arXiv subjects

Ming-Wei Li

Publications and source records attributed to Ming-Wei Li.

10 recordsLinked to original sources

arXiv2026

A light DM model for large $B \to K + \mbox{invisible}$ and $K \to \pi + \mbox{invisible}$ decays and its implications for $B_s-\bar B_s$ mixing and neutron EDM

We study the implications for $B_s - \bar B_s$ mixing and the neutron electric dipole moment (EDM) in a light dark matter model with sizable invisible rare meson decays to accommodate the recent possible deviations from Standard Model (SM) predictions observed in $B^+\to K^+\nu\bar\nu$ by Belle II and $K^+\to\pi^+\nu\bar\nu$ by NA62. Given that the neutrinos in these decays escape detection, they can be replaced by other invisible final states. Based on effective operator analysis, it has been proposed that branching ratios for $B^+\to K^+ +\mbox{invisible}$ and $K^+\to\pi^+ +\mbox{invisible}$ can naturally be larger than the SM predictions due to the emission of light dark matter pairs. We demonstrate that this scenario can be realized within a UV-complete two-Higgs-doublet model (2HDM) where neutral Higgs bosons mediating dark matter interactions induce significant low-energy effects especially for $B_s-\bar B_s$ mixing and neutron EDM. Within the allowed parameter space, we find non-negligible contributions to $B_s - \bar B_s$ mixing. For neutron EDM, there is a cancellation due to the exchange of neutral spin-zero particle, but QCD renormalization group evolution will lift this cancellation which in fact is generally true for any neutral Higgs contribution. However, we demonstrate that such a cancellation does not occur for charged scalar contributions. Ultimately, the allowed CP-violating phases in the Yukawa sector can generate a neutron EDM at a level consistent with current bound.

hep-ph↗

arXiv2025

The $U(1)_{L_μ-L_τ}$ model meets the new $(g-2)_μ$ data and muon neutrino trident scattering

The Muon $g-2$ collaboration at Fermilab has announced their final result of the anomalous magnetic moment of the muon. By adopting the lattice-QCD evaluation of the leading-order hadronic-vacuum-polarization, this result is now in agreement with the latest theoretical prediction to the $1σ$ level. This new result further constrains the allowed parameter space, but does not rule out all possible new physics contributions the muon $g-2$. We study the implications for one of the relevant models, the gauged $U(1)_{L_μ- L_τ}$. When using this model to resolve the previous $4σ$ tension, results from muon neutrino trident (MNT) scattering experiments would restrict the mass of the new gauge boson ($Z'$) to be less than $300$ MeV. Since the theory and experimental data difference for muon $g-2$ is lowered down to $1σ$, the requirement for $m_{Z'}\lesssim 300\,{\rm MeV}$ is much relaxed. Within the updated allowed range of $Z'$ boson mass, we study the models implications for electron and tauon $g-2$ as well as future muon colliders. We find that muon collider can effectively probe the $U(1)_{L_μ- L_τ}$.

hep-ph↗

arXiv2024

The SM expected branching ratio for $h \to γγ$ and an excess for $h \to Z γ$

The recent measurements of $h \to Z γ$ from ATLAS and CMS show an excess of the signal strength $μ_Z = (σ\cdot{\cal B})_{\mathrm{obs}}/(σ\cdot{\cal B})_{\mathrm{SM}}=2.2\pm 0.7$, normalized as 1 in the standard model~(SM). If confirmed, it would be a signal of new physics (NP) beyond the SM. We study NP explanation for this excess. In general, for a given model, it also affects the process $h \to γγ$. Since the measured branching ratio for this process agrees well with the SM prediction, the model is severely constrained. We find that a minimally fermion singlets and doublet extended NP model can explain simultaneously the current data for $h \to Z γ$ and $h\to γγ$. There are two solutions. Although both solutions enhance the amplitude of $h \to Z γ$ to the observed one, in one of the solutions the amplitude of $h \to γγ$ flips sign to give the observ ed branching ratio. This seems to be a contrived solution although cannot be ruled out simply using branching ratio measurements alone. However, we find another solution that naturally enhances $h \to Z γ$ to the measured value, but keeps the amplitude of $h \to γγ$ close to its SM prediction. We also comment on the phenomenology associated with these new fermions.

