Search

Agricultural landscapes are quite complex, especially in the Global South where fields are smaller, and agricultural practices are more varied. In this paper we report on our progress in digitizing the agricultural landscape (natural and man-made) in our study region of India. We use high resolution imagery and a UNet style segmentation model to generate the first of its kind national-scale multi-class panoptic segmentation output. Through this work we have been able to identify individual fields across 151.7M hectares, and delineating key features such as water resources and vegetation. We share how this output was validated by our team and externally by downstream users, including some sample use cases that can lead to targeted data driven decision making. We believe this dataset will contribute towards digitizing agriculture by generating the foundational baselayer., Comment: 34 pages, 7 tables, 15 figs

Knowledge Editing (KE) algorithms alter models' internal weights to perform targeted updates to incorrect, outdated, or otherwise unwanted factual associations. In order to better define the possibilities and limitations of these approaches, recent work has shown that applying KE can adversely affect models' factual recall accuracy and diminish their general reasoning abilities. While these studies give broad insights into the potential harms of KE algorithms, e.g., via performance evaluations on benchmarks, we argue little is understood as to why such destructive failures occur. Is it possible KE methods distort representations of concepts beyond the targeted fact, hence hampering abilities at broad? If so, what is the extent of this distortion? To take a step towards addressing such questions, we define a novel synthetic task wherein a Transformer is trained from scratch to internalize a ``structured'' knowledge graph. The structure enforces relationships between entities of the graph, such that editing a factual association has "trickling effects" on other entities in the graph (e.g., altering X's parent is Y to Z affects who X's siblings' parent is). Through evaluations of edited models and analysis of extracted representations, we show that KE inadvertently affects representations of entities beyond the targeted one, distorting relevant structures that allow a model to infer unseen knowledge about an entity. We call this phenomenon representation shattering and demonstrate that it results in degradation of factual recall and reasoning performance more broadly. To corroborate our findings in a more naturalistic setup, we perform preliminary experiments with a pretrained GPT-2-XL model and reproduce the representation shattering effect therein as well. Overall, our work yields a precise mechanistic hypothesis to explain why KE has adverse effects on model capabilities., Comment: Under review

The local modes of trapped ions can be used to construct an analog quantum simulator and a digital quantum computer. However, the control of the phonon hopping remains diffi cult because it proceeds among all the local modes through the Coulomb coupling. We propose a method to cancel the phonon hopping among a given set of local modes by applying a sequence of phase shift gates implemented through the modulation of the trap potential. We analyze the error scaling in the algorithm to treat three or more modes and show that the error can be suppressed by repeating the pulse sequence. The duration of the phase shift gate in the present method can be as short as a few microseconds, which is an order of magnitude faster than the laser-based method. This short duration of the phase shift gate facilitates the suppression of the gate error. The present method can be applied to the simulation of bosonic systems as well as to the continuous variable encoding of quantum computing using trapped ions., Comment: 11 pages, 7 figures

In this paper, we propose a quantum algorithm for solving the linear advection-diffusion equation by employing a new approximate probabilistic imaginary-time evolution (PITE) operator which improves the existing approximate PITE. First, the effectiveness of the proposed approximate PITE operator is justified by the theoretical evaluation of the error. Next, we construct the explicit quantum circuit for realizing the imaginary-time evolution of the Hamiltonian coming from the advection-diffusion equation, whose gate complexity is logarithmic regarding the size of the discretized Hamiltonian matrix. Numerical simulations using gate-based quantum emulator for 1D/2D examples are also provided to support our algorithm. Finally, we extend our algorithm to the coupled system of advection-diffusion equations, and we also compare our proposed algorithm to some other algorithms in the previous works. We find that our algorithm gives comparable result to the Harrow-Hassidim-Lloyd (HHL) algorithm with similar gate complexity, while we need much less ancillary qubits. Besides, our algorithm outperforms a specific HHL algorithm and a variational quantum algorithm (VQA) based on the finite difference method (FDM)., Comment: 58 pages, 23 figures

