dedup-isc-ft-v107-score
float64 0.32
1
| uid
stringlengths 32
32
| text
stringlengths 0
32.5k
| paper_id
stringlengths 1
14
| original_image_filename
stringlengths 5
222
|
|---|---|---|---|---|
0.491762 | f5569b8e52334eb2a3108ded7a409255 | Visualization of the convolution operators conducted on dense and sparse tensors. Blue and green denote the input and out feature maps, respectively. Gray indicates the convolutional kernel. It will densely slide over the entire space. On a sparse tensor, the convolution is instead only conducted on a few specified locations. | 2012.13846 | dense_sparse.jpg |
0.458174 | 4bae2848d1114a86a4cc8047cf52080f | GPipe <cit.> | 2012.13846 | gpipe.jpg |
0.408897 | 6f879d4d26fb4cebbc7adf26532ce206 | Performance evaluation between naive data parallelism (DP), pipeline model parallelism (MP) and heterogeneous aware pipeline model partition (HETE-MP). The speedups of MP and HETE-MP compared to DP are presented. The evaluation is conducted with two servers. All models are trained at the voxel resolution of size 50 × 50 × 50 and a batch size of 64. | 2012.13846 | hetemp_vs_mp.jpg |
0.426989 | 6162f85d238d4e4b8b667ea38ffbf4fa | Naive pipeline | 2012.13846 | naive.jpg |
0.387463 | ee198da0d5a642079fb6e3b9565848d3 | An overview of the heterogeneous-aware pipeline model partition algorithm for sparse-tensor-based computation. Given an input sparse DNN, we profile the computation time, communication time, and parameter sizes for layers, from which we compute the accumulated layer costs ratio (ALCR), with increasing layer IDs. Our method measures across multiple machines with GPU processors that are heterogeneous. We then aggregate all profiles and use a dynamic programming algorithm to obtain feasible model partition that divides the whole model into sub-models and distributes to processors. For simplicity, we omit drawings of all intermediate layers (e.g., batch normalization, pooling, or activation layers) between SparseConv blocks of the Sparse DNN. | 2012.13846 | partition.jpg |
0.402077 | d3fbaf9da9334fbfb6d1e7046582ce37 | Pipedream <cit.> | 2012.13846 | pipedream.jpg |
0.427025 | 0f192355368e411d8046d133ea4842fa | Hybrid pipeline model parallelism with 3 processors and 2 stages. Stage 1 takes twice of the time units than stage 2 in the forward or backward pass. To sustain roughly the same throughput of stage 2 (on proc 3), stage 1 is replicated on two processors (proc 1 and 2). | 2012.13846 | replicated_pipedream_2.jpg |
0.467749 | d02402ef4f9d4e5d8cc87588da9ee4ac | Compared to Dense DNN, Sparse DNN is significantly less memory intensive. Sparse DNN can support a resolution of 200 × 200 × 200 while Dense DNN is limited to 50 × 50 × 50. This opens up the possibility of solving problems at larger scale. Both models can achieve higher accuracy with a higher resolution. | 2012.13846 | sparseDNNvsDenseDNN.jpg |
0.429199 | 4d018f5e6fef4975af9e7719ea0cf73b | Denoising of tremor waveforms with neural network attribution. Raw waveforms (A, left panel) are turned into spectrograms (A, second panel) and fed to a classic CNN. We rely on a standard network (three convolution+max pool and two dense layers). The architecture of the network is shown in (B). From these spectrograms, the CNN is tasked to classify tremor from non-tremor. Once the model is trained (with tremors identified from the PNSN catalog), we rely on Integrated Gradients <cit.> attribution to interpret its results. (A, third panel) shows the result of the positive attribution for the same waveform; the blue areas correspond to the parts of the spectrogram that carried core tremor information according to the attribution analysis. We can then select exclusively these areas to inverse-Fourier transform, to get a clean tremor signal (A, right panel). These cleaned waveforms correspond to the core tremor signals, as seen by our model, and have clear structure compared to the original waveforms; our goal here is to rely on these clean waveforms to locate tremors. | 2012.13847 | Figure1_v6.jpg |
0.407978 | 7fe4e83c710343c7910dcbe9a78c8e02 | Application to an array of seismometers. (A) Raw waveforms bandpassed between 1 and 8 Hz. (B) Result of the neural network attribution analysis to denoise the waveforms (in blue). The denoising procedure is performed on each station independently. Clean move-out patterns can be seen across the network, which can be compared to shear wave theoretical travel times (red line). The move-outs match theoretical wave speed to first order. | 2012.13847 | Figure2_v16a.jpg |