hep-ph↗

arXiv2024

Majorana Phase And Matter Effects In Neutrino Chiral Oscillation

Due to finite masses and mixing, for neutrinos propagation in space-time, there is a chiral oscillation between left- and right- chiral neutrinos, besides the usual oscillation between different generations. The probability of chiral oscillation is suppressed by a factor of $m^2/E^2$ making the effect small for relativistic neutrinos. However, for non-relativistic neutrinos, this effect can be significant. In matter, the equation of motion is modified. When neutrinos produced in weak interaction pass through the matter, the eigen-energies are split into two different ones depending on the helicity of the neutrino. This results in different oscillation behavior for neutrinos with different helicity, in particular there is a new resonant effect related to the helicity state of neutrino different than the usual MSW effect. For Majorana neutrinos, chiral oscillation also depends on Majorana phases.

hep-ph↗

arXiv2024

On the Evaluation Metric for Hashing

Due to its low storage cost and fast query speed, hashing has been widely used for large-scale approximate nearest neighbor (ANN) search. Bucket search, also called hash lookup, can achieve fast query speed with a sub-linear time cost based on the inverted index table constructed from hash codes. Many metrics have been adopted to evaluate hashing algorithms. However, all existing metrics are improper to evaluate the hash codes for bucket search. On one hand, all existing metrics ignore the retrieval time cost which is an important factor reflecting the performance of search. On the other hand, some of them, such as mean average precision (MAP), suffer from the uncertainty problem as the ranked list is based on integer-valued Hamming distance, and are insensitive to Hamming radius as these metrics only depend on relative Hamming distance. Other metrics, such as precision at Hamming radius R, fail to evaluate global performance as these metrics only depend on one specific Hamming radius. In this paper, we first point out the problems of existing metrics which have been ignored by the hashing community, and then propose a novel evaluation metric called radius aware mean average precision (RAMAP) to evaluate hash codes for bucket search. Furthermore, two coding strategies are also proposed to qualitatively show the problems of existing metrics. Experiments demonstrate that our proposed RAMAP can provide more proper evaluation than existing metrics.

cs.IR↗

arXiv2024

Multiple Code Hashing for Efficient Image Retrieval

Due to its low storage cost and fast query speed, hashing has been widely used in large-scale image retrieval tasks. Hash bucket search returns data points within a given Hamming radius to each query, which can enable search at a constant or sub-linear time cost. However, existing hashing methods cannot achieve satisfactory retrieval performance for hash bucket search in complex scenarios, since they learn only one hash code for each image. More specifically, by using one hash code to represent one image, existing methods might fail to put similar image pairs to the buckets with a small Hamming distance to the query when the semantic information of images is complex. As a result, a large number of hash buckets need to be visited for retrieving similar images, based on the learned codes. This will deteriorate the efficiency of hash bucket search. In this paper, we propose a novel hashing framework, called multiple code hashing (MCH), to improve the performance of hash bucket search. The main idea of MCH is to learn multiple hash codes for each image, with each code representing a different region of the image. Furthermore, we propose a deep reinforcement learning algorithm to learn the parameters in MCH. To the best of our knowledge, this is the first work that proposes to learn multiple hash codes for each image in image retrieval. Experiments demonstrate that MCH can achieve a significant improvement in hash bucket search, compared with existing methods that learn only one hash code for each image.

