Computational mechanics + ML at Brown: topology optimization, learned constitutive models, learned optimizer selection.
3
papers here
2
reviewed by Evolve
0
first-author
0
corresponding
—
works · OpenAlex
—
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—
Not yet scored: Evolve lacks a reliable full-publication record for this profile, so total output, venue spread, and productivity trends cannot be computed. This axis would measure publication volume and its growth over time.
Impact—
Not yet scored: citation counts and h-index are unavailable in the current data, so field-level influence cannot be assessed. This axis would measure citations, citations per paper, and uptake of the reviewed methods.
Focus0/100
Scores 0/100, the lowest axis in the radar: the field distribution splits evenly at 0.5/0.5 between structural topology optimization and machine learning for constitutive modeling, and the reviewed RL2CO paper adds a third strand in learning to optimize. The program is thematically coherent — differentiable and learned computational mechanics — but the even two-field split keeps the measured focus score at the floor.
Collaboration44/100
Scores 44/100, with 8 distinct coauthors in the snapshot. Recurring partners include Martin van der Schelling, Deepesh Toshniwal, Suryanarayanan Manoj Sanu, Alejandro Marcos Aragón, Rui Barreira, Taylan Soydan, Francesco Scipione, and Dirk Mohr. All three reviewed papers are multi-author efforts in which Bessa holds a middle or last rather than first or corresponding position, consistent with distributed, team-based authorship rather than solo work.
Activity—
Not yet scored: recent per-year publication rates cannot be verified from the available snapshot, which lists only 2026 arXiv preprints. This axis would measure publishing cadence and recent-year momentum across venues.
Reproducibility55/100
Scores 55/100, computed as a mean over 2 scored papers (reproMean 55, reproPapers 2). Among the reviewed work, the RL2CO optimizer-selection paper rates high for reproducibility readiness, while BOTH and the constitutive state-space framework rate medium — Evolve's verdicts cite thin baselines, incomplete stress-case results, and synthetic-only benchmarks.
Research summary
Miguel Anibal Bessa is a researcher at Brown University working where computational mechanics meets machine learning. The Evolve catalog places him in two balanced strands — structural topology optimization and machine learning for constitutive modeling, each 0.5 of his field distribution — plus a third interest in learning to optimize. Three 2026 arXiv preprints anchor the profile. BOTH (Bilevel Optimization of Topology and Hyperparameters) reports the first gradient-based bilevel tuning of topology-optimization hyperparameters via automatic differentiation; Evolve finds the mechanics well verified but baselines thin and stress-case results incomplete. A constitutive state-space framework adapts S5 dynamics to path-dependent plasticity, matching or beating MSC on accuracy while leading on resolution robustness and speed, though evidence is limited to synthetic benchmarks. RL2CO uses reinforcement learning to choose which optimizer to run and for how long, with strong results tempered by a gap-maximizing test set and missing prior-switch baselines. The radar scores collaboration at 44 with 8 distinct coauthors, reproducibility at 55 (mean over 2 scored papers), and focus at 0, reflecting the even two-field split; output, impact, and activity lack sufficient data. Recurring coauthors include Martin van der Schelling, Deepesh Toshniwal, Suryanarayanan Manoj Sanu, Alejandro Marcos Aragón, Rui Barreira, Taylan Soydan, Francesco Scipione, and Dirk Mohr. In all three reviewed papers Bessa appears in a middle or last author position.
Collaboration pattern
The snapshot records 8 distinct coauthors and a collaboration score of 44. Recurring partners — Martin van der Schelling, Deepesh Toshniwal, Suryanarayanan Manoj Sanu, Alejandro Marcos Aragón, Rui Barreira, Taylan Soydan, Francesco Scipione, and Dirk Mohr — span topology optimization, constitutive modeling, and optimization ML. In all three reviewed papers Bessa sits in a middle or last author position without a corresponding flag, suggesting a contributing or senior role within multi-author teams rather than a first-author lead.
Research keywords
Highlights
First gradient-based bilevel tuning of topology-optimization hyperparameters
S5-based state-space model for path-dependent plasticity
Reinforcement learning to choose optimizers
Timeline
- 2026Publishes BOTH on arXiv — the first gradient-based bilevel tuning of topology-optimization hyperparameters via automatic differentiation.
- 2026Introduces a constitutive state-space modeling framework that adapts S5 dynamics to path-dependent plasticity (arXiv).
- 2026Co-authors RL2CO on arXiv, using reinforcement learning to select optimizers and their runtimes.
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 dimensions come from the OpenAlex-based radar snapshot only; output, impact, and activity scores are missing from the current data.
- All three reviewed papers are 2026 arXiv preprints; peer-reviewed venue status is not reflected in this profile.
- None of the reviewed papers lists Bessa as first or corresponding author (middle, middle, last), limiting leadership attribution.
- The reproducibility mean (55) is computed over only 2 scored papers, a thin base for that axis.