0.437666 | 113056cd6d384f4fa8fe72a607e7d7e7 | Example tremor location. We compute the envelope of the cleaned signals; these envelopes are then sliced around each identified STA/LTA pick, and cross-correlated for all station pairs. This figure shows the stacked correlation plotted over the lat/lon grid for one tremor. If a number of criteria are met (see below), the maximum of the stacked correlation function is returned as tremor location. | 2012.13847 | Figure2_v16b.jpg |
0.410184 | 761aab357a13423fa43a40bf22883279 | Area analyzed and seismic array used. Left: Map of the Southern Vancouver Island, Canada, analyzed in this study. Right: seismic array. The black triangles correspond to seismic stations used to locate tremors. | 2012.13847 | Figure3_v10a.jpg |
0.428636 | c3172dd0a51a4bf7a7aa2131a54acc40 | Example location. Tremor locations (in red) for three consecutive days with consistent, local tremor patches, during the 2018 slow slip event. | 2012.13847 | Figure3_v10b.jpg |
0.431836 | af86e2d1faf94ccf8f98e9ac654e628f | Comparison with the PNSN catalog. (A) Number of tremors detected by our algorithm (dark blue), compared to the number of detected tremors in the PNSN catalog (light blue). The timing of detections is consistent between both catalogs, with a high number of tremors detected between June 18th and July 11th. (B) Distance between daily tremor patches detected by our algorithm and by the PNSN catalog. For most days, the patches have similar locations (with a median distance of 4.42km). The two days where the distance is high correspond to patches in the PNSN catalog that are on the border or outside of our seismic array. | 2012.13847 | Figure4_v7.jpg |
0.53003 | 2d93dfe47c6348a582a6633e8fdefe71 | Isosurfaces in 𝒫(3,ℂ) centered at I with unit radius, e.g., {X∈𝒫(3,ℂ)|δ_F(I,X)=1 } where λ_1, λ_2 and λ_3 are eigenvalues of X | 2012.13861 | isotropy.jpg |
0.427633 | ca8118da700a4a5ebf5af5ce7f908d4c | The scheme of signal detection | 2012.13861 | signal_detection_scheme.jpg |
0.432233 | 9e3af513c19d412d80ac11b31bc233de | The basic method of solving mixture problems using the finite volume and material point method (FV-MPM). a) Define the mixture and initial configuration including densities, porosities, stresses. b) Define the initial solid and fluid phase continuum bodies (ℬ_s^0 and ℬ_f^0, respectively). c) Break the continuum bodies into piecewise-defined chunks of material: use material points to track the solid phase and finite volumes to track the fluid (𝒫_p^0 is the body part associated with the pth material point). d) Define finite element nodal basis functions and simulation domain. e) Solve the discrete equations of motion for the mixture using FV-MPM. | 2012.13862 | Fig1.jpg |
0.42808 | 48c1b53a6ab947ce9b8d2a51f9e0587a | Numerical simulation of sand pile erosion by gust of wind in three dimensions: a) a field of crescent-shaped barchan sand dunes in the desert between Chimbote and Casma on the coast of Peru (image source: <cit.>); b) schematic diagram of characteristic barchan dune shape (see <cit.>); b) ϕ = 0.3 surface contours for flow solutions at 0 s, 0.5 s, and 2 s. Black contour lines in b) are marked at 4 mm vertical increments, and the tick marks along the boundary denote 2.5 cm increments. | 2012.13862 | Fig10.jpg |
0.394239 | a540f4fab3934a508ac102f368ddcd8a | Numerical simulation of rocket exhaust impinging on Martian soil: a) schematic diagram of numerical test showing boundary conditions, dimensions, and initial configuration of soil; b) finite volume (FV) and finite element (FE) tetrahedral grids along with initial material point (MP) distribution. Simulation is performed on a 60^∘ wedge of a cylindrical domain with symmetric boundary conditions applied to reflected boundaries of the domain. Martian gravity (3.7 m/s^2) is used for g. | 2012.13862 | Fig11.jpg |
0.455047 | 3a8763a64aa240359cf89d2d2b91e9ee | Numerical simulation of rocket exhaust impinging on Martian soil: a) flow solution at 0.0 s, b) flow solution at 0.1 s, c) flow solution at 0.2 s, and d) flow solution at 0.3 s. Left column: renderings of Apollo Lunar Module Descent Engine (LMDE) and material point tracers colored by the local volume fraction ϕ. Right column: slice of simulated domain extending radially outward from the center-line highlighting the overlapping fluid and granular domains; the fluid domain is colored by the local mach number M = v_f/√(γ_r R ϑ_f). | 2012.13862 | Fig12.jpg |