This paper introduces a data assimilation (DA) framework based on the extended Kalman filter-cell transmission model, designed to assist jam-absorption driving (JAD) operation to alleviate sag traffic congestion. To ascertain and demonstrate the effectiveness of the DA framework for JAD operation, in this paper, we initially investigated its impact on the motion and control performance of a single absorbing vehicle. Numerical results show that the DA framework effectively mitigated underestimated or overestimated control failures of JAD caused by misestimation of key parameters (e.g., free flow speed and critical density) of the traffic flow fundamental diagram. The findings suggest that the proposed DA framework can reduce control failures and prevent significant declines and deteriorations in JAD performance caused by changes in traffic characteristics, e.g., weather conditions or traffic composition., Comment: SMC 2024. Copyright 2024 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works

We study the flavor invariants of the SM augmented by one singlet vector-like quark. Aided by the Hilbert series, we construct all the basic invariants with which any flavor invariant can be written as a polynomial. In special, this theory contains one CP odd invariant of degree six which has degree much lower than the usual Jarlskog invariant of the SM. We find the nonlinear polynomial relations (syzygies) of lowest degrees involving these basic invariants, including the expression of the square of the CP odd invariant of lowest degree in terms of CP even invariants. The $SU(3)$ identity underlying this syzygy is uncovered in terms of invariant tensors, which can be applied to rewrite any square of a CP odd invariant of the same form, involving three hermitean matrices of size three. We demonstrate by an example that there is CP violation that is not detected by the CP odd invariants proposed in the literature so far but it can be detected with the full list of CP odd invariants found here., Comment: 24pp, some tables and many figure files. v2: the set of basic invariants can be extended to the case of the full U(4) weak basis transformations including the VLQ; see around eq.(59). Ap.A now includes the graded Hilbert series. Accepted for JHEP

This study investigates the application of Factorization Machines with Quantum Annealing (FMQA) to address the crystal structure problem (CSP) in materials science. FMQA is a black-box optimization algorithm that combines machine learning with annealing machines to find samples to a black-box function that minimize a given loss. The CSP involves determining the optimal arrangement of atoms in a material based on its chemical composition, a critical challenge in materials science. We explore FMQA's ability to efficiently sample optimal crystal configurations by setting the loss function to the energy of the crystal configuration as given by a predefined interatomic potential. Further we investigate how well the energies of the various metastable configurations, or local minima of the potential, are learned by the algorithm. Our investigation reveals FMQA's potential in quick ground state sampling and in recovering relational order between local minima., Comment: 9 pages, 3 tables, 3 figures

LLM-driven dialog systems are used in a diverse set of applications, ranging from healthcare to customer service. However, given their generalization capability, it is difficult to ensure that these chatbots stay within the boundaries of the specialized domains, potentially resulting in inaccurate information and irrelevant responses. This paper introduces an unsupervised approach for automatically inducing domain-specific dialog flows that can be used to constrain LLM-based chatbots. We introduce two variants of dialog flow based on the availability of in-domain conversation instances. Through human and automatic evaluation over various dialog domains, we demonstrate that our high-quality data-guided dialog flows achieve better domain coverage, thereby overcoming the need for extensive manual crafting of such flows., Comment: Accepted at SIGDIAL 2024

In this article, we compare the methods implementing the real-time evolution operator generated by a unitary diagonal matrix where its entries obey a known underlying real function. When the size of the unitary diagonal matrix is small, a well-known method based on Walsh operators gives a good and precise implementation. In contrast, as the number of qubits grows, the precise one uses exponentially increasing resources, and we need an efficient implementation based on suitable approximate functions. Using piecewise polynomial approximation of the function, we summarize the methods with different polynomial degrees. Moreover, we obtain the overheads of gate count for different methods concerning the error bound and grid parameter (number of qubits). This enables us to analytically find a relatively good method as long as the underlying function, the error bound, and the grid parameter are given. This study contributes to the problem of encoding a known function in the phase factor, which plays a crucial role in many quantum algorithms/subroutines. In particular, we apply our methods to implement the real-time evolution operator for the potential part in the first-quantized Hamiltonian simulation and estimate the resources (gate count and ancillary qubits) regarding the error bound, which indicates that the error coming from the approximation of the potential function is not negligible compared to the error from the Trotter-Suzuki formula.

