arXiv subjects

Ran Zhou

Publications and source records attributed to Ran Zhou.

At least 37 records · Page 2Linked to original sources

arXiv2022

MReD: A Meta-Review Dataset for Structure-Controllable Text Generation

When directly using existing text generation datasets for controllable generation, we are facing the problem of not having the domain knowledge and thus the aspects that could be controlled are limited. A typical example is when using CNN/Daily Mail dataset for controllable text summarization, there is no guided information on the emphasis of summary sentences. A more useful text generator should leverage both the input text and the control signal to guide the generation, which can only be built with a deep understanding of the domain knowledge. Motivated by this vision, our paper introduces a new text generation dataset, named MReD. Our new dataset consists of 7,089 meta-reviews and all its 45k meta-review sentences are manually annotated with one of the 9 carefully defined categories, including abstract, strength, decision, etc. We present experimental results on start-of-the-art summarization models, and propose methods for structure-controlled generation with both extractive and abstractive models using our annotated data. By exploring various settings and analyzing the model behavior with respect to the control signal, we demonstrate the challenges of our proposed task and the values of our dataset MReD. Meanwhile, MReD also allows us to have a better understanding of the meta-review domain.

cs.CL↗

arXiv2022

MELM: Data Augmentation with Masked Entity Language Modeling for Low-Resource NER

Data augmentation is an effective solution to data scarcity in low-resource scenarios. However, when applied to token-level tasks such as NER, data augmentation methods often suffer from token-label misalignment, which leads to unsatsifactory performance. In this work, we propose Masked Entity Language Modeling (MELM) as a novel data augmentation framework for low-resource NER. To alleviate the token-label misalignment issue, we explicitly inject NER labels into sentence context, and thus the fine-tuned MELM is able to predict masked entity tokens by explicitly conditioning on their labels. Thereby, MELM generates high-quality augmented data with novel entities, which provides rich entity regularity knowledge and boosts NER performance. When training data from multiple languages are available, we also integrate MELM with code-mixing for further improvement. We demonstrate the effectiveness of MELM on monolingual, cross-lingual and multilingual NER across various low-resource levels. Experimental results show that our MELM presents substantial improvement over the baseline methods.

cs.CL↗

arXiv2019

$|V_{us}|$ from $K_{\ell 3}$ decay and four-flavor lattice QCD

Using HISQ $N_f=2+1+1$ MILC ensembles with five different values of the lattice spacing, including four ensembles with physical quark masses, we have performed the most precise computation to date of the $K\toπ\ellν$ vector form factor at zero momentum transfer, $f_+^{K^0π^-}(0)=0.9696(15)_\text{stat}(12)_\text{syst}$. This is the first calculation that includes the dominant finite-volume effects, as calculated in chiral perturbation theory at next-to-leading order. Our result for the form factor provides a direct determination of the Cabibbo-Kobayashi-Maskawa matrix element $|V_{us}|=0.22333(44)_{f_+(0)}(42)_\text{exp}$, with a theory error that is, for the first time, at the same level as the experimental error. The uncertainty of the semileptonic determination is now similar to that from leptonic decays and the ratio $f_{K^+}/f_{π^+}$, which uses $|V_{ud}|$ as input. Our value of $|V_{us}|$ is in tension at the 2--$2.6σ$ level both with the determinations from leptonic decays and with the unitarity of the CKM matrix. In the test of CKM unitarity in the first row, the current limiting factor is the error in $|V_{ud}|$, although a recent determination of the nucleus-independent radiative corrections to superallowed nuclear $β$ decays could reduce the $|V_{ud}|^2$ uncertainty nearly to that of $|V_{us}|^2$. Alternative unitarity tests using only kaon decays, for which improvements in the theory and experimental inputs are likely in the next few years, reveal similar tensions. As part of our analysis, we calculated the correction to $f_+^{Kπ}(0)$ due to nonequilibrated topological charge at leading order in chiral perturbation theory, for both the full-QCD and the partially-quenched cases. We also obtain the combination of low-energy constants in the chiral effective Lagrangian $[C_{12}^r+C_{34}^r-(L_5^r)^2](M_ρ)=(2.92\pm0.31)\cdot10^{-6}$.

