MatOS: An Agent Operating System for Materials Research

Dingyang Lyu; Jing Gao; Zhonghao Chang; Jingjie Huang; Xiangran Cheng; Xin He; Yongda Huang; Zefeng Li

cs.AIOct 25, 2026

MatOS: An Agent Operating System for Materials Research

Authors: Dingyang Lyu, Jing Gao, Zhonghao Chang, Jingjie Huang, Xiangran Cheng, Xin He, Yongda Huang, Zefeng Li,

Organizations: Zhongguancun Academy · Zhongguancun Institute of Artificial Intelligence, Beijing, China

Abstract

Materials research relies on scattered and growing data, knowledge, and simulation tools, and a single study often spans multi-day workflows across literature analysis, computation, and experiment. Large language model agents can now retrieve materials data, operate simulation codes, and close design-synthesis-characterization loops with autonomous laboratories. However, extending these agents with new capabilities, keeping long-running tasks on track, and letting them learn from past tasks and user feedback remain difficult. Here we present MatOS, an agent operating system for materials research with three core components: (1) a layered scientific asset graph that organizes data, knowledge, tools, skills, roles, and workflows as reusable assets; (2) a stateful adaptive workflow that runs tasks as explicit execution graphs with scientific checks, so that they can be resumed and traced; and (3) evidence-driven asset evolution, which revises these assets from execution records and user feedback. On ResearchClawBench, MatOS improves on the previously best-performing agent by 11.15% with the same base model and extends the cost-score Pareto frontier. With a common base model, it outperforms Codex CLI and Claude Code, exceeding Codex CLI by 2.13% on MatTools and 4.17% on SciAgentGym. In eight research cases, MatOS narrowed more than 30 million candidate structures to 97 solid-electrolyte candidates, identified 89 halide variants with higher computed conductivity than their parent structures, linked the force errors of 11 pretrained interatomic potentials to local atomic environments, and supported two new methods that reduce errors by 3.1% and 13.6% relative to their baselines. These results show that organizing research capabilities as reusable, evolving assets enables agents to carry out long, multistage materials studies. Code is available at https://github.com/DingyangLyu/MatOS.

Figures & tables

Figure 1: Overview of the MatOS architecture. Research tasks and user constraints drive the stateful adaptive workflow, while the layered scientific asset graph supplies composable research capabilities. Execution traces and user feedback lead to asset revisions through critique, focused editing, structural checks, and semantic review. Accepted revisions return to the asset graph, and benchmark validation informs subsequent updates.

Symbol	Type	Scientific meaning
$D$	Data	Records of materials systems and their computational or experimental processes, including measurements, simulation outputs, structures, parameters, and references to data sources.
$K$	Knowledge	Generalizations from observations and their conditions of applicability, including physical laws, empirical judgments, and knowledge about the use of methods.
$T$	Tool	Callable capabilities centered on materials solvers, including software interfaces, retrieval and processing routines, validation scripts, and instrument interfaces.
$S$	Skill	A method for a class of problems that specifies how knowledge and tools are selected and used and how results are checked.
$R$	Role	An agent configuration defined by a responsibility prompt and the assets linked to it, which specify the capabilities it may select, combine, and invoke.
$W$	Workflow	A reusable process description that organizes tools, roles, and subworkflows through nodes, dependencies, and control conditions.

Table 1: Symbols and scientific meanings of the six MatOS asset types.

Figure 2: Organization of six scientific asset types and an example centered on a role. (a) Successive abstractions from data through knowledge, tools, and skills to roles and workflows. (b) How a role selects and combines data, knowledge, tools, skills, and workflows in an asset graph.

Relation	Meaning of an edge from $a$ to $b$	Typical type connection
Dependency	The applicability or use of $a$ requires data, knowledge, or a capability supplied by $b$ .	$S\rightarrow D,K,T$ ; $T\rightarrow K$
Mount	$b$ becomes part of the capabilities available to $a$ for selection and composition.	$R\rightarrow D,K,T,S,W$ ; $S\rightarrow K,T$
Call	$a$ may delegate a solution or decision to $b$ and use the returned result.	$R\rightarrow T,R,W$
Containment	A research process described by $a$ contains a stage handled by $b$ , which may be a reusable subprocess.	$W\rightarrow T,R,W$
Evidence	$b$ supplies traceable evidence for a claim, method choice, or result judgment in $a$ .	$K,S,W\rightarrow D,K$
Verification	$b$ specifies or performs a check of the applicability, behavior, or result of $a$ .	$K,S,W\rightarrow T,S,W$

Table 2: Semantics of directed scientific asset edges. An edge

a\rightarrow b

points from the referencing asset to its referent. The last column lists typical type pairs.