In this study, we propose a quantum-classical hybrid scheme for performing orbital-free density functional theory (OFDFT) using probabilistic imaginary-time evolution (PITE), designed for the era of fault-tolerant quantum computers (FTQC), as a material calculation method for large-scale systems. PITE is applied to the part of OFDFT that searches the ground state of the Hamiltonian in each self-consistent field (SCF) iteration, while the other parts such as electron density and Hamiltonian updates are performed by existing algorithms on classical computers. When the simulation cell is discretized into $N_\mathrm{g}$ grid points, combined with quantum phase estimation (QPE), it is shown that obtaining the ground state energy of Hamiltonian requires a circuit depth of $O(\log N_\mathrm{g})$. The ground state calculation part in OFDFT is expected to be accelerated, for example, by creating an appropriate preconditioner from the estimated ground state energy for the locally optimal block preconditioned conjugate gradient (LOBPCG) method., Comment: 8 pages, 6 figures

In this study, we employed a quantum computer to solve a low-energy effective Hamiltonian for spin defects in diamond (so-called NV centre) and wurtzite-type aluminium nitride, which are anticipated to be qubits. The probabilistic imaginary-time evolution (PITE) method, designed for use in a fault-tolerant quantum computer (FTQC) era, was employed to calculate the ground and excited states of the spin singlet state, as represented by the effective Hamiltonian. It is difficult to compute the spin singlet state correctly using density functional theory (DFT), which should be described by multiple Slater determinants. To mitigate the effects of quantum errors inherent in current quantum computers, we implemented a $[[ n+2,n,2 ]]$ quantum error detection (QED) code called the Iceberg code. Despite the inevitable destruction of the encoded state resulting from the measurement of the ancilla qubit at each PITE step, we were able to successfully re-encode and recover the logical success state. In the implementation of the PITE, it was observed that the effective Hamiltonian comprises large components of the diagonal part and a relatively small non-diagonal part, which is frequently the case with quantum chemistry calculations. An efficient implementation of Hamiltonian simulations, in which the diagonal components dominate, was developed on a quantum computer based on the second-order Trotter-Suzuki decomposition. This is the first instance of an encoded PITE circuit being executed on a trapped-ion quantum computer. Our results demonstrate that QED effectively reduces quantum errors and that we successfully obtained both the ground and excited states of the spin singlet state. Our demonstration clearly manifests that Zr$_{\rm Al}$V$_{\rm N}$, Ti$_{\rm Al}$V$_{\rm N}$, and Hf$_{\rm Al}$V$_{\rm N}$ defects have a high potential as spin qubits for quantum sensors., Comment: 22 pages, 12 figures

We consider estimating nonparametric time-varying parameters in linear models using kernel regression. Our contributions are twofold. First, We consider a broad class of time-varying parameters including deterministic smooth functions, the rescaled random walk, structural breaks, the threshold model and their mixtures. We show that those time-varying parameters can be consistently estimated by kernel regression. Our analysis exploits the smoothness of the time-varying parameter, which is quantified by a single parameter. The second contribution is to reveal that the bandwidth used in kernel regression determines the trade-off between the rate of convergence and the size of the class of time-varying parameters that can be estimated. We demonstrate that an improper choice of the bandwidth yields biased estimation and provide a guide on the bandwidth selection. An empirical application shows that the kernel-based estimator with a particular bandwidth choice can capture the random-walk dynamics in time-varying parameters.

