arXiv subjects

Tanmoy Bhattacharya

Publications and source records attributed to Tanmoy Bhattacharya.

At least 73 records · Page 4Linked to original sources

arXiv2020

Recent results of nucleon structure & matrix element calculations

A review of recent lattice calculations of nucleon structure and matrix elements of operators in nucleons is presented. It primarily covers developments in the calculation of the matrix elements of the scalar, tensor, pseudo-scalar, axial-vector and vector operators in the ground state of neutrons and protons in the isospin symmetric limit. Some preliminary calculations of the electric dipole moment, the gravitational moments and stress-energy distribution, and the magnetic polarizability are briefly described.

hep-lat↗

arXiv2020

Neutron Electric Dipole Moments with Clover Fermions

We present preliminary results for the contributions to the neutron EDM arising from the QCD $θ$-term, the Weinberg three-gluon and the quark chromo-EDM operators from our ongoing lattice calculations using clover valence quarks on the MILC HISQ lattices. We use the gradient-flow technique to smooth the lattices and renormalize the gluonic operators, and use the Schwinger source method to incorporate the quark chromo-EDM interactions in the quark propagator. For the QCD $θ$-term and the Weinberg three-gluon operator, we report results in the gradient-flow scheme from 8 ensembles at four lattice spacings and three pion masses, including 2 physical pion mass ensembles. For the quark chromo-EDM, unrenormalized results are presented at two lattice spacings, $a=0.12$ and $0.09$ fm, and two pion masses, $M_π= 310$ MeV and $220$ MeV.

hep-lat↗

arXiv2020

Leptonic decays of $B_{(s)}$ and $D_{(s)}$ using the OK action

We present recent progress in the lattice calculation of leptonic decay constants for $B_{(s)}$ and $D_{(s)}$ mesons using the Oktay-Kronfeld (OK) action for charm and bottom valence quarks, whose masses are tuned non-perturbatively. The calculations are done on 6 HISQ ensembles generated by the MILC collaboration with $N_f=2+1+1$ flavors. We also use the HISQ action for the light spectator quarks. Results are presented for the ratios $f_{B_s}/f_B$ and $f_{D_s}/f_D$, which reflect $SU(3)$ flavor symmetry breaking, and are independent of the renormalization constants of the axial currents.

hep-lat↗

arXiv2020

Nucleon charges and form factors using clover and HISQ ensembles

We present high statistics ($\mathcal{O}(2\times 10^5)$ measurements) preliminary results on (i) the isovector charges, $g^{u-d}_{A,S,T}$, and form factors, $G^{u-d}_E(Q^2)$, $G^{u-d}_M(Q^2)$, $G^{u-d}_A(Q^2)$, $\widetilde G^{u-d}_P(Q^2)$, $G^{u-d}_P(Q^2)$, on six 2+1-flavor Wilson-clover ensembles generated by the JLab/W&M/LANL/MIT collaboration with lattice parameters given in Table 1. Examples of the impact of using different estimates of the excited state spectra are given for the clover-on-clover data, and as discussed in [1], the biggest difference on including the lower energy (close to $Nπ$ and $Nππ$) states is in the axial channel. (ii) Flavor diagonal axial, tensor and scalar charges, $g^{u,d,s}_{A,S,T}$, are calculated with the clover-on-HISQ formulation using nine 2+1+1-flavor HISQ ensembles generated by the MILC collaboration [2] with lattice parameters given in Table 2. Once finished, the calculations of $g^{u,d,s}_{A,T}$ will update the results given in Refs.[3,4]. The estimates for $g^{u,d,s}_{S}$ and $σ_{Nπ}$ are new. Overall, a large part of the focus is on understanding the excited state contamination (ESC), and the results discussed provide a partial status report on developing defensible analyses strategies that include contributions of possible low-lying excited states to individual nucleon matrix elements.

