Ameya D. Jagtap

Worcester Polytechnic Institute · United States

Unclaimed · AI-structured profileORCID OpenAlex

Scientific ML researcher at WPI working on physics-informed neural networks, gradient conflicts, and inverse problems.

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—

Publication-count metrics for this axis are not yet in our snapshot, so no score is given. This dimension would measure cumulative and annual scholarly output; the reviewed record so far shows two 2026 arXiv preprints with last-author placement, but a scored assessment needs the full indexed record.

Impact—

Citation-based impact metrics (citations, field-weighted influence) are not yet available for scoring here. This axis would measure how widely the work is cited and used; only index-level figures (h-index 20, i10-index 28) exist in the snapshot, without per-paper citation detail.

Focus0/100

The radar focus score is 0, reflecting a notably dispersed topical profile. OpenAlex groups the work into five clusters — model reduction and neural networks (42), neural networks and applications (36), computational fluid dynamics and aerodynamics (17), advanced numerical methods (17), and nonlinear waves and solitons (7) — and the field distribution splits evenly at 0.5 computational solid mechanics (scientific machine learning) and 0.5 scientific machine learning (PDE solvers).

Collaboration36/100

Collaboration scores 36, the lowest of the present axes. Recent-window data lists only 5 distinct coauthors against 28 works in three years, pointing to small, stable teams rather than a wide ad-hoc network; recurring partners include Trishit Mondal, Prajwal Bharadwaj, Nikhil Karanjgaokar, Sidharth S. Menon, and Irina Tezaur. On both reviewed papers he is listed last without the corresponding-author flag, consistent with a supervisory role.

Activity98/100

Activity is the standout axis at 98. Per-year indexed counts climb from 1 in 2016 to 3–4 across 2017–2021, 7 in 2022, 8 in 2023, 5 in 2024, 8 in 2025, and 15 already in 2026, with 28 works in the latest three-year window. Aside from the 2024 dip to 5, the cadence has accelerated sharply, making 2026 the strongest year on record.

Reproducibility55/100

Reproducibility reads 55, computed from 2 reviewed papers with a mean readiness of 55; both are rated medium. GenVoid pairs a generative level-set prior with a real FEM solver, reporting strong synthetic results but modest experimental IoU; Norm-PCGrad shows strong 2D/3D elliptic evidence, though two-loss theory and fixed hyperparameters temper its claims. A two-paper base makes this a provisional signal.

Research summary

Ameya D. Jagtap is a researcher at Worcester Polytechnic Institute (US) whose work sits at the intersection of scientific machine learning and computational mechanics. OpenAlex bibliometric data (identity: openalex) records an h-index of 20 and an i10-index of 28, with annual output climbing from 1 indexed work in 2016 to 7 in 2022, 8 in 2023, 8 in 2025, and 15 so far in 2026, and 28 works in the most recent three-year window. His topical footprint spans five OpenAlex clusters — model reduction and neural networks (42 works), neural networks and applications (36), computational fluid dynamics and aerodynamics (17), advanced numerical methods (17), and nonlinear waves and solitons (7) — with an even 50/50 split between computational solid mechanics (scientific machine learning) and scientific machine learning (PDE solvers). Two of his recent arXiv preprints are reviewed in the Evolve catalog: GenVoid, an uncertainty-aware physics-informed generative model for identifying subsurface material voids against a real FEM solver with experimental validation, and Norm-PCGrad, which addresses gradient conflicts in domain-decomposed PINNs and PIKANs. On both reviewed papers he appears in the last author position without the corresponding-author flag, a pattern consistent with senior supervision. The radar shows exceptionally high activity (98), medium reproducibility readiness (55 across 2 papers), and comparatively limited collaboration breadth (36, with 5 distinct recent coauthors).

Collaboration pattern

Radar collaboration reads 36, with only 5 distinct coauthors in the recent window against 28 works, suggesting compact, stable working groups rather than wide ad-hoc networks. Recurring partners are Trishit Mondal, Prajwal Bharadwaj, Nikhil Karanjgaokar, Sidharth S. Menon, and Irina Tezaur. On both reviewed 2026 papers he sits in the last author position without the corresponding-author flag, consistent with senior supervision of mainly early-career collaborators.

Research keywords

Scientific machine learningPhysics-informed neural networks (PINNs)Physics-informed Kolmogorov–Arnold networks (PIKANs)PDE solversDomain decompositionConflicting gradients and gradient surgeryInverse problemsPhysics-informed generative modelsUncertainty quantificationNondestructive evaluationComputational solid mechanicsModel reduction

Highlights

Experimentally validated physics-informed generative model for subsurface void identification
GenVoid (2026, arXiv) pairs a generative level-set prior with a real FEM solver for hidden-void identification in materials, with uncertainty awareness and experimental validation. Evolve's review notes strong synthetic results but modest experimental IoU, and rates reproducibility medium; Jagtap is the last-listed author.
Conflict-free gradients for domain-decomposed PINNs and PIKANs
Norm-PCGrad (2026, arXiv) tackles failure modes of physics-informed neural networks and KANs using gradient surgery to curb conflicting gradients in domain-decomposed training. Evolve notes strong 2D/3D elliptic evidence, tempered by two-loss theory and fixed hyperparameters; reproducibility is rated medium and Jagtap is the last-listed author.

Timeline

2016First indexed work appears in the OpenAlex record, with 1 publication that year.
2022Indexed output rises to 7 works, beginning a multi-year acceleration from the 3–4 works per year seen in 2017–2021.
20258 works indexed in the year; 28 works accumulate over the trailing three-year window.
2026Strongest year on record with 15 indexed works so far; the Evolve catalog reviews two arXiv preprints — GenVoid and Norm-PCGrad — with Jagtap listed last among authors.

Papers in the Evolve catalog

Frequent co-authors

Trishit Mondal×1 Prajwal Bharadwaj×1 Nikhil Karanjgaokar×1 Sidharth S. Menon×1 Irina Tezaur×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 figures (h-index, i10-index, per-year counts, topics, coauthor counts) come from an OpenAlex snapshot; the output and impact dimensions lack underlying data and are left unscored.
Only two papers are reviewed in the Evolve catalog (both 2026 arXiv preprints), so qualitative judgments rest on a small, very recent sample.
The reproducibility score (55) is based on just 2 papers and should be read as a provisional signal.
The profile is name-resolved from public bibliographic data (OpenAlex identity; ORCID 0000-0002-8831-1000); biography beyond the listed affiliation is out of scope.