Figure 3: Stateful adaptive workflow. (a) Node types. (b) The execution graph (top) connects research tasks, workflow compilation, mapping, scientific checks, loops, and results. The adaptation path (bottom) evaluates graph updates proposed by an agent or researcher through structural, policy, and compilation checks. (c) Persistent state stores graph snapshots, event logs, and output evidence.

Node type	Scientific responsibility
Agent	Assigns open decisions such as problem decomposition, method selection, and anomaly interpretation to a research role.
Human	Handles stages requiring researcher judgment, additional information, or confirmation.
Exec	Runs a specified computation, data-processing operation, or analysis step.
Backend	Organizes a research job on an external computing platform or experimental instrument.
Guard	Checks inputs and intermediate results against scientific or procedural conditions for continuation.
Subworkflow	Invokes a reusable research process as a complete subtask.

Table 3: Node types and scientific responsibilities in the stateful adaptive workflow.

Figure 4: Evidence-driven asset evolution. (a) Evidence critique identifies assets for revision from execution traces and user feedback. Focused edits undergo structural checks and semantic review before adoption. Benchmarks and later tasks supply further feedback. (b) What each stage examines or changes, numbered as in (a).

Benchmark	Type	Tasks used	Metric
ResearchClawBench [ 54 ]	End-to-end research	12 tasks	Rubric score, cost per task
MatTools [ 55 ]	Materials software use	49 problems, 138 subtasks	Subtask and problem success
SciAgentGym [ 56 ]	Multistep tool use	42 cases	Case success rate

Table 4: Benchmarks used in this report.

Figure 5: ResearchClawBench domain and overall scores. Bars show the chemistry, energy, and materials scores; horizontal markers show the overall score, also listed below each system name.

Figure 6: ResearchClawBench mean execution cost versus overall score. Cost is shown on a logarithmic axis. The teal dashed line joins the observed non-dominated points, and the gray dashed line marks the score of MIRA, the highest-scoring baseline.

Figure 7: Materials tool and scientific-task results. (a) MatTools subtask success (solid fill) and complete-problem success (hatched). (b) SciAgentGym case success, with solved-case counts in parentheses.

Figure 8: Energy, force, and stress errors of 11 pretrained potentials on 7,599 multicomponent configurations. (a,c,e) Structure-level distributions of formation-energy error, Cartesian force MAE, and six-component stress MAE. Symbols and intervals denote the mean, median, interquartile range, and 5th to 95th percentiles. (b,d,f) Mean errors within disjoint chemical classes. Rows and columns are ordered by model and class mean errors for each property. Formation energies use aligned model and DFT elemental references. Ch denotes S, Se, or Te, and Hal denotes F, Cl, Br, or I. Adapted from Du et al. [ 73 ] .

Figure 9: Local chemical and geometric contributions to force-error prediction. (a) Out-of-fold predictions of the mean force error across 11 potentials, averaged to the structure level for 7,599 configurations. (b) Reductions in prediction MAE from a constant baseline to a short-contact model, then to the full descriptor model. Smaller bars show additional coordination and neighbor-environment descriptors. Error bars are 95% intervals from 1,000 paired group-bootstrap resamples. (c) Median force error in ten percentile bins of short-contact severity, with 95% bootstrap intervals. (d) Median force error versus short-contact severity and Voronoi heterogeneity for 7,583 valid structures. Grey cells contain fewer than 20 structures. Both coordinates are percentile ranks. $S$ denotes short-contact severity, $C$ central and neighboring chemistry, and $G$ coordination and geometric heterogeneity. Adapted from Du et al. [ 73 ] .

Figure 10: Allegro prediction errors for a single training seed (20260813) on seven MPtrj evaluation sets, with 100,000 training structures per selection method. Models were trained for 300 epochs with a global batch size of 512. (a,b) Energy MAE and RMSE. (c,d) Force-component MAE and RMSE. (e,f) Stress-component MAE and RMSE.