A real-time beam profile monitoring system is proposed for GeV photon beams at the BM4 beamline of the Mikamine site, Research Center for Accelerator and Radioisotope Science (RARiS; previously known as ELPH) at Tohoku University. This monitoring system enhances the capability to monitor the entire beamline by incorporating newly developed beam profile monitors (BPMs) for upstream and midstream sections, in addition to the existing high-speed BPM used for downstream monitoring. This paper reports on the detection mechanisms of the newly developed BPMs and the actual measurement results obtained using the integrated beam monitoring system. The new BPMs are composed of plastic scintillation fibers and silicon photomultipliers, enabling high-precision, real-time measurements. Data acquisition utilizes streaming TDC, a firmware commonly employed in the J-PARC Hadron-hall, allowing real-time detection of high-intensity photon beams with count rates reaching several tens of MHz. With sufficient statistical data, the BPM achieved a 1-s beam-profiling accuracy of 10 {\mu}m. The proposed BPM system serves as a valuable resource for future physics experiments at the BM4 photon beamline and will significantly contribute to ongoing accelerator research endeavors.

One of the crucial generic techniques for quantum computation is amplitude encoding. Although several approaches have been proposed, each of them often requires exponential classical-computational cost or an oracle whose explicit construction is not provided. Given the growing demands for practical quantum computation, we develop moderately specialized encoding techniques that generate an arbitrary linear combination of localized complex functions. We demonstrate that $n_{\mathrm{loc}}$ discrete Lorentzian functions as an expansion basis set lead to eficient probabilistic encoding, whose computational time is $\mathcal{O}( \max ( n_{\mathrm{loc}}^2 \log n_{\mathrm{loc}},n_{\mathrm{loc}}^2 \log n_q, n_q ))$ for $n_q$ data qubits equipped with $\log_2 n_{\mathrm{loc}}$ ancillae. Furthermore, amplitude amplification in combination with amplitude reduction renders it deterministic analytically with controllable errors and the computational time is reduced to $\mathcal{O}( \max ( n_{\mathrm{loc}}^{3/2} \log n_{\mathrm{loc}}, n_{\mathrm{loc}}^{3/2} \log n_q, n_q )).$ We estimate required resources for applying our scheme to quantum chemistry in real space. We also show the results on real superconducting quantum computers to confirm the validity of our techniques., Comment: 11 figures

We propose hardware-oriented models of intrinsic plasticity (IP) and synaptic plasticity (SP) for spiking randomly connected recursive neural network (RNN). Although the potential of RNNs for temporal data processing has been demonstrated, randomness of the network architecture often causes performance degradation. Self-organization mechanism using IP and SP can mitigate the degradation, therefore, we compile these functions in a spiking neuronal model. To implement the function of IP, a variable firing threshold is introduced to each excitatory neuron in the RNN that changes stepwise in accordance with its activity. We also define other thresholds for SP that synchronize with the firing threshold, which determine the direction of stepwise synaptic update that is executed on receiving a pre-synaptic spike. We demonstrate the effectiveness of our model through simulations of temporal data learning and anomaly detection with a spiking RNN using publicly available electrocardiograms. Considering hardware implementation, we employ discretized thresholds and synaptic weights and show that these parameters can be reduced to binary if the RNN architecture is appropriately designed. This contributes to minimization of the circuit of the neuronal system having IP and SP., Comment: 18 pages, 9 figures, 1 table

We investigate how the solution to the strong CP problem and the explanation for the observed fermion mass hierarchies can be intrinsically related. Specifically, we explore the Nelson-Barr mechanism and identify its "seesaw limit", where light quark masses are suppressed by large CP-violating terms. Upon adding three (two) vector-like quarks that mix with the down-type (up-type) quark sector of the Standard Model, we demonstrate how the lack of CP violation in the strong sector and the observed quark mass hierarchy can be simultaneously achieved., Comment: 16 pages, 6 figures