cs.LG↗

arXiv2023

Type-II Seesaw Triplet Scalar Effects on Neutrino Trident Scattering

In Type-II seesaw model, an electroweak triplet scalar field $Δ$ with a non-zero vacuum expectation value (vev) $v_Δ$ is introduced to facilitate the generation of small neutrino masses. A non-zero $v_Δ$ also affects the W mass through the electroweak $ρ$ parameter, making it to be less than 1 as predicted by standard model (SM). The component fields in $Δ$ come along introduce additional contributions to reduce the SM rare neutrino trident scattering cross section. These fields also induce new processes not existed in SM, such as $l_i \to \overline{ l_j} l_k l_l$ and $l_i \to l_j γ$. There are severe constraints on these processes which limit the effects on neutrino trident scattering and the $ρ$ parameter and therefore the W mass. The newly measured W mass by CDF makes the central value of $ρ$ parameter to be larger than 1, even larger than previously expected. Combining neutrinoless double beta decay, direct neutrino mass and oscillation data, we find a lower limit for $v_Δ$ as a function of the triplet scalar mass $m_Δ$, $v_Δ> (6.3 \sim 8.4) \mathrm{eV} (100 \mathrm{GeV}/m_Δ)$. To have significant effect on $ρ$ in this model, $v_Δ$ needs to be in the range of a GeV or so. However this implies a very small $m_Δ$ which is ruled out by data. We conclude that the effect of triplet vev $v_Δ$ on the W mass can be neglected. We also find that at 3$σ$ level, the deviation of the ratio for Type-II Seesaw to SM neutrino trident scattering cross section predictions is reduced to be below 1, but is restricted to be larger than 0.98.

hep-ph↗

arXiv2021

Magnetic moments and transition magnetic moments of $P_c$ and $P_{cs}$ states

We study the magnetic moments and transition magnetic moments of $P_c$ and $P_{cs}$ states in the molecular picture. We first revisit the magnetic moments of $P_c$ states as the $S$ wave molecular states without coupled channel effects. The coupled channel effects and the $D$ wave contributions are then investigated carefully. The coupled channel effects contribute to the change of $0.1\sim 0.4$ nuclear magneton $μ_N$ for most cases while the $D$ wave only induces the variation of less than $0.03 ~μ_N$. In addition, we obtain the transition magnetic moments between different $P_c$ states and the related electromagnetic decay widths of $P_c'\to P_cγ$. The magnetic moments of $P_{cs}$ are much different for the assumption of spin being 1/2 and 3/2. The study of electromagnetic properties will help us disclose further the structure of these unconventional states.

hep-ph↗

arXiv2019

Deep Multi-Index Hashing for Person Re-Identification

Traditional person re-identification (ReID) methods typically represent person images as real-valued features, which makes ReID inefficient when the gallery set is extremely large. Recently, some hashing methods have been proposed to make ReID more efficient. However, these hashing methods will deteriorate the accuracy in general, and the efficiency of them is still not high enough. In this paper, we propose a novel hashing method, called deep multi-index hashing (DMIH), to improve both efficiency and accuracy for ReID. DMIH seamlessly integrates multi-index hashing and multi-branch based networks into the same framework. Furthermore, a novel block-wise multi-index hashing table construction approach and a search-aware multi-index (SAMI) loss are proposed in DMIH to improve the search efficiency. Experiments on three widely used datasets show that DMIH can outperform other state-of-the-art baselines, including both hashing methods and real-valued methods, in terms of both efficiency and accuracy.

cs.CV↗

arXiv2018

Proximal SCOPE for Distributed Sparse Learning: Better Data Partition Implies Faster Convergence Rate

Distributed sparse learning with a cluster of multiple machines has attracted much attention in machine learning, especially for large-scale applications with high-dimensional data. One popular way to implement sparse learning is to use $L_1$ regularization. In this paper, we propose a novel method, called proximal \mbox{SCOPE}~(\mbox{pSCOPE}), for distributed sparse learning with $L_1$ regularization. pSCOPE is based on a \underline{c}ooperative \underline{a}utonomous \underline{l}ocal \underline{l}earning~(\mbox{CALL}) framework. In the \mbox{CALL} framework of \mbox{pSCOPE}, we find that the data partition affects the convergence of the learning procedure, and subsequently we define a metric to measure the goodness of a data partition. Based on the defined metric, we theoretically prove that pSCOPE is convergent with a linear convergence rate if the data partition is good enough. We also prove that better data partition implies faster convergence rate. Furthermore, pSCOPE is also communication efficient. Experimental results on real data sets show that pSCOPE can outperform other state-of-the-art distributed methods for sparse learning.

stat.ML↗