Figure 11: Staged screening of solid electrolytes. Bars show the remaining candidate count on a logarithmic scale, with colors indicating screening stages. Models are listed at left, and percentages give the fraction retained from the previous stage. The screen narrows 30,364,908 initial structures to 97 candidates.

Figure 12: Model benchmarks for hierarchical solid-electrolyte screening. (a) Composition-based L-G-DCNN evaluation: counts of lowest-MAE Matbench tasks, formation-energy prediction and learning curves, and three composition-only extrapolation tests. Extrapolation error bars show standard deviations across five folds. (b) Structure-based property prediction and Matbench comparisons, with MAEs normalized to DenseGNN. (c) Mean absolute errors in $\log_{10}D$ for Li + diffusion in sulfides, oxides, and halides at 1100 and 300 K, using system-specific DeePMD simulations as references. Adapted from Du et al. [ 82 ] .

Figure 13: ORB-v3 predictions against DFT references for 361 halide configurations. (a–c) Energy, force, and stress parity plots. (d–f) Corresponding RMSEs compared with nine other public potentials. The source figure labels ORB-v3 as Halide-MLP*.

Figure 14: Calculated properties of liquid electrolytes E1, E2, and E3. (a–d) Density, viscosity, and conductivity from molecular dynamics. (e) Finite-field response of the neutral solvent. (f) Local Li + solvation from molecular dynamics. (g,h) Molecular quantum-chemical descriptors and thermal properties. Reaction free energies are calculated, while thermal-event temperatures are literature values.

Figure 15: Composition search for a 300 °C melting target. (a) Search score of each valid child in order of completion, colored by island, with the best score so far. Ticks below the axis mark children without a result, and grey bands mark the two interrupted intervals. (b) Predicted liquid phase fraction near 300 °C for the seed and the highest-scoring composition (wt.%). (c,d) Melting interval

\Delta T_{80}

against

T_{s}

and against the residual solid fraction at

T_{80}

for 282 valid children. Dashed lines mark the final thresholds (

\Delta T_{80}=10

°C, residual 0.1), and the green band marks the preferred

T_{s}

window. Three children whose

T_{80}

reached the scan limit are omitted from (c,d).

Figure 16: Test errors for six alloy melting-temperature models. (a) Learning curves over 35 training stages, plotted against the number of training compositions labeled by thermodynamic calculations, on a logarithmic MAE axis. (b) Final model comparison. Both use the same 1,401-composition test set of Sn-containing alloys with melting-interval midpoints from 200 to 400 °C. Reference values come from thermodynamic calculations with a 5 K temperature step.

Figure 17: Dependence of test MAE on (a) melting-interval midpoint and (b) alloy component count. The binary and ternary compositions form the 1,401-composition test set. Separate quaternary and quinary test sets contain 165 and 27 compositions. All test compositions contain Sn and are excluded from training. Sample counts appear below the columns, and the heatmaps use separate color scales.

	Asset organization		Stateful execution		Self-evolution		Domain
System	Modular extension	Capability links	Dynamic orchestration	Persistent recovery	Experience extraction	Update acceptance	Materials focus
Scientific workflow systems

AiiDA [ 38 ]	✓	✓	✓	✓	✗	✗	✓
FireWorks [ 44 ]	✓	✗	✓	✓	✗	✗	✗
Scientific research agents

Table 5: Capabilities of representative systems, grouped as in Section 5 . ✓ indicates an explicit mechanism. ✗ indicates that no explicit mechanism was identified.

Figure 18: Composition and coverage of the MatOS scientific asset library. The five regions show data ( $D$ ), knowledge ( $K$ ), tools and specialized software ( $T$ ), skills ( $S$ ), and roles ( $R$ ), with asset counts at the top right. The data region lists representative datasets, pretrained potentials, and pseudopotential libraries by research object. The knowledge region shows domain guides, method-interface knowledge, instrument interface specifications, and handbooks. In the tool region, callable research interfaces appear on the left and representative integrated software on the right. The skill and role regions list the main method families and specialized roles. Illustrations are schematic. Software marks belong to their respective owners.