hep-lat↗

arXiv2020

Nucleon Axial Form Factors from Clover Fermion on 2+1+1-flavor HISQ Lattice

The nucleon axial form factors -- axial $G_A$, induced pseudoscalar $\widetilde{G}_P$ and pseudoscalar $G_P$ -- have displayed large systematics in lattice QCD calculations. The major symptoms were the violation of the partially conserved axial current (PCAC) relation between the three form factors, and the underestimation of the induced pseudoscalar coupling $g_P^\ast$ and the axial charge radius $r_A$ compared to phenomenological estimates. The small $g_P^\ast$ was a consequence of the failure of the pion-pole dominance (PPD) hypothesis, especially at low $M_π^2$. The small charge radius $r_A$ and the underestimate of $g_A$ were related. The dominant systematic responsible is the lack of inclusion of low-energy ($N π$) states that are not manifest in the multiexponential fit to the nucleon two-point correlator. We show that this low-energy state can be determined from the three-point correlator $\langle N A_4 N \rangle $ with the insertion of the temporal component of the axial current $A_4$ within the nucleon state, ie, the strategy labeled $S_{A4}$ [1]. Including this low-energy state in fits to control excited-state contamination (ESC) gives results for $g_A$, $r_A$, and $g_P^\ast$ that are consistent with experimental/phenomenological values. However, the systematic uncertainties, especially in data at small $Q^2$, are now much larger.

hep-lat↗

arXiv2020

On Mixup Training: Improved Calibration and Predictive Uncertainty for Deep Neural Networks

Mixup~\cite{zhang2017mixup} is a recently proposed method for training deep neural networks where additional samples are generated during training by convexly combining random pairs of images and their associated labels. While simple to implement, it has been shown to be a surprisingly effective method of data augmentation for image classification: DNNs trained with mixup show noticeable gains in classification performance on a number of image classification benchmarks. In this work, we discuss a hitherto untouched aspect of mixup training -- the calibration and predictive uncertainty of models trained with mixup. We find that DNNs trained with mixup are significantly better calibrated -- i.e., the predicted softmax scores are much better indicators of the actual likelihood of a correct prediction -- than DNNs trained in the regular fashion. We conduct experiments on a number of image classification architectures and datasets -- including large-scale datasets like ImageNet -- and find this to be the case. Additionally, we find that merely mixing features does not result in the same calibration benefit and that the label smoothing in mixup training plays a significant role in improving calibration. Finally, we also observe that mixup-trained DNNs are less prone to over-confident predictions on out-of-distribution and random-noise data. We conclude that the typical overconfidence seen in neural networks, even on in-distribution data is likely a consequence of training with hard labels, suggesting that mixup be employed for classification tasks where predictive uncertainty is a significant concern.

stat.ML↗

arXiv2019

Axial Vector Form Factors from Lattice QCD that Satisfy the PCAC Relation

Previous lattice QCD calculations of axial vector and pseudoscalar form factors show significant deviation from the partially conserved axial current (PCAC) relation between them. Since the original correlation functions satisfy PCAC, the observed deviations from the operator identity cast doubt on whether all the systematics in the extraction of form factors from the correlation functions are under control. We identify the problematic systematic as a missed excited state, whose energy as a function of the momentum transfer squared, $Q^2$, is determined from the analysis of the 3-point functions themselves. Its mass is much smaller than those of the excited states previously considered and including it impacts the extraction of all the ground state matrix elements. The form factors extracted using these mass/energy gaps satisfy PCAC and other consistency conditions, and validate the pion-pole dominance hypothesis. We also show that the extraction of the axial charge $g_A$ is very sensitive to the value of the mass gaps of the excited states used and current lattice data do not provide an unambiguous determination of these, unlike the $Q^2 \neq 0$ case. To highlight the differences and improvement between the conventional versus the new analysis strategy, we present a comparison of results obtained on a physical pion mass ensemble at $a\approx 0.0871\,\mathrm{fm}$. With the new strategy, we find $g_A = 1.30(6)$. A very significant improvement over previous lattice results is found for the axial charge radius $r_A = 0.74(6)\,\mathrm{fm}$, extracted using the $z$-expansion to parameterize the $Q^2$ behavior of $G_A(Q^2)$, and $g_P^\ast = 8.06(44)$ obtained using the pion pole-dominance ansatz to fit the $Q^2$ behavior of the induced pseudoscalar form factor $\widetilde{G}_P(Q^2)$.

