arXiv subjects

Peter Hirschfeld

Publications and source records attributed to Peter Hirschfeld.

4 recordsLinked to original sources

arXiv2026

Gap structure and phase diagram of twisted bilayer cuprates from a microscopic perspective

Since the prediction of a time-reversal symmetry breaking (TRSB) $d+id^\prime$ state in twisted bilayer cuprate superconductors by Can et al. (Nat.Phys. 17,519(2021)), several experiments have attempted to detect this state, yielding conflicting results. At present, it is not clear which differences in samples or experimental conditions might explain these discrepancies. In this work, we perform a tight-binding lattice model calculation with phenomenological interlayer tunneling, examining the order parameter as a function of twist angle, interlayer tunneling, doping, and temperature. We observe the TRSB state to be correlated to the position of the Van Hove singularity in the normal state, which changes not only as a function of doping but also the tunneling strength. Two such phases are identified as nominally consistent with in-plane $d+id'$ and $d+is$ order, but with unexpected transformation properties under bilayer symmetry operations. We calculate the Josephson critical current, in particular examining the angle dependence for various tunneling strengths. Finally, we discuss the existing experiments in the context of our results.

cond-mat.supr-con↗

arXiv2026

Reinforcement Learning on the Discrete Composition Channel of a Crystal Generator: Validated Gains and Reward Hacking

Inverse materials design is a long-standing goal of computational materials discovery. Generative models for crystalline materials are typically trained to match the distribution of a structure database, while nothing in their training objective points them at specific design goals such as targeted properties. We use group-relative policy optimization (GRPO) to align a generative model based on stochastic interpolants and discrete flow matching with general black-box reward functions through reinforcement learning. Atom types are generated by a discrete flow and the policy gradient of our generalization of GRPO directly acts on the likelihoods of the atom-type transitions, which differentiates our work from previous reinforcement-learning approaches for diffusion and flow-based generative models of crystalline materials. We introduce a reward function that raises the yield of metastable, unique and novel structures (mSUN) from 13.4% for the pretrained model to 45.5% for the reinforced model, as evaluated by a community benchmark. Our reward also improves the performance of a reinforcement learning framework for crystalline materials based on latent denoising diffusion models. At the same time, we find that directly reinforcing atom-type transition likelihoods enables reward exploitation that has to be prevented with explicit guards. The same analysis also exposes a gap in the community metric. Single-element structures in distinct packings are counted as metastable, unique and novel materials and inflate mSUN without yielding any new compounds. A stability claim is only as good as its reference hull. We report every result split by the number of reference phases behind it and argue that benchmarks should do the same.

cs.LG↗

arXiv2026

Thermodynamics of $T_{\rm c}$ suppression in far-overdoped Tl$_2$Ba$_2$CuO$_6$

The physical origin of the suppression of superconductivity with hole doping in overdoped cuprates remains unclear. We measure the electronic specific heat of microgram-scale Tl$_2$Ba$_2$CuO$_6$ crystals and find sharp superconducting anomalies persisting far into the overdoped regime. A weak-coupling BCS-like framework incorporating the known Fermi surface and cation disorder quantitatively reproduces the observed anomalies for $T_{\rm c}=14$-$25$ K and their weak doping dependence. The results show $T_{\rm c}(p)$ to be driven predominantly by a smoothly decreasing pairing strength.

cond-mat.supr-con↗

arXiv2001

The Kondo lattice model from strong-coupling viewpoint

We present some preliminary results on the phase diagram of the 2D S=1/2 Kondo lattice model at finite doping. As a starting point the Hamiltonian is written in terms of local spin and charge excitations, and the interactions between these modes are subsequently treated in various perturbative schemes. We find that a paramagnetic-magnetic quantum phase transition does occur, and, at least on a superficial level, the Kondo effect does not break down at the critical point. The latter result however might well be a consequence of the inherent bias of our starting point and/or the level of approximation.

cond-mat.str-el↗