Type	Count	Representative content
Data $D$	3,135	Structure databases (Materials Project and Alexandria), training data (MPtrj, MatPES, OMat24, and OC20/OC22), pretrained potentials (MACE, CHGNet, and MatterSim), SSSP and GBRV pseudopotentials, and published research datasets
Knowledge $K$	65	Domain guides and code-resource assessments for twelve research areas, pymatgen method-interface knowledge, instrument interface specifications, indexes of datasets and the machine-learning-potential ecosystem, and handbooks on task readiness and common pitfalls
Tool $T$	194	Database and literature retrieval. Catalogs of models, pseudopotentials, and computing environments. Simulators for XRD, Raman, FTIR, BET, and thermal analysis. High-performance computing job submission and monitoring. PyBaMM solution and Battery Parameter eXchange (BPX) validation
Skill $S$	386	Procedures for band structure and electronic structure, phonons and thermal transport, charged defects, machine-learning potentials, high-throughput screening, molecular quantum chemistry, phase diagrams and phase fields, catalysis and electrochemistry, spectroscopy, and magnetism
Role $R$	51	A computation coordinator and subroles for electronic structure, molecules, phase stability, and other calculations. Domain roles for batteries, phase fields and CALPHAD, thermal transport, high-entropy alloys, porous-material sorption, and experimental XRD. Roles for workflow planning, review, and self-evolution
Workflow $W$	104	Stability screening, gas-sorption screening, charged-defect formation energies, Quantum ESPRESSO (QE) and AiiDA calculation chains, machine-learning-potential benchmarks, diffraction phase analysis, battery discharge simulation, and research and asset-construction planning

Table 6: Scale and representative content of MatOS scientific assets. Tool assets provide callable interfaces to solver backends, including 48 integrated materials software packages.

Figure 19: Execution structure of four representative workflows. (a) Stability screening of known candidate structures. (b) Charged-defect formation energies with QE as the computational backend. (c) Gas-sorption screening in porous frameworks. (d) PyBaMM cell-discharge simulation. Circular markers identify agent (A), execution (E), guard (G), and subworkflow (S) nodes. Steps inside a dashed box are carried out by a single execution node, and the leftmost dashed card is the input. Illustrations are schematic.

Functional layer	Representative systems and tools
Electronic-structure computation	VASP, Quantum ESPRESSO [ 152 , 153 ]
Computational integration and workflows	pyiron, FireWorks, AiiDA, atomate2 [ 180 , 44 , 38 , 58 ]
Open materials data	Materials Project, OQMD [ 35 , 178 ]
Structure generation	MatterGen [ 179 ]
Input and result processing	VASPKIT [ 181 ]
Experimental automation	A-Lab, CRESt [ 16 , 19 ]

Table 7: Functional layers and representative systems in materials research infrastructure.

Figure 20: Centerline velocity profiles of the lid-driven cavity at

\mathrm{Re}=100

, 400, and 1000. Top panels show

u

along

x=0.5

, and bottom panels show

v

along

y=0.5

. Solid lines denote the OpenFOAM solution, filled circles the Ghia et al. reference, and open squares the CFDBench data. The annotation in each panel gives the RMS deviation from Ghia et al. and the relative

L_{2}

difference from CFDBench along that centerline. These values differ from the interior-grid statistics in the text. The circled point in the

\mathrm{Re}=400

v

profile is the tabulated value excluded from the statistics.

Quantity	$\mathrm{Re}=100$	$\mathrm{Re}=400$	$\mathrm{Re}=1000$
RMS deviation of $u$ along $x=0.5$	0.0023	0.0010	0.0028
RMS deviation of $v$ along $y=0.5$	0.0047	0.0025 a	0.0080
$u_{\min}$	$-0.2138$ / $-0.2109$	$-0.3272$ / $-0.3273$	$-0.3864$ / $-0.3829$
$y$ at $u_{\min}$	0.457 / 0.453	0.281 / 0.281	0.171 / 0.172
Primary-vortex center, $x$	0.616 / 0.617	0.553 / 0.555	0.529 / 0.531
Primary-vortex center, $y$	0.734 / 0.734	0.603 / 0.606	0.565 / 0.563

Table 8: Agreement with Ghia et al. for the three square-cavity cases. Velocities are normalized by the lid speed. Paired entries give the present result followed by the reference value. a Excluding the tabulated value at

x=0.9063

discussed in the text.

Figure 21: Steady RANS check of the rectangular cavity geo/case0001 (width 0.01 m, height 0.02 m, $u_{\mathrm{lid}}=10$ m s -1 , $\nu=10^{-5}$ m 2 s -1 ). Left: $u$ field and streamlines from OpenFOAM with the $k$ – $\omega$ SST model. Center: stored CFDBench field. Right: their difference. The right panel title gives the structure correlations, the relative $L_{2}$ difference (rounded), and the minimum $u$ of the two fields.

