Intertwined charge, spin, and pairing orders in doped iron ladders

Motivated by recent experimental progress on iron-based ladder compounds, we study the doped two-orbital Hubbard model for the two-leg ladder BaFe$_2$S$_3$. The model is constructed by using {it ab initio} hopping parameters and the ground state properties are investigated using the density matrix renormalization group method. We show that the $(pi,0)$ magnetic ordering at half-filling, with ferromagnetic rungs and antiferromagnetic legs, becomes incommensurate upon hole doping. Moreover, depending on the strength of the Hubbard $U$ coupling, other magnetic patterns, such as $(0,pi)$, are also stabilized. We found that the binding energy for two holes becomes negative for intermediate Hubbard interaction strength, indicating hole pairing. Due to the crystal-field split among orbitals, the holes primarily reside in one orbital, with the other one remaining half-filled. This resembles orbital selective Mott states. The formation of tight hole pairs continues with increasing hole density, as long as the magnetic order remains antiferromagnetic in one direction. The study of pair-pair correlations indicates the dominance of the intra-orbital spin-singlet channel, as opposed to other pairing channels. Although in a range of hole doping pairing correlations decay slowly, our results can also be interpreted as corresponding to a charge-density-wave made of pairs, a precursor of eventual superconductivity after interladder couplings are included. Such scenario of intertwined orders has been extensively discussed before in the cuprates, and our results suggest a similar physics could exist in ladder iron-based superconductors. Finally, we also show that a robust Hunds coupling is needed for pairing to occur.

Prediction of exotic magnetic states in the alkali metal quasi-one-dimensional iron selenide compound Na$_2$FeSe$_2$

The magnetic and electronic phase diagram of a model for the quasi-one-dimensional alkali metal iron selenide compound Na$_2$FeSe$_2$ is presented. The novelty of this material is that the valence of iron is Fe$^{2+}$ contrary to most other iron-chain compounds with valence Fe$^{3+}$. Using first-principles techniques, we developed a three-orbital tight-binding model that reproduces the {it ab initio} band structure near the Fermi level. Including Hubbard and Hund couplings and studying the model via the density matrix renormalization group and Lanczos methods, we constructed the ground state phase diagram. A robust region where the block state $uparrow uparrow downarrow downarrow uparrow uparrow downarrow downarrow$ is stabilized was unveiled. The analog state in iron ladders, employing 2$times$2 ferromagnetic blocks, is by now well-established, but in chains a block magnetic order has not been observed yet in real materials. The phase diagram also contains a large region of canonical staggered spin order $uparrow downarrow uparrow downarrow uparrow downarrow uparrow$ at very large Hubbard repulsion. At the block to staggered transition region, a novel phase is stabilized with a mixture of both states: an inhomogeneous orbital-selective charge density wave with the exotic spin configuration $uparrow uparrow downarrow uparrow downarrow downarrow uparrow downarrow$. Our predictions for Na$_2$FeSe$_2$ may guide crystal growers and neutron scattering experimentalists towards the realization of block states in one-dimensional iron-selenide chain materials.

BCS-BEC crossover in a $(t_{2g})^4$ Excitonic Magnet

The condensation of spin-orbit-induced excitons in $(t_{2g})^4$ electronic systems is attracting considerable attention. In the large Hubbard U limit, antiferromagnetism was proposed to emerge from the Bose-Einstein Condensation (BEC) of triplons ($J_{textrm{eff}} = 1$). In this publication, we show that even for the weak and intermediate U regimes, the spin-orbit exciton condensation is possible leading also to staggered magnetic order. The canonical electron-hole excitations (excitons) transform into local triplon excitations at large U , and this BEC strong coupling regime is smoothly connected to the intermediate U excitonic insulator region. We solved the degenerate three-orbital Hubbard model with spin-orbit coupling ($lambda$) in one-dimensional geometry using the Density Matrix Renormalization Group, while in two-dimensional square clusters we use the Hartree-Fock approximation (HFA). Employing these techniques, we provide the full $lambda$ vs U phase diagrams for both one- and two- dimensional lattices. Our main result is that at the intermediate Hubbard U region of our focus, increasing $lambda$ at fixed U the system transitions from an incommensurate spin-density-wave metal to a Bardeen-Cooper-Schrieffer (BCS) excitonic insulator, with coherence length r coh of O(a) and O(10a) in 1d and 2d, respectively, with a the lattice spacing. Further increasing $lambda$, the system eventually crosses over to the BEC limit (with r coh << a).