0.456603 | dafab96db9d7468c924c2eacd6ee590c | An example of a two-phase finite volume and material point method (FV-MPM) simulation highlighting the different solution spaces associated with each phase of the mixture. a) A snapshot from the simulation of a sand–air mixture with wind blowing from left to right; the granular phase is represented by black material tracers and the fluid phase is colored according to the local effective temperature, ϑ_f. b) A region of this simulation showing the material point characteristic functions, U_p(x,t), visualized using their GIMP representations (squares; see <cit.>); the nodal basis functions, 𝒩_i(x), and associated nodes (black circles); and the finite volume subdomains, Ω_α (triangles). Here, we chose nodal basis functions and finite volume subdomains that are defined on the same triangular grid. | 2012.13862 | Fig2.jpg |
0.432078 | 0b8cd5c623b949f8a0d20d1f8f7d7d64 | Schematic showing a representative finite element grid and material point discretization. The continuum body and the pth material subdomain are shown in their a) initial configurations (ℬ^0 and 𝒫_p^0) and b) configurations at time t (ℬ^t and 𝒫_p^t). The pth material point characteristic function, U_p(x,t), is determined by the position of the pth material subdomain. | 2012.13862 | Fig3.jpg |
0.406885 | e97a070bfa4044d8b78731ad6ec60fe6 | Simulation snapshots for collapse of a submerged column of glass beads. The images in each column represent the flow solution at 0 s, 2 s, and 8 s, respectively, and are found using the simulation frameworks and initial conditions labeled at the left of the figure. The first column images indicate the relevant boundary conditions applied in each case (* indicates the application of a zero pressure BC at the upper surface), and the arrows in the second and third columns indicate the strength and direction of the fluid velocity field, v_f. The length scale of these arrows is 0.25 s × v_f. The material points associated with the granular phase are colored by the local granular volume fraction, ϕ, and the material points associated with the fluid phase are colored gray. Inset a) highlights the regular Cartesian grid used in the MPM and FV-MPM simulations as well as markers for the initial material point discretization. | 2012.13862 | Fig7.jpg |
0.464967 | 5f91acae9e324784881c9f2db3742a77 | Numerical simulation sand pile erosion by air in two dimensions: a) schematic diagram of numerical test showing boundary conditions and initial dimensions of sand pile; b) finite volume (FV) and finite element (FE) triangular grids along with initial material point (MP) distribution (representing the sand pile at 0.0 s); c) mixed flow solution at 1.5 s with air colored by effective temperature, ϑ_f, and material points (sand) colored by the effective shear strain measure, γ̅^p; d) mixed flow solution at 3.0 s. Tick marks in c) and d) mark 2 cm intervals. | 2012.13862 | Fig8.jpg |
0.393614 | a1833406606445aeb508afa7759967c5 | Numerical simulation of sand pile erosion by gust of wind in three dimensions: a) schematic diagram of numerical test showing boundary conditions and initial dimensions of sand pile; b) finite volume (FV) and finite element (FE) Cartesian grids along with initial material point (MP) distribution (representing the sand pile at 0.0 s); c) mixed flow solution at 0.5 s with air streamlines (originating along center-line of inlet) colored by velocity magnitude, ||v_f||, and material points (sand) colored by the effective shear strain measure, γ̅^p; d) mixed flow solution at 2.0 s, after wind gust has stopped. | 2012.13862 | Fig9.jpg |
0.426611 | 8041b8d01e0f4826963f80f8ca8b70a1 | Pictorial description of the representative volume Ω and boundary ∂Ω, the decomposition of the domain into fluid and solid volumes, and the homogenization of the two phases. Here Ω = Ω_s ∪Ω_f and ∂Ω = ∂Ω_s ∪∂Ω_f with ∂Ω^* defining the interior surface separating the solid and fluid domains. | 2012.13862 | FigA1.jpg |