Benchmark	System	Model and version	Setting	Dates (2026)
ResearchClawBench	MatOS	GPT-5.4 ( gpt-5.4-2026-03-05 )	reasoning effort high	Sep 7–8
	MatOS	DeepSeek-V4-Flash ( DeepSeek-V4-Flash-0731 )	default	Sep 7
	Judge	GPT-5.1 ( gpt-5.1-2025-11-13 )	benchmark rubric	Sep 7–8
MatTools	MatOS	DeepSeek-V4-Pro ( DeepSeek-V4-Pro-Preview )	1M-token context	Jul 6
	Codex CLI 0.142.2	DeepSeek-V4-Pro ( DeepSeek-V4-Pro-Preview )	reasoning effort high	Jul 7
	Claude Code 2.1.191	DeepSeek-V4-Pro ( DeepSeek-V4-Pro-Preview )	effort max	Jul 7

Table 9: Models, versions, settings, and dates of the benchmark runs. Dates are in UTC.

Name	Description	Use in this report	Ref.
MACE	Higher-order equivariant message-passing interatomic potential	Adsorption energies in Material_002 and committee selection of halide configurations	[ 70 ]
MatRIS-10M-OAM	Invariant, attention-based universal potential	One of the 11 evaluated potentials	[ 74 ]
PET-OAM-XL	Point-edge-transformer universal potential	One of the 11 evaluated potentials	[ 75 , 76 ]
DPA-3.1-MPtrj	Line-graph neural network for large atomistic models	One of the 11 evaluated potentials	[ 77 ]
MatterSim	Universal potential across elements, temperatures, and pressures	MD pre-screening of solid-electrolyte candidates	[ 86 ]
DeePMD-kit	Deep-potential training and MD package	Trajectory review of solid-electrolyte candidates	[ 87 ]

Table 10: Models and datasets used in this report.

Name	Description	Use in this report	Ref.
pymatgen	Python library for materials analysis	MatTools tasks	[ 59 ]
pymatgen-analysis-defects	Point-defect extension of pymatgen	MatTools defect and insertion-site tasks	[ 60 ]
ASE	Atomic Simulation Environment	Structure construction and calculator interface	[ 69 ]
RDKit (ETKDGv3)	Cheminformatics toolkit and its distance-geometry conformer generator	Conformers in the rigidity-index task	[ 72 , 71 ]
SOAP	Smooth-overlap-of-atomic-positions descriptor	Environment distances and configuration novelty	[ 190 ]
softBV (BVSE)	Bond-valence site-energy analysis of ion migration	Migration-barrier screening of halides	[ 91 ]

Table 11: Software and methods used in this report.

Autonomous Research Agents

Autonomous research agents use language models and tools to search literature, design and run experiments, analyze results, and support scientific discovery.

See all 319 papers

Sep 30, 2026

CompMat-Bench: Benchmarking AI Agents for Computational Materials Science

Evaluating AI agents on scientific research tasks is constrained by the time and resources required for the underlying experiments or calculations. In computational materials research, repeating the same expensive simulations across agents and trials can make evaluation impractical. We introduce CompMat-Bench, a benchmark of 94 tasks derived from recently published computational materials studies, each asking agents to complete a step toward achieving the study's scientific goal. We reproduce the research steps in advance and assess agents on preparing inputs and analyzing outputs for expensive simulations, so expensive simulations can be avoided during evaluation. The reproduced inputs and results serve as ground truth for grading agents with fixed rules, without an LLM judge. The benchmark supports four evaluation conditions: single tasks and workflows composed of related tasks, each with full or reduced methodological guidance. With full guidance on single tasks, agents based on three LLMs demonstrate the ability to complete individual materials research steps, with pass rates of 66.0-90.4% across 94 tasks. Both longer workflows and reduced guidance can limit agent performance, but in different ways for different agents: they lower the pass rates of the weaker agents, whereas the strongest agent falls only when a long workflow is combined with reduced guidance. Failure analysis attributes most failures to scientific errors rather than to errors in software usage. CompMat-Bench provides a basis for comparing agents on the steps of real materials research and for analyzing agent failure modes.

