinese Academy of Sciences, is a numerical analyst whose recent output targets the mathematical foundations of scientific machine learning. The Evolve catalog holds three 2026 arXiv preprints, each with Ming as first author: a proof of multidimensional spectral convergence for the random feature method (with rigor noted as idealized in its leverage-score sampling), a sharp mixed spectral Barron regularity result for Coulombic many-electron wave functions built on Pauli-driven antisymmetric Fourier-L1 estimates, and a completeness (universal approximation) theory for multi-layer message-passing graph neural network interatomic potentials. The bibliometric snapshot spans 1996–2026 with h-index 19 and i10-index 30, and shows clearly rising activity: 18 works in the last three years, with 2025 and 2026 the two highest annual counts (7 each). Topic tagging is heavily concentrated in advanced numerical methods in computational mathematics (69 works). The recorded collaboration footprint is narrow — 3 distinct recent co-authors, with Han Wang and Hao Yu recurring — and the reproducibility score (20/100) rests on a single assessed paper, with two of three catalog papers rated low reproducibility readiness. Output and impact axes remain unscored for lack of complete citation-linked data.
Collaboration pattern
The recorded collaboration footprint is compact: 3 distinct co-authors in the recent window and a collaboration score of 28/100. Han Wang and Hao Yu recur as frequent co-authors, though 'Hao Yu' appears twice in the source list and may denote two different researchers. All three catalog papers are first-authored with no corresponding-author role recorded, consistent with a small, stable, analysis-driven team rather than a large distributed network.
Research keywords
Highlights
Spectral convergence of the random feature method, proved in multiple dimensions
Sharp mixed spectral Barron regularity for Coulombic wave functions
Completeness theory for message-passing GNN interatomic potentials
Timeline
- 2026First-authored arXiv preprint proves multidimensional spectral convergence of the random feature method.
- 2026First-authored arXiv preprint establishes sharp mixed spectral Barron regularity of Coulombic many-electron wave functions.
- 2026First-authored arXiv preprint gives a completeness theory for multi-layer message-passing GNN interatomic potentials.
Papers in the Evolve catalog
Frequent co-authors
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 indicators come from a single OpenAlex snapshot (identity: openalex); h-index 19 and i10-index 30 are quoted as provided and not independently verified.
- All three catalog papers are 2026 arXiv preprints; peer-review status is not part of the underlying data.
- The reproducibility score rests on one assessed paper (reproPapers = 1), so it is indicative rather than stable.
- 'Hao Yu' appears twice among frequent co-authors and may denote two different researchers; this profile is name-resolved from public bibliographic data.