I am a professor in the Department of Physics at Tongji University, working on the theory of direct nuclear reactions. Most of my work starts from one question: when a loosely bound nucleus hits a target, where does the flux actually go, and which part of that can we hope to measure?
That question has taken me through the Ichimura-Austern-Vincent formalism for inclusive breakup, coupled-channel descriptions of the continuum, and more recently into what a measurement can and cannot constrain at all. Elastic scattering, it turns out, pins down far fewer directions in optical-potential space than the number of parameters we routinely fit. Chasing that has pulled me into emulators, differentiable solvers, and Bayesian inference, not because the machine learning is interesting on its own, but because the physics question needs calculations that are cheap enough to repeat a million times.
I write most of my own solvers. They are listed below.
Background
| Period | Where | With |
|---|---|---|
| 2020 - now | Tongji University, Shanghai | Professor, NSFC PI |
| 2026 - 2027 | SCNT, Institute of Modern Physics, CAS, Huizhou | Visiting Scientist |
| 2019 - 2020 | INFN Pisa | Postdoc, A. Bonaccorso |
| 2016 - 2019 | Ohio University | Postdoc, Ch. Elster |
| 2014 - 2016 | Universidad de Sevilla | PhD, A. M. Moro |
Where the flux goes when a weakly bound projectile survives only partly. The Ichimura-Austern-Vincent formalism, post-prior equivalence, and the coupled-channel version of it. Complete fusion suppression turns out to be a Trojan Horse effect, and incomplete fusion is one-step direct capture. |
Exact flux decompositions of what coupled-channel calculations call absorption. The Feshbach dynamic polarization potential constructed without the weak-coupling approximation, and a uniqueness proof for the coupled-channel Green's function. |
How much a measurement actually carries. Fisher-matrix limits on what elastic scattering can fix in the optical potential, Bayesian calibration of coupled-channel models, and whether chiral effective field theory respects its own power counting. |
Making scattering solvable in a square-integrable basis, then making it fast: direct boundary matching, reduced-basis emulation, physics-informed networks, and GPU linear algebra underneath all of it. |
Faddeev-AGS calculations of three-body halos, and what counts as a halo at all. 6Li behaves as a deuteron halo. |
|
Everything here I wrote or co-wrote. Where a code was released with a paper, the paper is the reference to cite.
| Code | Language | What it does | Released with |
|---|---|---|---|
| SLAM.jl | Julia | General scattering solver on a Lagrange-Legendre basis, direct boundary matching, built to be emulated | PRC 113, 024614 |
| swift.jl | Julia | Three-body Faddeev solver, AV18 / AV14 / Nijmegen with UIX; 3H bound state and Nd scattering with Coulomb and complex scaling | in progress |
| HPRMAT | Fortran + CUDA | High-performance R-matrix linear algebra: direct LU, mixed precision, multi-GPU. 18x speedup at N = 25600 on a single RTX 3090 | CPC, in press |
| COLOSS | Fortran | Complex-scaled two-body scattering with local and Perey-Buck nonlocal optical potentials | CPC 311, 109568 |
| inhomoR | Fortran | Lagrange-mesh R-matrix solver for inhomogeneous equations, with Vincent-Fortune contour integration | PRC 102, 014608 |
| fresco_gui | Fortran | A usable front end for building FRESCO input decks | - |
Some production codes are not public and are available on request:
| Code | Language | What it does | Reference |
|---|---|---|---|
| smoothie | Fortran 95 | Production IAV-DWBA and IAV-CDCC nonelastic breakup, the daily driver behind most of Line A | PRL 123, 232501 |
| PINN-ECS | Python / JAX | Physics-informed network for scattering with an exterior-complex-scaling boundary | PRC 113, 064618 |
| BiLNN | PyTorch | Bidirectional liquid neural network mapping the KD02 optical potential to scattering wave functions, differentiable and global over 1-200 MeV | PRC 114, 014620 |
| transfer | Fortran 95 | DWBA and IAV transfer engine, carries the phase-equivalent nonlocality machinery | in review |
| opticalfisher | Python | Fisher-information analysis of what elastic scattering constrains in the optical potential | in review |
| STARS | Fortran + CUDA | GPU coupled-channels and CDCC production code with a Fortran-side reduced-basis emulator | - |
Recent
-
Inclusive breakup with nonspectator fragments: generalization of the IAV sum rules Phys. Rev. C 114, 014632 (2026) · doi · arXiv:2604.11226 Removes the spectator approximation on the detected fragment. Standard IAV turns out to be the total inclusive cross section summed over the fragment's internal states.
-
Reduced basis emulator for elastic scattering in CDCC Phys. Rev. C 113, 044610 (2026) · doi Proper orthogonal decomposition plus Galerkin projection, 220x faster, sub-0.1% on an 18-parameter problem.
-
Direct boundary matching: a bound-state technique for nuclear scattering Phys. Rev. C 113, 024614 (2026) · doi Scattering boundary conditions without Bloch operators. Released as SLAM.jl.
-
Exterior complex scaling enables physics-informed neural networks for nuclear reactions Phys. Rev. C 113, 064618 (2026) · doi The trick that makes PINNs work here: damp the exterior wave so the optical potential stays on the real axis.
Earlier, and still the backbone
-
Puzzle of complete fusion suppression in weakly bound nuclei: a Trojan Horse effect? Phys. Rev. Lett. 122, 042503 (2019) · doi
-
Unraveling the reaction mechanisms leading to partial fusion of weakly bound nuclei Phys. Rev. Lett. 123, 232501 (2019) · doi First IAV-CDCC calculation. Incomplete fusion is one-step direct capture.
-
Numerical assessment of post-prior equivalence for inclusive breakup reactions Phys. Rev. C 92, 061602(R) (2015), Editor's Suggestion · doi Closes a thirty-year controversy about the IAV and Udagawa-Tamura formulations.
Full list on Google Scholar or the website.
Tongji University, Shanghai · SCNT / IMP-CAS, Huizhou
Happy to talk about breakup reactions, optical potentials, emulators, or anything in the codes above. Students and visitors welcome.