Google researcher in convex optimization: oracle complexity, stochastic gradients, privacy-preserving optimization.
4
papers here
4
reviewed by Evolve
0
first-author
0
corresponding
—
works · OpenAlex
10
h-index · OpenAlex
Bibliometric figures come from OpenAlex (2026-10-16) 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—
Not yet scored: publication-level metadata is incomplete in this OpenAlex snapshot, so total output volume cannot be summarized. The only signals on file are 17 works in the last three years; the axis would measure career output size and trajectory.
Impact—
Not yet scored: citation-level aggregation is insufficient for this name-resolved profile, so impact is not estimated. The axis would measure citation counts, h-index growth, and influence on subsequent work; the h-index of 10 is the sole citation signal on file.
Focus0/100
Not yet scored: topic classification is available as raw counts but no consolidated focus score exists. The axis would measure how concentrated research is across topics; the snapshot lists five topic clusters, led by stochastic gradient optimization (34).
Collaboration44/100
Scores 28 on Evolve's collaboration axis, reflecting a small working circle: the snapshot records only 3 distinct coauthors, with David Martínez-Rubio, Brian Bullins, and Mathieu Molina as frequent collaborators. Both cataloged 2026 preprints place him in middle or last author position without corresponding-author roles, consistent with concentrated, theory-style two-to-three-person partnerships rather than large consortia.
Activity83/100
Scores 84 on Evolve's activity axis. The year-by-year record shows 17 works in the most recent three-year window and two pronounced peaks — 13 works in 2021 and 14 in 2024 — after quieter years (1 work in 2020, 4 in 2023). Output continues into 2025-2026 with the two Lipschitz optimization preprints, indicating a sustained and currently active research pace.
Reproducibility20/100
Not yet scored: no systematic artifact or code-release data is available for this profile. Evolve tags both cataloged 2026 preprints as low reproducibility readiness, but the axis would measure released code, proofs, and benchmarks across the full body of work.
Research summary
Cristóbal Guzmán is a researcher at Google (United States) working on the theory of convex optimization. His OpenAlex-resolved snapshot records an h-index of 10, an i10-index of 12, and 17 works over the most recent three-year window, with yearly output peaking at 13 works in 2021 and 14 in 2024. Topic counts are led by stochastic gradient optimization techniques (34), followed by sparse and compressive sensing (21), privacy-preserving technologies (11), graph complexity and algorithms (10), and advanced optimization algorithms (9). The two Evolve-reviewed entries, both 2026 arXiv preprints, target the nonsmooth COLT 2015 open problem on Lipschitz convex optimization in nondual norm settings: one proposes a polynomial-time randomized bundle method, built on stable selectors, with nearly optimal oracle rates for p < min{q, 2}; the other derives first-order minimax oracle complexity bounds via max-regret analysis and the sequential fat-shattering dimension, though its algorithm is exponential-time. Evolve judged the mathematics coherent but not line-verified. His radar shows high activity (84) against low collaboration intensity (28), consistent with a small, stable coauthor circle centered on optimization theory.
Collaboration pattern
Guzmán's collaboration footprint is narrow but stable: only 3 distinct coauthors appear in the recent snapshot, and the collaboration dimension scores 28. David Martínez-Rubio, Brian Bullins, and Mathieu Molina recur as frequent coauthors, and both cataloged 2026 preprints fit a compact theory-paper authorship pattern in which he holds middle or last position without corresponding authorship. No large-scale consortium or industry-academic network is visible in the provided data.
Research keywords
Highlights
Polynomial-time bundle method for nondual Lipschitz optimization
Minimax oracle complexity for first-order methods in nondual settings
A decade-old open problem, attacked from both directions
Timeline
- 2011First recorded publication year in the OpenAlex snapshot, with 1 work.
- 2021Peak output year with 13 recorded works, within the period dominated by the stochastic gradient and sparsity topic clusters.
- 2024Highest single-year count in the snapshot: 14 recorded works.
- 2026Two arXiv preprints on nondual Lipschitz convex optimization enter the Evolve catalog, both reviewed as resolving the nonsmooth COLT 2015 open problem.
Papers in the Evolve catalog
Near-Optimal Acceleration for Smooth $\ell_p$ / $\ell_q$ Nondual Convex First-Order Oracle Optimization
Stable Movement for Nondual Lipschitz Convex Optimization: Efficiency and Nearly Optimal Oracle Rates
The First-Order Oracle Complexity of Lipschitz Convex Optimization in Nondual Settings
Oracle Complexity of Stochastic Fixed-Point Equations with Nonexpansive Maps
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.
- Bibliometrics (h-index 10, i10-index 12, counts by year, topics) come from an OpenAlex-resolved snapshot; identity was matched from public bibliographic data and may merge or split namesakes.
- Output, impact, focus, and reproducibility dimensions are unscored because the underlying data is insufficient; no estimates are implied.
- The qualitative assessment rests on only two Evolve-reviewed papers (both 2026 arXiv preprints), judged mathematically coherent but not line-verified and rated low for reproducibility readiness.
- Yearly publication counts contain gaps (no records for 2012, 2016, or 2019), so the trajectory may be incomplete under name resolution.