Multi-task learning has become increasingly popular in the machine learning field, but its practicality is hindered by the need for large, labeled datasets. Most multi-task learning methods depend on fully labeled datasets wherein each input example is accompanied by ground-truth labels for all target tasks. Unfortunately, curating such datasets can be prohibitively expensive and impractical, especially for dense prediction tasks which require per-pixel labels for each image. With this in mind, we propose Joint-Task Regularization (JTR), an intuitive technique which leverages cross-task relations to simultaneously regularize all tasks in a single joint-task latent space to improve learning when data is not fully labeled for all tasks. JTR stands out from existing approaches in that it regularizes all tasks jointly rather than separately in pairs -- therefore, it achieves linear complexity relative to the number of tasks while previous methods scale quadratically. To demonstrate the validity of our approach, we extensively benchmark our method across a wide variety of partially labeled scenarios based on NYU-v2, Cityscapes, and Taskonomy., Comment: Accepted paper to CVPR 2024 (main conference)

Recently, there has been significant attention on online political incivility. While previous research suggests that uncivil political comments lead people to be less willing to see more comments on the same issue, two critical questions have received limited exploration: (1) Are people exposed to uncivil political comments less willing to see other comments from the person who posted the uncivil comment?; (2) Are people exposed to uncivil political comments less willing to see comments from people who have different thoughts than them? To address these questions, the present study conducted a preregistered online survey experiment targeting Japanese citizens, focusing on the pro- vs anti-Kishida cabinet conflict in Japan. The results show that the participants were less willing to see other comments by the person who posted the comment when the comment was uncivil than when it was civil. In addition, the anti-Kishida participants were less willing to see political opinions posted online by people who have different thoughts than them when the comment was uncivil than when it was civil, while the participants in the other subgroups did not show a similar tendency. These findings suggest that uncivil expressions in online political communication might prompt people to avoid reading opinions from those who have different thoughts than them, which might promote political echo chambers.

Stepwise inference protocols, such as scratchpads and chain-of-thought, help language models solve complex problems by decomposing them into a sequence of simpler subproblems. Despite the significant gain in performance achieved via these protocols, the underlying mechanisms of stepwise inference have remained elusive. To address this, we propose to study autoregressive Transformer models on a synthetic task that embodies the multi-step nature of problems where stepwise inference is generally most useful. Specifically, we define a graph navigation problem wherein a model is tasked with traversing a path from a start to a goal node on the graph. Despite is simplicity, we find we can empirically reproduce and analyze several phenomena observed at scale: (i) the stepwise inference reasoning gap, the cause of which we find in the structure of the training data; (ii) a diversity-accuracy tradeoff in model generations as sampling temperature varies; (iii) a simplicity bias in the model's output; and (iv) compositional generalization and a primacy bias with in-context exemplars. Overall, our work introduces a grounded, synthetic framework for studying stepwise inference and offers mechanistic hypotheses that can lay the foundation for a deeper understanding of this phenomenon.

To understand physical properties of the interstellar medium (ISM) on various scales, we investigate it at parsec resolution on the kiloparsec scale. Here, we report on the sub-kpc scale Gas Density Histogram (GDH) of the Milky Way. The GDH is a density probability distribution function (PDF) of the gas volume density. Using this method, we are free from an identification of individual molecular clouds and their spatial structures. We use survey data of $^{12}$CO and $^{13}$CO ($J$=1-0) emission in the Galactic plane ($l = 10^{\circ}$-$50^{\circ}$) obtained as a part of the FOREST Unbiased Galactic plane Imaging survey with the Nobeyama 45m telescope (FUGIN). We make a GDH for every channel map of $2^{\circ} \times 2^{\circ}$ area, including the blank sky component, and without setting cloud boundaries. This is a different approach from previous works for molecular clouds. The GDH fits well to a single or double log-normal distribution, which we name the low-density log-normal (L-LN) and high-density log-normal (H-LN) components, respectively. The multi-log-normal components suggest that the L-LN and H-LN components originate from two different stages of structure formation in the ISM. Moreover, we find that both the volume ratios of H-LN components to total ($f_{\mathrm{H}}$) and the width of the L-LN along the gas density axis ($\sigma_{\rm{L}}$) show coherent structure in the Galactic-plane longitude-velocity diagram. It is possible that these GDH parameters are related to strong galactic shocks and other weak shocks in the Milky Way.