0.504618 | 2893af10085947f7baf8e25c18f090dc | Pictorial representation of a one-dimensional, steady-state flow solution. The blue solid line shows a representative porosity field, n, with fluctuations that are not resolved on the finite volume grid, and the blue `×'s denote the cell-wise average values. The red solid line shows a steady-state true momentum field (ρ_f v_f) solution that is consistent with the porosity field below, and the red `□'s denote the cell-wise average values. The red long-dashed line shows the reconstructed true momentum field found using the gradient approximations from (<ref>). The red short-dashed line shows the reconstructed true momentum field found using the corrected gradient approximations from (<ref>). | 2012.13862 | FigB1.jpg |
0.476579 | 92492bb992684b1ea27e79ddd160733b | Method of manufactured solutions (MMS) results. a) Simulated domain and boundary conditions. b) FV, FE, and MP discretizations and associated characteristic lengths h_α, h_i, and h_p. c) L_2-error vs. finite volume length scale, h_α; h_i and h_p are constant. d) L_2-error vs. finite element length scale, h_i; h_α and h_p/h_i are constant. e) L_2-error vs. material point refinement measure, h_p/h_i; h_α and h_i are constant. | 2012.13862 | FigC1.jpg |
0.435419 | 97fbadb3de024ba0952c820d91975314 | Comparison of the true coefficients (solid) with the coefficients from the POD-GP ROM (dashed-dot) and from the POD-GP ROMs augmented with the three different learned neural closure models at the end of training (dashed). For each neural closure, the training period is from t=0 to 2.0, the validation period from t=2.0 to 4.0, and the future prediction period from t=4.0 to 6.0. Top-left: neural ODEs with no-delays (nODE); Top-right: neural DDEs with discrete-delays (Discrete-nDDE); Bottom-left: neural DDEs with distributed-delays (Distributed-nDDE). Bottom-right: Evolution of root-mean-squared-error (RMSE(t) = √(1/3∑_k=1^3 |a_k^pred(t) - a_k^true(t)|^2)) of coefficients from the four different ROMs. These results correspond to the architectures detailed in Table <ref>. | 2012.13869 | Exp1_combined_results.jpg |
0.447783 | 7a5b4b054e3c4aeb8f4a07c314cb3e22 | Geometric interpretation of the closure for reduced-order-models (ROMs). u ([0.8mm]0.5cm1pt): Solution to the full-order-model (FOM); VV^Tu ([0.8mm]0.5cm0.75pt): Projection of u on the subspace V; u^ROM ([0.8mm]0.5cm0.75pt): Solution to the proper-orthogonal-decomposition Galerkin-projection (POD-GP) ROM; and, u^ROM+C ([0.8mm]0.5cm0.75pt): Solution to POD-GP ROM with closure. Adapted from <cit.>. | 2012.13869 | POD-GP-Geom_Interpretation_pl.jpg |
0.431441 | bbde18416abe45ccaa5ec2fdbb89a6b7 | Visual examples of the augmented images by histogram matching. | 2012.13871 | HM.jpg |
0.400852 | fd37c777b9014b0c81bc1e45c7dbb392 | Visual examples of worse cases in vendor A and B. | 2012.13871 | MMs_worse_case_AB.jpg |
0.414229 | 3292e5aa67e748f1b4e11fdd140eeecb | Visual examples of worse cases in vendor C and D. | 2012.13871 | MMs_worse_case_CD.jpg |
0.47214 | e2796a97874746d18533e15a67d2e2c8 | Visual examples of segmentation results. Base and HM stand for the baseline dataset and the histogram matching augmented dataset. | 2012.13871 | SegRes.jpg |
0.521408 | fba267f8f3264043acdd4d0dab1c4b06 | The Mach-Zehender interferometer (See text for details). | 2012.13875 | fig1.jpg |
0.439196 | 5fc5293e1a79442a9d8cc55117e1b807 | Quantum values of four LG expressions given in Eqs. (<ref>) - (<ref>) are plotted against β. Curves demonstrate that one of the four LGIs will be violated for any given value of β, except for β= 0 or ± 1. | 2012.13875 | lgiplots.jpg |
0.486479 | 4db07935d4a24f82b3b3dcad4f5f93cb | (Color-online) Critical values of unsharpmness parameter λ_k required for violating the preparation non-contextual bound are plotted k^th Bob in the case of n=5. Here, blue and red dots denote the critical values corresponding to two-qubit and qubit system respectively. | 2012.13876 | n5sharing.jpg |
0.479593 | b3be566eee05481fa70b71872ee8e2ac | (Color-online) Critical values of unsharpmness parameter λ_k required for violating the preparation non-contextual bound are plotted k^th Bob in the case of n=6. Here, blue, red and green dots denote the critical values corresponding to three-qubit, two-qubit and qubit system respectively. | 2012.13876 | n6sharing.jpg |
0.395517 | a4d88829b0b54cedbecef591952fa6c9 | The distribution of blazars in luminosity-redshift space according to the luminosity function deduced in <cit.>. The dashed line separates the region into resolved and unresolved sources in Fermi-LAT survey. | 2012.13877 | lz_sim.jpg |