hep-lat↗

arXiv2019

$D$ meson Semileptonic Decay Form Factors at $q^2 = 0$

We discuss preliminary results for the vector form factors $f_+^{\{π,K\}}$ at zero-momentum transfer for the decays $D\toπ\ellν$ and $D\to K \ellν$ using MILC's $N_f = 2+1+1$ HISQ ensembles at four lattice spacings, $a \approx 0.042, 0.06, 0.09$, and 0.12 fm, and various HISQ quark masses down to the (degenerate) physical light quark mass. We use the kinematic constraint $f_+(q^2)= f_0(q^2)$ at $q^2 = 0$ to determine the vector form factor from our study of the scalar current, which yields $f_0(0)$. Results are extrapolated to the continuum physical point in the framework of hard pion/kaon SU(3) heavy-meson-staggered $χ$PT and Symanzik effective theory. Our calculation improves upon the precision achieved in existing lattice-QCD calculations of the vector form factors at $q^2=0$. We show the values of the CKM matrix elements $|V_{cs}|$ and $|V_{cd}|$ that we would obtain using our preliminary results for the form factors together with recent experimental results, and discuss the implications of these values for the second row CKM unitarity.

hep-lat↗

arXiv2019

$B_s\to K\ellν$ decay from lattice QCD

We use lattice QCD to calculate the form factors $f_+(q^2)$ and $f_0(q^2)$ for the semileptonic decay $B_s\to K\ellν$. Our calculation uses six MILC asqtad 2+1 flavor gauge-field ensembles with three lattice spacings. At the smallest and largest lattice spacing the light-quark sea mass is set to 1/10 the strange-quark mass. At the intermediate lattice spacing, we use four values for the light-quark sea mass ranging from 1/5 to 1/20 of the strange-quark mass. We use the asqtad improved staggered action for the light valence quarks, and the clover action with the Fermilab interpolation for the heavy valence bottom quark. We use SU(2) hard-kaon heavy-meson rooted staggered chiral perturbation theory to take the chiral-continuum limit. A functional $z$ expansion is used to extend the form factors to the full kinematic range. We present predictions for the differential decay rate for both $B_s\to Kμν$ and $B_s\to Kτν$. We also present results for the forward-backward asymmetry, the lepton polarization asymmetry, ratios of the scalar and vector form factors for the decays $B_s\to K\ellν$ and $B_s\to D_s \ellν$. Our results, together with future experimental measurements, can be used to determine the magnitude of the Cabibbo-Kobayashi-Maskawa matrix element $|V_{ub}|$.

hep-lat↗

arXiv2017

$B_s \to K \ellν$ form factors with 2+1 flavors

Using the MILC 2+1 flavor asqtad quark action ensembles, we are calculating the form factors $f_0$ and $f_+$ for the semileptonic $B_s \rightarrow K \ellν$ decay. A total of six ensembles with lattice spacing from $\approx0.12$ to 0.06 fm are being used. At the coarsest and finest lattice spacings, the light quark mass $m'_l$ is one-tenth the strange quark mass $m'_s$. At the intermediate lattice spacing, the ratio $m'_l/m'_s$ ranges from 0.05 to 0.2. The valence $b$ quark is treated using the Sheikholeslami-Wohlert Wilson-clover action with the Fermilab interpretation. The other valence quarks use the asqtad action. When combined with (future) measurements from the LHCb and Belle II experiments, these calculations will provide an alternate determination of the CKM matrix element $|V_{ub}|$.

hep-lat↗

arXiv2016

Kaon semileptonic decays with $N_f=2+1+1$ HISQ fermions and physical light-quark masses

We discuss the reduction of errors in the calculation of the form factor $f_+^{K π}(0)$ with HISQ fermions on the $N_f=2+1+1$ MILC configurations from increased statistics on some key ensembles, new data on ensembles with lattice spacings down to 0.042 fm and the study of finite-volume effects within staggered ChPT. We also study the implications for the unitarity of the CKM matrix in the first row and for current tensions with leptonic determinations of $\vert V_{us}\vert$.

hep-lat↗

arXiv2016

$B^0_{(s)}$-mixing matrix elements from lattice QCD for the Standard Model and beyond

