Call for paper: Machine Learning for Physics: PINNs, CNNs, and Beyond
Lontar Physics Today invites researchers and practitioners to submit original manuscripts for a Special Issue on Machine Learning for Physics: Physics-Informed Neural Networks (PINNs), Convolutional Neural Networks (CNNs), and Beyond. This Special Issue aims to publish rigorous, reproducible work on how neural network–based methods support physics modeling, simulation, inference, and data analysis across theoretical, computational, and experimental contexts.
Neural networks are increasingly used not only for prediction, but also as physics-aware modeling tools—where physical laws, constraints, and domain knowledge are embedded into training and optimization. In particular, PINNs integrate governing equations and physical constraints to tackle forward and inverse problems in ODE/PDE systems. Meanwhile, CNNs are a core deep-learning architecture for structured data such as images and spatial fields, making them widely used for physics imaging, diagnostics, tomography, and pattern extraction from complex measurements.
This Special Issue includes broader scientific machine learning approaches that complement PINNs and CNNs, such as neural operators, surrogate modeling/emulators, hybrid physics–ML solvers, and learning frameworks for uncertainty quantification, generalization, and benchmarking. Alongside rapid progress, key challenges remain—reliability, robustness to noise, stability in stiff/multiscale regimes, and transparent evaluation. We therefore welcome contributions that provide clear methodological advances and careful comparisons against established physics/numerical baselines.
Scope and Topics (non-exhaustive)Submissions may include (but are not limited to):
- PINNs and physics-informed learning for forward/inverse problems in ODE/PDE systems
- CNNs and deep learning for physics data, including image- and field-based analysis (e.g., diagnostics, reconstruction, detection)
- Scientific ML beyond PINNs/CNNs, such as neural operators, surrogate modeling, hybrid physics–ML solvers, and uncertainty quantification
- Benchmarking and reproducibility: fair comparisons, baseline solvers, ablation studies, datasets, and open/reusable code (where possible)
- Trend and evidence synthesis studies, including systematic literature reviews (SLR), meta-analyses, and bibliometric analyses on neural networks/scientific ML in physics
- Empirical research (computational/experimental/theoretical validation)
- Methodological papers with clear technical contributions and careful evaluation
- Review articles, systematic literature reviews (SLR), meta-analyses, and bibliometric studies
All submissions to this Special Issue will undergo the journal’s standard peer-review process, identical to regular issues. Submission does not guarantee acceptance. Manuscripts will be published only after successfully passing the review and revision stages.
Important DatesSubmission opens: 1 February 2026
Submission deadline: 28 February 2027
Publication: Articles will be published after the review process is completed and subsequently collected in the Special Issue.
Submit manuscripts via OJS:
https://journal3.upgris.ac.id/index.php/lpt/about/submissions
All submissions must follow the journal’s Author Guidelines:
https://journal3.upgris.ac.id/index.php/lpt/authorguidelines
During submission, please select the section: “Special Issue: Machine Learning for Physics: PINNs, CNNs, and Beyond.” In the submission comments or cover letter, please include:
“Special Issue: Machine Learning for Physics: PINNs, CNNs, and Beyond.”
Editorial Office: [email protected]