hep-lat↗

arXiv2019

Combating Label Noise in Deep Learning Using Abstention

We introduce a novel method to combat label noise when training deep neural networks for classification. We propose a loss function that permits abstention during training thereby allowing the DNN to abstain on confusing samples while continuing to learn and improve classification performance on the non-abstained samples. We show how such a deep abstaining classifier (DAC) can be used for robust learning in the presence of different types of label noise. In the case of structured or systematic label noise -- where noisy training labels or confusing examples are correlated with underlying features of the data-- training with abstention enables representation learning for features that are associated with unreliable labels. In the case of unstructured (arbitrary) label noise, abstention during training enables the DAC to be used as an effective data cleaner by identifying samples that are likely to have label noise. We provide analytical results on the loss function behavior that enable dynamic adaption of abstention rates based on learning progress during training. We demonstrate the utility of the deep abstaining classifier for various image classification tasks under different types of label noise; in the case of arbitrary label noise, we show significant improvements over previously published results on multiple image benchmarks. Source code is available at https://github.com/thulas/dac-label-noise

stat.ML↗

arXiv2019

Machine Learning Estimators for Lattice QCD Observables

A novel technique using machine learning (ML) to reduce the computational cost of evaluating lattice quantum chromodynamics (QCD) observables is presented. The ML is trained on a subset of background gauge field configurations, called the labeled set, to predict an observable $O$ from the values of correlated, but less compute-intensive, observables $\mathbf{X}$ calculated on the full sample. By using a second subset, also part of the labeled set, we estimate the bias in the result predicted by the trained ML algorithm. A reduction in the computational cost by about $7\%-38\%$ is demonstrated for two different lattice QCD calculations using the Boosted decision tree regression ML algorithm: (1) prediction of the nucleon three-point correlation functions that yield isovector charges from the two-point correlation functions, and (2) prediction of the phase acquired by the neutron mass when a small Charge-Parity (CP) violating interaction, the quark chromoelectric dipole moment interaction, is added to QCD, again from the two-point correlation functions calculated without CP violation.

hep-lat↗

arXiv2019

Nucleon Electromagnetic Form Factors in the Continuum Limit from 2+1+1-flavor Lattice QCD

Results are presented for the nucleon isovector electromagnetic form factors using 11 ensembles generated by the MILC collaboration using the 2+1+1-flavors HISQ action. They span 4 lattice spacings $a \sim$ 0.06, 0.09, 0.12 and 0.15~fm and 3 values of $M_π\sim 135, 225$ and 315 MeV. High-statistics estimates are used to perform a simultaneous extrapolation in the lattice spacing, lattice volume and light-quark masses. The $Q^2$ dependence over the range 0.05-1.4 ${\rm GeV}^2$ is investigated using both the $z$-expansion and the dipole form. Final $z$-expansion estimates for the isovector r.m.s. radius are $r_E = 0.769(27)(30)$ fm $r_M = 0.671(48)(76)$ fm and $μ^{p-n} = 3.939(86)(138)$ Bohr magneton. The first error is the combined uncertainty from the leading-order analysis, and the second is an estimate of the additional uncertainty due to using the leading order chiral-continuum-finite-volume fits. The dipole estimates, $r_E = 0.765(11)(8)$ fm, $r_M = 0.704(21)(29)$ fm and $μ^{p-n} = 3.975(84)(125)$, are consistent with those from the $z$-expansion but with smaller errors. Our analysis highlights three points. First, all data from the eleven ensembles and existing lattice data on, or close to, physical mass ensembles from other collaborations collapses more clearly onto a single curve when plotted versus $Q^2/M_N^2$ as compared to $Q^2$ with the scale set by quantities other than $M_N$. The difference between these two analyses is indicative of discretization errors, some of which presumably cancel when the data are plotted versus $Q^2/M_N^2$. Second, the size of the remaining deviation of this common curve from the Kelly curve is small and can be accounted for by statistical and possible systematic uncertainties. Third, to improve lattice estimates, high statistics data for $Q^2 < 0.1$ ${\rm GeV}^2$ are needed.

hep-lat↗

arXiv2019

The Role of Lattice QCD in Searches for Violations of Fundamental Symmetries and Signals for New Physics

This document is one of a series of whitepapers from the USQCD collaboration. Here, we discuss opportunities for Lattice Quantum Chromodynamics (LQCD) in the research frontier in fundamental symmetries and signals for new physics. LQCD, in synergy with effective field theories and nuclear many-body studies, provides theoretical support to ongoing and planned experimental programs in searches for electric dipole moments of the nucleon, nuclei and atoms, decay of the proton, $n$-$\overline{n}$ oscillations, neutrinoless double-$β$ decay of a nucleus, conversion of muon to electron, precision measurements of weak decays of the nucleon and of nuclei, precision isotope-shift spectroscopy, as well as direct dark matter detection experiments using nuclear targets. This whitepaper details the objectives of the LQCD program in the area of Fundamental Symmetries within the USQCD collaboration, identifies priorities that can be addressed within the next five years, and elaborates on the areas that will likely demand a high degree of innovation in both numerical and analytical frontiers of the LQCD research.