We calculate---for the first time in three-flavor lattice QCD---the hadronic matrix elements of all five local operators that contribute to neutral $B^0$- and $B_s$-meson mixing in and beyond the Standard Model. We present a complete error budget for each matrix element and also provide the full set of correlations among the matrix elements. We also present the corresponding bag parameters and their correlations, as well as specific combinations of the mixing matrix elements that enter the expression for the neutral $B$-meson width difference. We obtain the most precise determination to date of the SU(3)-breaking ratio $ξ= 1.206(18)(6)$, where the second error stems from the omission of charm sea quarks, while the first encompasses all other uncertainties. The threefold reduction in total uncertainty, relative to the 2013 Flavor Lattice Averaging Group results, tightens the constraint from $B$ mixing on the Cabibbo-Kobayashi-Maskawa (CKM) unitarity triangle. Our calculation employs gauge-field ensembles generated by the MILC Collaboration with four lattice spacings and pion masses close to the physical value. We use the asqtad-improved staggered action for the light valence quarks, and the Fermilab method for the bottom quark. We use heavy-light meson chiral perturbation theory modified to include lattice-spacing effects to extrapolate the five matrix elements to the physical point. We combine our results with experimental measurements of the neutral $B$-meson oscillation frequencies to determine the CKM matrix elements $|V_{td}| = 8.00(34)(8) \times 10^{-3}$, $|V_{ts}| = 39.0(1.2)(0.4) \times 10^{-3}$, and $|V_{td}/V_{ts}| = 0.2052(31)(10)$, which differ from CKM-unitarity expectations by about 2$σ$. These results and others from flavor-changing-neutral currents point towards an emerging tension between weak processes that are mediated at the loop and tree levels.

hep-lat↗

arXiv2016

$B\to Kl^+l^-$ decay form factors from three-flavor lattice QCD

We compute the form factors for the $B \to Kl^+l^-$ semileptonic decay process in lattice QCD using gauge-field ensembles with 2+1 flavors of sea quark, generated by the MILC Collaboration. The ensembles span lattice spacings from 0.12 to 0.045 fm and have multiple sea-quark masses to help control the chiral extrapolation. The asqtad improved staggered action is used for the light valence and sea quarks, and the clover action with the Fermilab interpretation is used for the heavy $b$ quark. We present results for the form factors $f_+(q^2)$, $f_0(q^2)$, and $f_T(q^2)$, where $q^2$ is the momentum transfer, together with a comprehensive examination of systematic errors. Lattice QCD determines the form factors for a limited range of $q^2$, and we use the model-independent $z$ expansion to cover the whole kinematically allowed range. We present our final form-factor results as coefficients of the $z$ expansion and the correlations between them, where the errors on the coefficients include statistical and all systematic uncertainties. We use this complete description of the form factors to test QCD predictions of the form factors at high and low $q^2$. We also compare a Standard-Model calculation of the branching ratio for $B \to Kl^+l^-$ with experimental data.

hep-lat↗

arXiv2016

Phenomenology of semileptonic B-meson decays with form factors from lattice QCD

We study the exclusive semileptonic $B$-meson decays $B\to K(π)\ell^+\ell^-$, $B\to K(π)ν\barν$, and $B\toπτν$, computing observables in the Standard model using the recent lattice-QCD results for the underlying form factors from the Fermilab Lattice and MILC Collaborations. These processes provide theoretically clean windows into physics beyond the Standard Model because the hadronic uncertainties are now under good control for suitably binned observables. For example, the resulting partially integrated branching fractions for $B\toπμ^+μ^-$ and $B\to Kμ^+μ^-$ outside the charmonium resonance region are 1-2$σ$ higher than the LHCb Collaboration's recent measurements, where the theoretical and experimental errors are commensurate. The combined tension is 1.7$σ$. Combining the Standard-Model rates with LHCb's measurements yields values for the Cabibbo-Kobayashi-Maskawa (CKM) matrix elements $|V_{td}|=7.45{(69)}\times10^{-3}$, $|V_{ts}|=35.7(1.5)\times10^{-3}$, and $|V_{td}/V_{ts}|=0.201{(20)}$, which are compatible with the values obtained from neutral $B_{(s)}$-meson oscillations and have competitive uncertainties. Alternatively, taking the CKM matrix elements from unitarity, we constrain new-physics contributions at the electroweak scale. The constraints on the Wilson coefficients ${\rm Re}(C_9)$ and ${\rm Re}(C_{10})$ from $B\toπμ^+μ^-$ and $B\to Kμ^+μ^-$ are competitive with those from $B\to K^* μ^+μ^-$, and display a 2.0$σ$ tension with the Standard Model. Our predictions for $B\to K(π)ν\barν$ and $B\toπτν$ are close to the current experimental limits.