0.487713 | 3617170171544fa69d786b42188e155e | The schematic of semi-device independent prepare-measure communication game implementing Mermin's proof of Kochen-Specker contextuality. Alice prepares the particle in quantum state ρ_ab by sequential measurements of two commuting observables upon receiving the inputs (x,y ∈{1,2,3}) and sends it to Charlie who performs the measurement according to his inputs z∈{1,2,3}. | 2012.13878 | fig1.jpg |
0.444502 | 5f1cdde8667f44faac4a508d9bc63935 | Magic-star proof diagram | 2012.13878 | star.jpg |
0.419237 | 3461f0d974c14f3c97db1fddea2a871e | Deviation of the global model parameters (Adult). | 2012.13891 | 9_angle.jpg |
0.421929 | 42702ef385fb40daa62635826f1c25cc | NMI results compared with competing methods under different number of selected features | 2012.13892 | NMI.jpg |
0.42212 | 7e78b7b4b14741e99189999eb6c3bcbe | ACC results compared with competing methods under different number of selected features | 2012.13892 | acc.jpg |
0.519039 | e1e168fba2e24a5a8dd699ad8ca2c3da | Convergence curves of AGUFS on Umist, COIL20 and JAFFE. | 2012.13892 | convergence1.jpg |
0.416769 | 22248bbb160247fda18f8a021ef5977c | ACC with varying parameters α and λ | 2012.13892 | sensitivity-ACC.jpg |
0.414476 | 61be6099534044628d9d6aa0afcdf58e | NMI with varying parameters α and λ | 2012.13892 | sensitivity-NMI.jpg |
0.437506 | 6fce2c26212847878c7db4ebc6d701f5 | The exponent, ν_0, for the unperturbed conformation as a function of t for the comb polymer illustrated in Fig. <ref>. | 2012.13893 | CombExponent.jpg |
0.516 | 22cd57659e6b4bf08bb80a6870605529 | Extended triangular model. | 2012.13893 | GeneralizedTriangular-A.jpg |
0.478042 | 65cb8000a00f444fb35f914544c4ee4f | The first (z=1), the second (z=2), and the third (z=3) nested structures. In z=3, g-1 comb polymers branch off from the monomers on the new backbone (red solid-line). Such nesting may be repeated successively, which results in the dendrimer structure with f=3. | 2012.13893 | NestedComb.jpg |
0.379773 | 0182926d6e894e939f486bd72392b83b | The radial segment distributions around the center of gravity for the regular comb polymers. The red solid-line is the PDF of the comb polymer having (a) g=100 and n=10 (N=2190) and the blue solid-line is that for (b) g=1000 and n=10 (N=10990); the dotted lines are the corresponding Gaussian PDF's having the same ⟨ s_N^2⟩_0's. | 2012.13893 | RegularCombSimul.jpg |
0.496732 | d6f77c3be9564e0c9cb115094e9afbbb | A triangular polymer. The main backbone is indicated by the red bold-line (red). The end monomer on each branch has the same generation number, g, as the main backbone. | 2012.13893 | TriangularIllustB.jpg |
0.426152 | abd7136e13474edca39bf306f7aa63ad | The probability distribution of segments around the center of gravity for the triangular polymer illustrated in Fig. <ref>. The solid lines are drawn according to Eq. (<ref>) for (a) g=10 (N=46) and (b) for g=100 (N=4951). The dotted lines represent the Gaussian functions having the same mean radii of gyration, ⟨ s_N^2⟩_0. | 2012.13893 | TriangularSimul.jpg |
0.49854 | 052a991ffc7149ca993d381347635717 | The regular comb polymer having the same length, n, of side chains. The red bold-line is the main backbone, and the black lines are side chains. | 2012.13893 | regularcombB.jpg |
0.462007 | 34452a5ade22435e88dc986aee946dc6 | Temperature dependence of entropy of mixing as a function of x for Bi_xAl_1-x liquid binary alloys ( after Fysol et al. [25]). | 2012.13897 | AlBi_delent_T.jpg |
0.446135 | 73b23ab86c9148c1b4bcf9a6b971ad75 | Temperature dependence of enthalpy of mixing as a function of x for Bi_xAl_1-x liquid binary alloys (after Fysol et al. [25]). | 2012.13897 | AlBi_delh_T.jpg |
0.430577 | 5246859c52544cf5a0c4cf253752fc20 | Temperature dependence of energy of mixing as a function of x for Cu_xCo_1-x liquid binary alloys (after Faruk et al. [22]). | 2012.13897 | CuCo_delf_T.jpg |
0.453454 | f6913dbed164481e9751bcee810d8ef1 | The proposed LCCNet takes the RGB and the projected depth image as inputs to predict the extrinsic parameters between the LiDAR and the camera. The point clouds are re-projected by the predicted extrinsic parameters. The re-projected depth image and the RGB image will be the subsequent inputs of the LCCNet. This process is called iterative refinement. After five times iterative refinements, we obtain the final extrinsic parameters estimation. | 2012.13901 | 0.jpg |