hep-lat↗

arXiv2018

Updates on Nucleon Form Factors from Clover-on-HISQ Lattice Formulation

Updates on results for the electric, magnetic and axial vector form factors are presented. The data analyzed cover high statistics measurements on 11 ensembles generated with 2+1+1 flavors of HISQ fermions by the MILC collaboration. The data cover the range $0.057\;\mathrm{fm} < a < 0.15\;\mathrm{fm}$ in lattice spacing, $135\;\mathrm{MeV} < M_π< 320\;\mathrm{MeV}$ in the pion mass and $3.3 < M_πL < 5.5 $ in the lattice size. Fits to control excited-state contamination use up to 3-states in the spectral decomposition of the 3-point functions. The simultaneous chiral-continuum-finite-volume fits include leading order corrections in each of the three variables.

hep-lat↗

arXiv2018

Update on $B\to D^\ast \ell ν$ form factor at zero-recoil using the Oktay-Kronfeld action

We present an update on the calculation of $\bar{B}\to D^\ast \ell \barν$ semileptonic form factor at zero recoil using the Oktay-Kronfeld bottom and charm quarks on $N_f=2+1+1$ flavor HISQ ensembles generated by the MILC collaboration. Preliminary results are given for two ensembles with $a\approx 0.12$ and $0.09$ fm and $M_π\approx 310$ MeV. Calculations have been done with a number of valence quark masses, and the dependence of the form factor on them is investigated on the $a\approx 0.12$ fm ensemble. The excited state is controlled by using multistate fits to the three-point correlators measured at 4--6 source-sink separations.

hep-lat↗

arXiv2018

Neutron Electric Dipole Moment from Beyond the Standard Model

We present an update on our calculations of the matrix elements of the CP violating quark and gluon chromo-EDM operators, as well as the operators these mix with, such as the QCD Theta-term. Their contribution to the neutron EDM is obtained by extrapolating the $F_3$ form factor of a vector current to zero momentum transfer. The calculation is being done using valence Wilson-clover quarks on HISQ background configurations generated by the MILC collaboration.

hep-lat↗

arXiv2018

Isovector and flavor diagonal charges of the nucleon from 2+1+1 flavor QCD

We present high-statistics results for the isovector and flavor diagonal charges of the proton using 11 ensembles of 2+1+1 flavor HISQ fermions. In the isospin symmetric limit, results for the neutron are given by the $u \leftrightarrow d$ interchange. A chiral-continuum fit with leading order corrections was made to extract the connected and disconnected contributions in the continuum limit and at $M_π=135$ MeV. All results are given in the $\overline{MS}$ scheme at 2 GeV. The isovector charges, $g_A^{u-d} = 1.218(25)(30)$, $g_S^{u-d} = 1.022(80)(60) $ and $g_T^{u-d} = 0.989(32)(10)$, are used to obtain low-energy constraints on novel scalar and tensor interactions, $ε_{S}$ and $ε_{T}$, at the TeV scale. The flavor diagonal axial charges are: $g_A^u \equiv Δu \equiv \langle 1 \rangle_{Δu^+} = 0.777(25)(30)$, $g_A^d \equiv Δd \equiv \langle 1 \rangle_{Δd^+} = -0.438(18)(30)$, and $g_A^s \equiv Δs \equiv \langle 1 \rangle_{Δs^+} = -0.053(8)$. Their sum gives the total quark contribution to the proton spin, $\sum_{q=u,d,s} (\frac{1}{2} Δq) = 0.143(31)(36)$. This result is in good agreement with the recent COMPASS analysis $0.13 < \frac{1}{2} ΔΣ< 0.18$. Implications of results for the flavor diagonal tensor charges, $g_T^u = 0.784(28)(10)$, $g_T^d = -0.204(11)(10)$ and $g_T^s = -0.0027(16)$ for constraining the quark electric dipole moments and their contributions to the neutron electric dipole moment are discussed. These flavor diagonal charges also give the strength of the interaction of dark matter with nucleons via axial and tensor mediators.