hep-ph↗

arXiv2015

$|V_{ub}|$ from $B\toπ\ellν$ decays and (2+1)-flavor lattice QCD

We present a lattice-QCD calculation of the $B\toπ\ellν$ semileptonic form factors and a new determination of the CKM matrix element $|V_{ub}|$. We use the MILC asqtad 2+1-flavor lattice configurations at four lattice spacings and light-quark masses down to 1/20 of the physical strange-quark mass. We extrapolate the lattice form factors to the continuum using staggered chiral perturbation theory in the hard-pion and SU(2) limits. We employ a model-independent $z$ parameterization to extrapolate our lattice form factors from large-recoil momentum to the full kinematic range. We introduce a new functional method to propagate information from the chiral-continuum extrapolation to the $z$ expansion. We present our results together with a complete systematic error budget, including a covariance matrix to enable the combination of our form factors with other lattice-QCD and experimental results. To obtain $|V_{ub}|$, we simultaneously fit the experimental data for the $B\toπ\ellν$ differential decay rate obtained by the BaBar and Belle collaborations together with our lattice form-factor results. We find $|V_{ub}|=(3.72\pm 0.16)\times 10^{-3}$ where the error is from the combined fit to lattice plus experiments and includes all sources of uncertainty. Our form-factor results bring the QCD error on $|V_{ub}|$ to the same level as the experimental error. We also provide results for the $B\toπ\ellν$ vector and scalar form factors obtained from the combined lattice and experiment fit, which are more precisely-determined than from our lattice-QCD calculation alone. These results can be used in other phenomenological applications and to test other approaches to QCD.

hep-lat↗

arXiv2015

$B\toπ\ell\ell$ form factors for new-physics searches from lattice QCD

The rare decay $B\toπ\ell^+\ell^-$ arises from $b\to d$ flavor-changing neutral currents and could be sensitive to physics beyond the Standard Model. Here, we present the first $ab$-$initio$ QCD calculation of the $B\toπ$ tensor form factor $f_T$. Together with the vector and scalar form factors $f_+$ and $f_0$ from our companion work [J. A. Bailey $et~al.$, Phys. Rev. D 92, 014024 (2015)], these parameterize the hadronic contribution to $B\toπ$ semileptonic decays in any extension of the Standard Model. We obtain the total branching ratio ${\text{BR}}(B^+\toπ^+μ^+μ^-)=20.4(2.1)\times10^{-9}$ in the Standard Model, which is the most precise theoretical determination to date, and agrees with the recent measurement from the LHCb experiment [R. Aaij $et~al.$, JHEP 1212, 125 (2012)]. Note added: after this paper was submitted for publication, LHCb announced a new measurement of the differential decay rate for this process [T. Tekampe, talk at DPF 2015], which we now compare to the shape and normalization of the Standard-Model prediction.

hep-ph↗

arXiv2015

The $B \to D \ell ν$ form factors at nonzero recoil and $|V_{cb}|$ from $2+1$-flavor lattice QCD

We present the first unquenched lattice-QCD calculation of the hadronic form factors for the exclusive decay $\overline{B} \rightarrow D \ell \overlineν$ at nonzero recoil. We carry out numerical simulations on fourteen ensembles of gauge-field configurations generated with 2+1 flavors of asqtad-improved staggered sea quarks. The ensembles encompass a wide range of lattice spacings (approximately 0.045 to 0.12 fm) and ratios of light (up and down) to strange sea-quark masses ranging from 0.05 to 0.4. For the $b$ and $c$ valence quarks we use improved Wilson fermions with the Fermilab interpretation, while for the light valence quarks we use asqtad-improved staggered fermions. We extrapolate our results to the physical point using rooted staggered heavy-light meson chiral perturbation theory. We then parameterize the form factors and extend them to the full kinematic range using model-independent functions based on analyticity and unitarity. We present our final results for $f_+(q^2)$ and $f_0(q^2)$, including statistical and systematic errors, as coefficients of a series in the variable $z$ and the covariance matrix between these coefficients. We then fit the lattice form-factor data jointly with the experimentally measured differential decay rate from BaBar to determine the CKM matrix element, $|V_{cb}|=(39.6 \pm 1.7_{\rm QCD+exp} \pm 0.2_{\rm QED})\times 10^{-3}$. As a byproduct of the joint fit we obtain the form factors with improved precision at large recoil. Finally, we use them to update our calculation of the ratio $R(D)$ in the Standard Model, which yields $R(D) = 0.299(11)$.