0.38236 | 57b0e723402f4eae8ef8cd4343a86751 | The workflow of our proposed method for the estimation of the extrinsic calibration parameters between 3D LiDAR and 2D camera. The network takes an RGB image from a calibrated camera and a projected sparse depth image from a mis-calibrated LiDAR as input. The output of the network is a 6-DoF rigid-body transformation T_pred that represents the deviation between the initial extrinsic T_init and the ground truth extrinsic T_LC. As shown, we notice that the 3D structure highlighted using red rectangles fails to project to their 2D counterparts with the mis-calibrated depth image. When using the predicted transformation T_pred to revise the T_init, we can reconstruct a more consistent and accurate 3D scene structure. | 2012.13901 | 1.jpg |
0.476117 | 3cc227dc9e7d4d348afdf8c1deefd328 | Confidence interval at level 95% | 2012.13904 | CIs.jpg |
0.382597 | 6e04dbae31eb43bca939e6af17e1dee7 | a=5, x=0:0.1:10 | 2012.13904 | a5.jpg |
0.404955 | 25518cc161d74beeada02bd486067830 | We set x=0, a̅=1:0.1:10. As we can see, 1/N∑_k=1^N∑_i=1^⌊ T_t^k/h⌋hg(X_t_i^k^k)/(a̅)^2/3 is almost a constant, which is consistent with our calculation in (4.2) | 2012.13904 | integral_over_a.jpg |
0.369395 | 91fd93cce02e46568153b840330a41ac | x=0, a̅=1:0.1:10 | 2012.13904 | x0g_neq_0.jpg |
0.483652 | 3b276bd26d7648f793ce51d4cae91851 | Correlation matrix showing the correlation coefficients among the basic features. | 2012.13905 | correlation_matrix.jpg |
0.492768 | 7c1d99125abd4312a249356db7ff76c6 | Experimental framework scheme. | 2012.13905 | process.jpg |
0.433061 | 626a1a249b09490f8304ef555e9eb46f | Feature importance of cumulative features computed according to Gini index, for the Decision Tree model (max depth = 10). | 2012.13905 | significance.jpg |
0.412581 | d05e9dc0d4f1403280fe3068a6543a03 | Distribution of the review ratings in the dataset used in the experiments. | 2012.13905 | star_distribution.jpg |
0.551544 | bd9f6bd857e849309a8e76c5bfa5e238 | Mass-radius diagram for low-mass stars, including all measurements for double-lined eclipsing binaries (SB2; filled symbols) as well as determinations for single-lined eclipsing systems (SB1) and single stars (open symbols). Solar-metallicity Dartmouth isochrones are shown for comparison, for ages ranging from 1 to 13 Gyr (grey band) <cit.>. | 2012.13907 | mass_radius.jpg |
0.505613 | 775ab39a52ca48ed94b5baf16c277ec9 | Mass-radius relation for low-mass and sub-stellar eclipsing binaries in star-forming regions, open clusters, and globular clusters. The primaries and secondaries are plotted as dots and squares, respectively. Colour scheme as follows: 1–5 Myr (cyan), 5–10 Myr (blue); 125 Myr (orange); 600 Myr (red); globular clusters (yellow). The primaries and secondaries of EBs from the DEBcat database whose accuracies are better than 2% on their masses and radii are displayed as black and grey symbols. Overplotted with green lines are the BT-Settl isochrones for ages of 1 Myr, 5 Myr, 10 Myr, 120 Myr, 625 Myr, 1 Gyr, and 10 Gyr. We added the 5 Gyr-old low-metallicity tracks at [M/H] = -2.0 and -1.0 dex as orange and red lines, respectively. | 2012.13907 | plot_clusters_EBs_Mass_Radius.jpg |
0.502701 | 1a7f30fd494242bcaa580a6e11b98a84 | Comparison between the shape functions of our models (red and blue lines), the standard approach proposed in <cit.> (grey line) and the model proposed in <cit.> (green line). The value of the β parameter chosen for our model (<ref>) is β=2. Our Padé expansion better adapts to the standard approach than other phenomenological ansatz. | 2012.13908 | plot6.jpg |
0.446625 | 4ba5f378817f42208660990a5c3164e4 | Schematic illustration of the active particles interacting via nearest-neighbour quadratic potential (shown by green springs) of spring constant K in one dimension. We study the dynamics of the tagged particle (shown in red). The position of the α-th particle is denoted by r⃗_α (t). | 2012.13910 | schematic-3.jpg |
0.504808 | 736f03201e91449a8ae0eba8686aa0d2 | Illustration of three time scales in the model of harmonic chain of active particles - (i) τ_K = 1/K due to the interaction among particles, (ii) τ _A = 1/γ / 1/ζ / 1/D_rot due to the activity of AOUP / RTP /ABP and (iii) τ _N =N^2/K due to the finiteness of the ring. As shown in the figure, we will consider τ_K ≪τ_A ≪τ_N in our paper. | 2012.13910 | time-draw-2.jpg |
0.462299 | b69f72845c154837a44da02e1bfe0afc | a) Schematics of the system composed by N layers. The electric field is applied along the out-of-plane direction and it is modeled by fixing the electrostatic potential at the i=0 and i=N+1 layers. b) Diagrammatic form of the Poisson equation and the gap equation. | 2012.13911 | Fig1.jpg |