Fine-grained Uncertainty Modeling in Neural Networks

Existing uncertainty modeling approaches try to detect an out-of-distribution point from the in-distribution dataset. We extend this argument to detect finer-grained uncertainty that distinguishes between (a). certain points, (b). uncertain points but within the data distribution, and (c). out-of-distribution points. Our method corrects overconfident NN decisions, detects outlier points and learns to say ``I dont know when uncertain about a critical point between the top two predictions. In addition, we provide a mechanism to quantify class distributions overlap in the decision manifold and investigate its implications in model interpretability. Our method is two-step: in the first step, the proposed method builds a class distribution using Kernel Activation Vectors (kav) extracted from the Network. In the second step, the algorithm determines the confidence of a test point by a hierarchical decision rule based on the chi-squared distribution of squared Mahalanobis distances. Our method sits on top of a given Neural Network, requires a single scan of training data to estimate class distribution statistics, and is highly scalable to deep networks and wider pre-softmax layer. As a positive side effect, our method helps to prevent adversarial attacks without requiring any additional training. It is directly achieved when the Softmax layer is substituted by our robust uncertainty layer at the evaluation phase.

Adversarial TCAV -- Robust and Effective Interpretation of Intermediate Layers in Neural Networks

Interpreting neural network decisions and the information learned in intermediate layers is still a challenge due to the opaque internal state and shared non-linear interactions. Although (Kim et al, 2017) proposed to interpret intermediate layers by quantifying its ability to distinguish a user-defined concept (from random examples), the questions of robustness (variation against the choice of random examples) and effectiveness (retrieval rate of concept images) remain. We investigate these two properties and propose improvements to make concept activations reliable for practical use. Effectiveness: If the intermediate layer has effectively learned a user-defined concept, it should be able to recall --- at the testing step --- most of the images containing the proposed concept. For instance, we observed that the recall rate of Tiger shark and Great white shark from the ImageNet dataset with Fins as a user-defined concept was only 18.35% for VGG16. To increase the effectiveness of concept learning, we propose A-CAV --- the Adversarial Concept Activation Vector --- this results in larger margins between user concepts and (negative) random examples. This approach improves the aforesaid recall to 76.83% for VGG16. For robustness, we define it as the ability of an intermediate layer to be consistent in its recall rate (the effectiveness) for different random seeds. We observed that TCAV has a large variance in recalling a concept across different random seeds. For example, the recall of cat images (from a layer learning the concept of tail) varies from 18% to 86% with 20.85% standard deviation on VGG16. We propose a simple and scalable modification that employs a Gram-Schmidt process to sample random noise from concepts and learn an average concept classifier. This approach improves the aforesaid standard deviation from 20.85% to 6.4%.

Flat Bands and Ferrimagnetic Order in Electronically Correlated Dice-Lattice Ribbons

We study ribbons of the dice two-dimensional lattice (that we call ``dice ladders) known to have nontrivial topological properties, such as Chern numbers 2 [Wang and Y. Ran, Phys. Rev. B {bf 84}, 241103 (2011)]. Our main results are two folded: (1) Analyzing the tight-binding model in the presence of Rashba spin-orbit coupling and an external magnetic field, we observed that dice ladders qualitatively display properties similar to their two-dimensional counterpart all the way to the limit of only two legs in the short direction. This includes flat bands near the Fermi level, edge currents and edge charge localization near zero energy when open boundary conditions are used, two chiral edge modes, and a nonzero Hall conductance. (2) We studied the effect of Hubbard correlation $U$ in the two-leg dice ladder using Lanczos and density matrix renormalization group techniques. We show that increasing $U$ the flat bands split without the need of introducing external fields. Moreover, robust ferrimagnetic order develops. Overall, our work establishes dice ladders as a promising playground to study the combined effect of topology and correlation effects, one of the frontiers in Quantum Materials.