hep-lat↗

arXiv2014

Dynamical QCD+QED simulation with staggered quarks

Electromagnetic effects play an important role in many phenomena such as isospin-symmetry breaking in the hadron spectrum and the hadronic contributions to g-2. We have generalized the MILC QCD code to include the electromagnetic field. In this work, we focus on simulations including charged sea quarks using the RHMC algorithm. We show details of the dynamical QCD+QED simulation algorithm with compact QED. We analyze the code performance and results for hadron-spectrum observables.

hep-lat↗

arXiv2014

K semileptonic form factor with HISQ fermions at the physical point

We present results for the form factor $f_+^{K π}(0)$, needed to extract the CKM matrix element $|V_{us}|$ from experimental data on semileptonic $K$ decays, on the HISQ $N_f=2+1+1$ MILC configurations. The HISQ action is also used for the valence sector. The data set used for our final result includes three different values of the lattice spacing and data at the physical light quark masses. We discuss the error budget and how this calculation improves on our previous determination of $f_+^{K π}(0)$ on the asqtad $N_f=2+1$ MILC configurations.

hep-lat↗

arXiv2014

Determination of $|V_{us}|$ from a lattice-QCD calculation of the $K\toπ\ellν$ semileptonic form factor with physical quark masses

We calculate the kaon semileptonic form factor $f_+(0)$ from lattice QCD, working, for the first time, at the physical light-quark masses. We use gauge configurations generated by the MILC collaboration with $N_f=2+1+1$ flavors of sea quarks, which incorporate the effects of dynamical charm quarks as well as those of up, down, and strange. We employ data at three lattice spacings to extrapolate to the continuum limit. Our result, $f_+(0) = 0.9704(32)$, where the error is the total statistical plus systematic uncertainty added in quadrature, is the most precise determination to date. Combining our result with the latest experimental measurements of $K$ semileptonic decays, one obtains the Cabibbo-Kobayashi-Maskawa matrix element $|V_{us}|=0.22290(74)(52)$, where the first error is from $f_+(0)$ and the second one is from experiment. In the first-row test of Cabibbo-Kobayashi-Maskawa unitarity, the error stemming from $|V_{us}|$ is now comparable to that from $|V_{ud}|$.

hep-ph↗

arXiv2014

Update of $|V_{cb}|$ from the $\bar{B}\to D^*\ell\barν$ form factor at zero recoil with three-flavor lattice QCD

We compute the zero-recoil form factor for the semileptonic decay $\bar{B}^0\to D^{*+}\ell^-\barν$ (and modes related by isospin and charge conjugation) using lattice QCD with three flavors of sea quarks. We use an improved staggered action for the light valence and sea quarks (the MILC \asqtad\ configurations), and the Fermilab action for the heavy quarks. Our calculations incorporate higher statistics, finer lattice spacings, and lighter quark masses than our 2008 work. As a byproduct of tuning the new data set, we obtain the $D_s$ and $B_s$ hyperfine splittings with few-MeV accuracy. For the zero-recoil form factor, we obtain $\mathcal{F}(1)=0.906(4)(12)$, where the first error is statistical and the second is the sum in quadrature of all systematic errors. With the latest HFAG average of experimental results and a cautious treatment of QED effects, we find $|V_{cb}| = (39.04 \pm 0.49_\text{expt} \pm 0.53_\text{QCD} \pm 0.19_\text{QED})\times10^{-3}$. The QCD error is now commensurate with the experimental error.

hep-lat↗

arXiv2013

Heavy-meson semileptonic decays for the Standard Model and beyond

We calculate the form factors for the semileptonic decays $B_s\to K\ellν$ and $B\to K\ell\ell$ with lattice QCD. We work at several lattice spacings and a range of light quark masses, using the MILC 2+1-flavor asqtad ensembles. We use the Fermilab method for the $b$ quark. We obtain chiral-continuum extrapolations for $E_K$ up to $\sim1.2$ GeV and then extend to the entire kinematic range with the model-independent $z$ expansion.

hep-lat↗