0.408233 | 0f84174ca2b24a64ba61d6b20210d855 | a) Histogram of the eigenvalues multiplicity (total density of states) in absence of applied gate. b) Profile of the self-consistent potential for different values of the relative dielectric constant ϵ_r=10, 20, 50. | 2012.13911 | Fig2.jpg |
0.511052 | df5497afac3d456892c636f6727f26b5 | Results of the simulations for a strong pairing U=1 t for a thick slab with N=30. a) Color plot of the local gap versus layer i_N and applied gate voltage ϕ_g. b) Section of a) at ϕ_g=24 t showing gap suppression in the outer most layers. c) Value of the self-consistent charge density versus the layer i_N and the applied gate, showing charge depletion in the outer most layers. d) Self-consistent charge density at ϕ_g=24 t showing full depletion at the outermost layer. | 2012.13911 | Fig3.jpg |
0.416335 | 98c92a576d4a4942aac74a5591409b8b | Average gap Δ versus applied gate ϕ_g for U=t varying a) the number of layers N=10,20,30 for ϵ_r=10, and b) the relative dielectric constant ϵ_r=10, 20, 50 for N=20. In-plane grid with N_x=N_y=100. | 2012.13911 | Fig4.jpg |
0.411087 | 7866af632d1a4ec6849365e60bd78766 | a) Normalized average gap versus applied gate voltage ϕ_g for different values of the pairing strength U/t = 1, 0.8, 0.6 for ϵ_r=10. b) Gap calculated through Eq. (<ref>) with the density of states ν(E) resulting from the self-consistent potential. Fixed parameters are N=20 and N_x=N_y=100. | 2012.13911 | Fig5.jpg |
0.438601 | 136e722a48094595baa4526fb797276b | a) Density of states versus energy after a smearing with broadening 10^-5t for a system with N=20 layers and ϵ_r=50 at zero applied field. Inset: zoom on a small energy window around the Fermi level at E=0. b) DoS at the Fermi level as a function of the applied gate voltage for ϵ_r=10, 20, 50. c) Gap resulting from solution of Eq. (<ref>) for ϵ_r=50 and U=0.6 t. | 2012.13911 | Fig6.jpg |
0.410399 | 3c58afaf25f74ea99a78728713f9ef3e | Gap as a function of the applied gate voltage for a system characterized by N=20 and N_x=N_y=80, with in-plane periodic boundary conditions. a) U=0.5 t varying ϵ_r. b) ϵ_r=10 varying U. | 2012.13911 | Fig7.jpg |
0.457609 | bfdd674adc5643d3bed23414d18eee03 | Gap as a function of the applied gate potential for a in-plane triangular tight-binding model characterized by N=20, ϵ_r=10 and N_x=N_y=100 for different values of the pairing U/t=0.8, 0.6, 0.5. | 2012.13911 | Fig8.jpg |
0.43331 | 3f930ae94c9a4d1f98de67404d35d44e | (a) Two-electron eigenenergies as a function of the energy detuning for the magnetic field B=1 T. The levels E_4, E_5 anticross at ε=0.5 meV due to the spin-orbit interaction. The two vertical arrows indicate possible transitions which can be induced by the AC fields defined in the main text Eq. (<ref>) and Eq. (<ref>). (b) Spin-orbit gap (Δ_so=E_5-E_4 at the anticrossing point) as a function of the detuning. In this case the magnetic field is detuning dependent. The AC induced current is computed for the marked points. | 2012.13914 | fig1.jpg |
0.464783 | 3e781ad4a3cc467ca6feabbb054df0bf | Current as a function of AC frequency, when the AC field modulates the tunnel barrier with the AC amplitude A_b=10 μeV, and the energy detuning with A_d=10 μeV. The constant value of the detuning is ε=2, 0.5 meV and the corresponding magnetic field is B=0.3, 1 T for (a) and (b) respectively. These fields define the singlet-triplet anticrossing point for each value of the detuning. | 2012.13914 | fig2.jpg |
0.45041 | 8995a2c2d290426199eeeb76e77c72bc | (a) Absolute value of the coupling parameter q_d as a function of the energy detuning and magnetic field for the AC amplitude A_d=10 μeV. The dotted curve defines the anticrossing point for each ε and B. (b) The same as (a) but for q_b with A_b=10 μeV. | 2012.13914 | fig3.jpg |
0.397642 | 7f73f3337da94638b934e004a8175a9d | The ratio q_b/q_d defined in Eq. (<ref>) as a function of detuning for different values of x_so and A_d=A_b. | 2012.13914 | fig4.jpg |
0.449228 | 7b4233fc975646b4b021525b8cc031e3 | As in Fig. <ref>, but (a) A_b=10 μeV and A_d=19A_b, (b) A_b=10 μeV and A_d=4.9A_b. The value of A_d is chosen so that to approximately induce the same current peaks as those induced by A_b. | 2012.13914 | fig5.jpg |
