Physics-Audited Agentic Discovery in Scientific Machine Learning

Source

Source name: arXiv
Source kind: paper
Authors / organization: Diab W. Abueidda, Bilal Ahmed, Panos Pantidis, Mostafa E. Mobasher
Venue: arXiv: cs.AI
Published: 2026-07-08
Collected: 2026-07-09
URL: http://arxiv.org/abs/2607.07379v1
PDF: https://arxiv.org/pdf/2607.07379v1
DOI: UNKNOWN

Plain-Language Brief

The paper introduces Physics-Audited Agentic Scientific Machine Learning (PA-SciML), a new workflow that verifies the accuracy of surrogate models discovered by large language model agents in scientific machine learning. The PA-SciML workflow checks each trained candidate on its outputs to ensure they satisfy certain physics principles, such as boundary conditions, superposition, stiffness scaling, and causality. It also records which isolated edits are associated with score gains before reuse.

What the Document Says

PA-SciML is a verification-first workflow for agentic SciML discovery.
The workflow fixes a scoring evaluator before search.
It derives reviewable machine-checkable physics requirements.
Checks each trained candidate on its outputs.
Separately searches prescribed input ranges or measured load-history spans for high-violation cases without reference solution fields.
A surrogate is reported as verified only under the stated checks.
When enabled, the workflow adds advisory numerical probes before training.
Tests one modeling change at a time to record which isolated edits are associated with score gains before reuse.
In computational-solid-mechanics numerical examples, the static elasticity run selects a surrogate with lower validation error than the error-only baseline.
Both selected models pass common linear-elastic checks.
In transient elastodynamics run, an error-only baseline fails a stricter causality check, while the selected surrogate passes the stated checks.
The main distinction is per-candidate physics evidence on predicted fields, not a richer aggregate score.

Key Concepts Mentioned

agentic scientific machine learning
large language model
surrogate models
verification-first workflow
physics requirements
checked outputs
prescribed input ranges
measured load-history spans
static elasticity
transient elastodynamics
common linear-elastic checks
causality check

Claims Made by the Document

The proposed PA-SciML workflow can verify the accuracy of surrogate models.
The workflow can ensure that predicted fields satisfy certain physics principles.

Evidence Used

In computational-solid-mechanics numerical examples, the static elasticity run selects a surrogate with lower validation error than the error-only baseline.
In transient elastodynamics run, an error-only baseline fails a stricter causality check, while the selected surrogate passes the stated checks.

Methods / Approach

UNKNOWN

Limitations Stated by the Authors

UNKNOWN

Limitations Noted by the Agent

Agent interpretation — not from the source.

UNKNOWN

Open Questions

UNKNOWN

Notes for Future Q&A

May include interpretation. Labeled where it does.

UNKNOWN

Raw Abstract / Excerpt

In agentic scientific machine learning (SciML), large language model (LLM) agents can discover surrogate models and select one by an automated score, typically an error metric. A low error, however, does not establish that the predicted fields satisfy the physics that matter for mechanics, such as boundary conditions, superposition, stiffness scaling, or causality. We introduce Physics-Audited Agentic SciML (PA-SciML), a verification-first workflow for agentic SciML discovery. The workflow fixes a scoring evaluator before search, derives reviewable machine-checkable physics requirements, checks each trained candidate on its outputs, and separately searches prescribed input ranges or measured load-history spans for high-violation cases without reference solution fields. A surrogate is reported as verified only under the stated checks. When enabled, the workflow also adds advisory numerical probes before training and tests one modeling change at a time to record which isolated edits are associated with score gains before reuse. In the reported computational-solid-mechanics numerical examples, the static elasticity run selects a surrogate with lower validation error than the error-only baseline while both selected models pass the common linear-elastic checks. In the transient elastodynamics run, an error-only baseline with similar mean error fails a stricter causality check by responding to future parts of the loading history, while the selected surrogate passes the stated checks. The main distinction is per-candidate physics evidence on predicted fields, not a richer aggregate score.