Minhao Zhang

Shanghai University · China

Unclaimed · AI-structured profileORCID OpenAlex

Optimization theorist at Shanghai University working on single-loop extragradient methods for minimax optimization

papers here

reviewed by Evolve

first-author

corresponding

—

works · OpenAlex

h-index · OpenAlex

Bibliometric figures come from OpenAlex (2026-10-15) 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 — publication volume, venue quality, and authorship roles over time — is not scored here: this name-resolved OpenAlex record cannot yet be separated from homonymous authors, so no attributable per-year output count exists for this specific researcher.

Impact—

Impact — citations, citation velocity, and influence beyond raw counts — is not scored: the record lists h-index 23 and i10-index 44, but its mixed topical clusters suggest homonym contamination, so these aggregates cannot be reliably attributed.

Focus0/100

Focus scores 0/100, the lowest axis on this radar. The field distribution itself is coherent — split 50/50 between stochastic minimax optimization / machine learning theory and numerical minimax optimization theory — but the record's top OpenAlex topics (Topological Materials and Phenomena, 29; 2D Materials, 21; thin-film magnetism, 16; underwater vehicles, 13) lie far outside optimization, so homonym publications dilute the measured concentration.

Collaboration32/100

Collaboration scores 32/100, with 4 distinct coauthors recorded. Both reviewed 2026 papers are collaborative: first authorship on the pure single-loop extragradient paper and middle authorship on the stochastic projected damped extragradient paper. Zi Xu is the most frequent partner, Huiling Zhang the other recurring coauthor, and 'Minhao Zhang' also appears, likely a self-duplication — a small, stable theory-group pattern.

Activity93/100

Activity scores 93/100. The record attributes 23 works to the past three years, including 11 in 2026, 6 in 2025, and 6 in 2024, continuing a sustained stretch from 2016 onward that peaked at 15 works in 2018. Both Evolve-reviewed papers appeared in 2026.

Reproducibility55/100

Reproducibility scores 55/100, averaged over 2 papers, both rated medium readiness. Evolve's verdicts note that the strongly convex–strongly concave paper's proof chain is verified but constants are loose and experiments small, while the SPDE/VR-SPDE paper carries rigorous stochastic-first-order-oracle bounds yet no numerical experiments — theory-first work where thin empirical artifacts cap replication confidence.

Research summary

Minhao Zhang (Shanghai University, ORCID 0000-0003-4151-1061) works on the theory of minimax optimization, at the interface of stochastic optimization and machine learning theory. Evolve has reviewed two 2026 arXiv papers. As first author, Zhang proposed a pure single-loop fixed-parameter extragradient method for strongly convex–strongly concave minimax problems with a near-optimal √(κxκy) last-iterate rate; Evolve verified the proof chain while noting loose constants and small-scale experiments. As a middle author, Zhang contributed to single-loop stochastic projected damped extragradient methods (SPDE/VR-SPDE) for nonconvex–(strongly)-concave stochastic minimax problems, establishing strong stochastic-first-order-oracle bounds, though the paper carries no numerical experiments. Recurring themes are single-loop algorithm design, extragradient-type updates, variance reduction, Lyapunov-function analysis, and last-iterate convergence guarantees. The bibliometric record shows high activity (93/100; 23 works in the past three years, 11 recorded in 2026) and moderate reproducibility (55/100 across 2 papers, both medium readiness). Collaboration is concentrated: 4 distinct coauthors, most frequently Zi Xu and Huiling Zhang — a small-team theory-group pattern (32/100). The OpenAlex record behind the aggregate statistics also lists condensed-matter topics, so homonym mixing is likely; see caveats.

Collaboration pattern

Collaboration is narrow and stable: only 4 distinct coauthors are recorded (score 32/100). Zi Xu is by far the most frequent partner, appearing twice in the frequent-coauthor list, with Huiling Zhang the other recurring collaborator; 'Minhao Zhang' also appears, likely a self-record duplication. Both 2026 catalog papers are team efforts — first authorship on one, middle authorship on the other — consistent with a small, tightly linked optimization theory group rather than a broad collaboration network.

Research keywords

minimax optimizationstochastic minimax optimizationnonconvex–strongly concave problemsstrongly convex–strongly concave problemsextragradient methodssingle-loop algorithmsvariance reductionlast-iterate convergenceLyapunov analysisstochastic first-order oracle boundsnear-optimal convergence rates

Highlights

First-author near-optimal pure single-loop extragradient method for SC–SC minimax
In this 2026 arXiv paper (first author), Zhang proposes a pure single-loop fixed-parameter extragradient method achieving a near-optimal √(κxκy) last-iterate rate for strongly convex–strongly concave minimax optimization. Evolve verified the proof chain while noting loose constants and small-scale experiments; reproducibility readiness is rated medium.
Single-loop stochastic projected damped extragradient for nonconvex–(strongly)-concave minimax
In this 2026 arXiv paper (middle authorship), the SPDE and VR-SPDE methods handle stochastic nonconvex–(strongly)-concave minimax problems with strong stochastic-first-order-oracle bounds. Evolve judges the theory rigorous but notes the absence of numerical experiments; reproducibility readiness is rated medium.

Timeline

2018Recorded output peaks at 15 works in the OpenAlex name-resolved record (aggregates may include homonymous authors).
202013 works recorded, part of a sustained high-output stretch running from 2016 through 2023.
202311 works recorded, keeping recent output high alongside 2024 (6 works) and 2025 (6 works).
2026Two minimax-optimization papers enter the Evolve catalog — first-author near-optimal pure single-loop extragradient (SC–SC) and middle-author single-loop stochastic projected damped extragradient (NC–(strongly)-concave); 11 works recorded this year.

Papers in the Evolve catalog

Frequent co-authors

Zi Xu×2 Zi Xu×2 Huiling Zhang×1 Minhao Zhang×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 aggregates (h-index 23, i10-index 44, per-year work counts, topics) come from a name-resolved OpenAlex record whose top topics (topological materials, 2D materials, thin-film magnetism, underwater vehicles) differ from the reviewed optimization papers, so homonym mixing is likely and record-level statistics may overstate this individual.
The output and impact radar axes are unscored because attributable per-author data is insufficient; the raw citation aggregates should not be read as verified personal impact.
Paper-level evidence covers only two 2026 arXiv preprints reviewed by Evolve; peer-reviewed venue versions and any earlier work by this same person are outside the catalog.
The reproducibility score averages just 2 papers, so it is indicative rather than stable.