hep-lat↗

arXiv2018

Isovector Charges of the Nucleon from 2+1+1-flavor Lattice QCD

We present high statistics results for the isovector charges $g^{u-d}_A$, $g^{u-d}_S$ and $g^{u-d}_T$ of the nucleon. Calculations were carried out on eleven ensembles of gauge configurations generated by the MILC collaboration using highly improved staggered quarks (HISQ) action with 2+1+1 dynamical flavors. These ensembles span four lattice spacings $a \approx$ 0.06, 0.09, 0.12 and 0.15 fm and light-quark masses corresponding to $M_π\approx$ 135, 225 and 315 MeV. Excited-state contamination in the nucleon 3-point correlation functions is controlled by including up to three-states in the spectral decomposition. Remaining systematic uncertainties associated with lattice discretization, lattice volume and light-quark masses are controlled using a simultaneous fit in these three variables. Our final estimates of the isovector charges in the $\overline{\text{MS}}$ scheme at 2 GeV are $g_A^{u-d} = 1.218(25)(30)$, $g_S^{u-d} = 1.022(80)(60) $ and $g_T^{u-d} = 0.989(32)(10)$. The first error includes statistical and all systematic uncertainties except that due to the extrapolation ansatz, which is given by the second error estimate. We provide a detailed comparison with the recent result of $g_A^{u-d} = 1.271(13)$ by the CalLat collaboration and argue that our error estimate is more realistic. Combining our estimate for $g_S^{u-d}$ with the difference of light quarks masses $(m_d-m_u)^{\rm QCD}=2.572(66)$ MeV given by the MILC/Fermilab/TUMQCD collaboration for 2+1+1-flavor theory, we obtain $(M_N-M_P)^{\rm QCD} = 2.63(27)$ MeV. We update the low-energy constraints on novel scalar and tensor interactions, $ε_{S}$ and $ε_{T}$, at the TeV scale by combining our new estimates for $g^{u-d}_S$ and $g^{u-d}_T$ with precision low-energy nuclear experiments, and find them comparable to those from the ATLAS and the CMS experiments at the LHC.

hep-lat↗

arXiv2018

Flavor diagonal tensor charges of the nucleon from 2+1+1 flavor lattice QCD

We present state-of-the-art results for the matrix elements of flavor diagonal tensor operators within the nucleon state. The calculation of the dominant connected contribution is done using eleven ensembles of gauge configurations generated by the MILC Collaboration using the highly improved staggered quark (HISQ) action with 2+1+1 dynamical flavors. The calculation of the disconnected contributions is done using seven (six) ensembles for the strange (light) quarks. These high-statistics simulations allowed us to address various systematic uncertainties. A simultaneous fit in the lattice spacing and the light-quark mass is used to extract the tensor charges in the continuum limit and at $M_π=135$ MeV. Results for the proton in the $\overline{MS}$ scheme at 2~GeV are: $g_T^u = 0.784(28)(10)$, $g_T^d = -0.204(11)(10)$ and $g_T^s = -0.0027(16)$. Implications of these results for constraining the quark electric dipole moments and their contributions to the neutron electric dipole moment are discussed.

hep-lat↗

arXiv2018

Quark contribution to the proton spin from 2+1+1-flavor lattice QCD

We present the first chiral-continuum extrapolated up, down and strange quark spin contribution to the proton spin using lattice QCD. For the connected contributions, we use eleven ensembles of 2+1+1-flavor of Highly Improved Staggered Quarks (HISQ) generated by the MILC Collaboration. They cover four lattice spacings $a \approx \{0.15,0.12,0.09,0.06\}$ fm and three pion masses, $M_π\approx \{315,220,135\}$ MeV, of which two are at the physical pion mass. The disconnected strange calculations are done on seven of these ensembles, covering the four lattice spacings but only one with the physical pion mass. The disconnected light quark calculation was done on six ensembles at two values of $M_π\approx \{315,220\}$ MeV. High-statistics estimates on each ensemble for all three quantities allow us to quantify systematic uncertainties and perform a simultaneous chiral-continuum extrapolation in the lattice spacing and the light-quark mass. Our final results are $Δu \equiv \langle 1 \rangle_{Δu^+} = 0.777(25)(30)$, $Δd \equiv \langle 1 \rangle_{Δd^+} = -0.438(18)(30)$, and $Δs \equiv \langle 1 \rangle_{Δs^+} = -0.053(8)$, adding up to a total quark contribution to proton spin of $\sum_{q=u,d,s} (\frac{1}{2} Δq) = 0.143(31)(36)$. The second error is the systematic uncertainty associated with the chiral-continuum extrapolation. These results are obtained without model assumptions and are in good agreement with the recent COMPASS analysis $0.13 < \frac{1}{2} ΔΣ< 0.18$, and with the $Δq$ obtained from various global analyses of polarized beam or target data.

hep-lat↗