Chenmu Zhang, Levi Felix, Jun-Jie Zhang +6

Department of Materials Science and NanoEngineering Rice University, Houston, TX 77005, USA

AI Agents for Scientific Discovery AI Agent Evaluation AI Agent Benchmarks Computational Materials Science

Ask PDF

May 13, 2026

OpenAaaS: An Open Agent-as-a-Service Framework for Distributed Materials-Informatics Research

The Materials Genome Initiative catalyzed the proliferation of centralized platforms--SaaS, PaaS, and IaaS--that aggregate computational and experimental resources for accelerated materials discovery. In parallel, breakthroughs in large language models (LLMs) and autonomous agents have created powerful new reasoning capabilities for scientific research. Yet a critical "last mile" problem remains: while we possess world-class models and vast repositories of materials data, we lack the organizational infrastructure to compose these capabilities securely across institutional boundaries. The development of structural and functional materials for harsh service environments--high-temperature alloys, radiation resistant steels, corrosion-resistant coatings--remains characterized by long-term iteration, mechanistic complexity, and high domain expertise--demands that exceed both monolithic agent systems and traditional centralized platforms. To address this gap we propose OpenAaaS, an open-source hierarchical and distributed Agent-as-a-Service framework that enables organized multi-agent collaboration for intelligent materials design. OpenAaaS is built on a single foundational principle: code flows, data stays still. A Master Agent plans and decomposes complex research tasks without requiring direct access to subordinate agents' managed data and computational resources. Sub-agents, deployed as near-data execution nodes, retain full sovereignty over local datasets, proprietary algorithms, and specialized hardware. This architecture guarantees that raw data never leaves its domain of origin while enabling cross-scale, cross-domain secure integration of previously isolated materials intelligence silos. We validate the framework through two representative case studies: (i) AlphaAgent, an evidence-grounded materials literature analysis executor that achieves 4.66/5.0 on deep analytical questions against single-pass RAG baselines; and (ii) an ultra-large-scale hexa-high-entropy alloy descriptor database service that demonstrates secure near-data execution and domain-specific scientific workflows under strict data-sovereignty constraints. OpenAaaS establishes a principled pathway toward "organized research" via agent collectives, offering a scalable foundation for next-generation materials intelligent design platforms. All source code is available at https://github.com/Wolido/OpenAaaS.

Peng Kang, Bixuan Li, Xiaoya Huang +5

National Key Laboratory of AI for Materials Science Beihang University Beijing, China · Tianmushan Laboratory Beihang University Beijing, China

Multi-Agent Orchestration AI Agents for Scientific Discovery Multi-Agent Systems

Ask PDF

May 1, 2026

Can Coding Agents Reproduce Findings in Computational Materials Science?

Large language models are increasingly deployed as autonomous coding agents and have achieved remarkably strong performance on software engineering benchmarks. However, it is unclear whether such success transfers to computational scientific workflows, where tasks require not only strong coding ability, but also the ability to navigate complex, domain-specific procedures and to interpret results in the context of scientific claims. To address this question, we present AutoMat, a benchmark for evaluating LLM-based agents' ability to reproduce claims from computational materials science. AutoMat poses three interrelated challenges: recovering underspecified computational procedures, navigating specialized toolchains, and determining whether the resulting evidence supports a claim. By working closely with subject matter experts, we curate a set of claims from real materials science papers to test whether coding agents can recover and execute the end-to-end workflow needed to support (or undermine) such claims. We then evaluate multiple representative coding agent settings across several foundation models. Our results show that current LLM-based agents obtain low overall success rates on AutoMat, with the best-performing setting achieving a success rate of only 53%. Error analysis further reveals that agents perform worst when workflows must be reconstructed from paper text alone and that they fail primarily due to incomplete procedures, methodological deviations, and execution fragility. Taken together, these findings position AutoMat as both a benchmark for computational scientific reproducibility and a tool for diagnosing the current limitations of agentic systems in AI-for-science settings.

Ziyang Huang, Yi Cao, Ali K. Shargh +15

Johns Hopkins University

Coding Agents AI Agent Benchmarks Scientific Reproducibility Computational Materials Science

Ask PDF

Abstract

Autonomous Research Agents

Explore similar work

CompMat-Bench: Benchmarking AI Agents for Computational Materials Science

OpenAaaS: An Open Agent-as-a-Service Framework for Distributed Materials-Informatics Research

Can Coding Agents Reproduce Findings in Computational Materials Science?