0.562022 | 580e42907021443eb92e388b518313cf | Current as a function of AC frequency, when the AC field modulates the tunnel barrier, with the AC amplitude A_b= 10 μeV. The detuning is ε=2 meV and from the upper to the lower curve the parameter x_so=0.1, 0.04, 0.02, 0. | 2012.13914 | fig6.jpg |
0.404211 | cdcb8032521f46348eb2335680b28504 | Visualization of the vanilla BERT attention (left) and syntax-guided self-attention (right). Weights of attention are selected from first head of the last attention layer. For the syntax-guided self-attention, the columns with weights represent the SDOI for each word in the row. For example, the SDOI of passed contains {name, of, legislation, passed}. Weights are normalized by SoftMax for each row. | 2012.13915 | att_ex.jpg |
0.506461 | 4bd23aac11984fb3b6f8aa6c73af6471 | Example of dependency formats. | 2012.13915 | dp_types.jpg |
0.528127 | 110bb7e9b20248d08358d35c6e8d1888 | (a) Example of syntax-guided span-based QA. The SDOI of each word consists of all its ancestor words. (b-c) The dependency parsing tree of the given passage sentence and question. | 2012.13915 | ex1.jpg |
0.429762 | ff68f3f22d7d43e6a9a63841f97df2dd | Overview of the syntax-guided network. | 2012.13915 | sg-net.jpg |
0.435021 | 11511396a7b54a66b8e73574cf27f06d | An excerpt from a search report showing a claim and cited paragraphs. The document (positive sample) is novelty-destroying for the claim while the document (negative sample) is not novelty-destroying and merely constitutes technical background. | 2012.13919 | Search-report-excerpt.jpg |
0.441893 | a3bd1a9bec37415396d77e5d599fe38d | The coexistence lines for the charged black hole in a cavity, with different values of electric potential Φ. The HP phase transition can occur at all pressures, and T_ HP decreases with Φ at a fixed pressure. Φ has an upper bound of 1/7, above which there is no HP phase transition. | 2012.13921 | RNTp.jpg |
0.463839 | 4b3dc5f5779340059af7475dd1280274 | The Gibbs free energy of the black hole–thermal gas system in a cavity, with different values of pressure p. The G–T curves of the thermal gas and the stable large black holes intersect at the HP temperature T_ HP, which increases at large pressures. Below or above T_ HP, the thermal gas phase or the large black hole phase is globally preferred. The G–T curves of the unstable small black holes are concave and always above the T-axis, so the HP phase transition never occurs. (LBH and SBH stand for large and small black hole, respectively.) | 2012.13921 | SchGT.jpg |
0.479571 | 2e6aca4721ab44efa1957c24aa9e070b | The Gibbs free energy of the Schwarzschild black holes in a cavity and in the AdS space. Both the HP temperature T_ HP and the minimum black hole temperature T_0 in the cavity case are higher than those in the AdS case at the same pressure. | 2012.13921 | SchGTAdScavity.jpg |
0.436449 | fd6673c7be094dc8968f34aab4043ef8 | The coexistence lines for the Schwarzschild black hole in a cavity and in the AdS space. There is no terminal point in the T_ HP–p curve, and the HP phase transition can occur at all pressures. The thermal gas phase lies below the coexistence line, behaving like a solid. The HP temperature T_ HP increases with p and is a little higher in a cavity than in the AdS space. | 2012.13921 | SchTp.jpg |
0.437515 | 95f306f052a9453f8cd0e324c0ebf64b | The CP discovery capability of the CEPC as a function of the Δ and κ_τ true values. The black dashed lines represents several typical values of the significance, √(χ^2_ CPV) = 5,10,15,20. The green region represents the space parameter where the sensitivity is below 95% C.L. The pink region represents the parameter space that can explain the BAU in the lepton flavored EWBG scenario <cit.>. | 2012.13922 | chi_2D_CPV.jpg |
0.504704 | 09a1f52def944b91a2d81518704c740f | The differential distributions of h→τ^+(→ρ^+ ν̅_τ) τ^-(→ρ^- ν_τ) for neutrino momentum δϕ_ν (red), polarimeter δϕ_r (black), acoplanarity ϕ^* (blue), and the Θ variable (green) at the truth level for Δ = 0^∘. | 2012.13922 | comparisson_histograms.jpg |
0.510871 | cdf13f0959734810adc714418089cad3 | eMBB SER vs transmission power in dBm for different URLLC arrival rate. | 2012.13923 | e_1.jpg |
0.414573 | f8be048b6d654b44bce5f0709a076a02 | eMBB reliability vs the SER. | 2012.13923 | e_2.jpg |
0.453958 | c2d0d94b61594f99882708ea3acef853 | eMBB users reliability. Adopted SER=.01 and λ=7 p/msec | 2012.13923 | e_e_1.jpg |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.