nopperl/arxiv-image-text · Datasets at Hugging Face

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
0.442695	4c28ced282fd47a6b7badee0d50d1c6a	GK with momentum (sketch Sample size, τ = 5): comparison among momentum variants on LIBSVM data, residual error and relative error vs CPU time and No. of iterations.	2012.13087	gk_tau_5.jpg
0.460661	19eb7be47d1a4deda3bf911acdaf36f8	GK with momentum (sketch Sample size, τ = 50): comparison among momentum variants on LIBSVM data, residual error and relative error vs CPU time and No. of iterations.	2012.13087	gk_tau_50.jpg
0.445708	e7576a4eba164c7fa1e633974146a9b8	GK with momentum (sketch Sample size, τ = m): comparison among momentum variants on LIBSVM data, residual error and relative error vs CPU time and No. of iterations.	2012.13087	gk_tau_max.jpg
0.567204	a758ad61be834bd8b8f676876aa2e241	GK momentum (effect of sketch sample size τ and capped parameter θ).	2012.13087	gk_tau_theta_.jpg
0.414506	120b1ec4858d479b99b050167d9cd386	Overall pipeline of our model. We fuse RGB and depth information only at the coarsest level and then efficiently do the multi-scale aggregation with CPR module. The IDR branch is to strengthen the less powerful features learned by the mobile networks in a computationally free manner.	2012.13095	newpipeline_crop.jpg
0.572956	4560410bcaf64c75bac78769f1401138	Architecture differences between existing RGB-D SOD methods (a)-(g) and ours (h). (a): late fusion. (b)-(d): early fusion. (e)-(h): multi-scale fusion. For efficiency, we only utilize the highest-level depth features on cross-modal fusion. Compared with (a), we leverage multi-scale RGB features. Best viewed in color.	2012.13095	rgbd-archs.jpg
0.405502	eb56886ab2de45d69c0304805223fd08		2012.13097	_density104.jpg
0.441134	d9171b86bb15491c9656336d694bdb55		2012.13097	_density13.jpg
0.350979	758eeda2340f43c19945a6acc15674d8		2012.13097	_density_7.jpg
0.46399	d1a958c8d9444a558c604ae8723bd692		2012.13097	_density_classes_phase126.jpg
0.447822	71d2ccc3aefd4ce9a0da5dd617037b54		2012.13097	_density_classes_phase35.jpg
0.455027	a904b7aeaf2c41a7974a09eb9e261f7b		2012.13097	_density_classes_phase692.jpg
0.440951	a730b2e9cc1d45f480b6a821975eacfb		2012.13097	_density_imag.jpg
0.389825	530e0fdae7ab4baf827d2400c4b6c10e		2012.13097	_density_mediumgaussian160.jpg
0.397739	567ab7995c28472dbe42fa631ba359b0		2012.13097	_density_mediumgaussian_phase9.jpg
0.60074	8207ecafd20e4ee0a6a20ec7eebd23f5		2012.13097	_density_noise_ring9.jpg
0.449095	db1ac191d69446bf8109990c4408630b		2012.13097	_density_real.jpg
0.480973	57c9cde3bfc24390b790a2fb6667c8f7		2012.13097	_density_ring14.jpg
0.515482	5fda00d5294248109c55c207fb4abef1		2012.13097	_density_stripes_ring8.jpg
0.41341	4150ff8f003c47dc9bae22de0364256f		2012.13097	_density_stronggaussian120.jpg
0.425356	1f79464ac35b47c697be1d3d07b75d54		2012.13097	_density_stronggaussian_phase185.jpg
0.447878	cd0dbbe83ef54370b76efcb9df8518d5		2012.13097	_density_strongstripes112.jpg
0.348393	8057975903844d6f820b2fe3bef5b547		2012.13097	_density_superstrongstripes50.jpg
0.445392	1b68927b37964d0aa16ae00504c25a17		2012.13097	_density_weakgaussian142.jpg
0.389511	e2f6305773284effb56ade4ab165dde7		2012.13097	_density_weakgaussian_phase180.jpg
0.398996	e6d7d095f413497db7113bc494bb7e93		2012.13097	_density_weakstripes54.jpg
0.446794	04b3898d17c644089a203f57a2c412a5		2012.13097	_histo_imag.jpg
0.447968	88406995c9e54d1c943f53c9dc6beece		2012.13097	_histo_real.jpg
0.431441	0da7eb6e3c5a4fcebea49a76df55da9a		2012.13097	_phase13.jpg
0.423471	36ccffdda73b432e8385718979fad1eb		2012.13097	_phase_7.jpg
0.442628	e7bd37d3428b4fdca4297c72c636d9eb		2012.13097	_phase_classes_phase126.jpg
0.457503	b2b3b0f778164ff5a96412c1ca940f05		2012.13097	_phase_classes_phase35.jpg
0.441587	02fbb09f3a7b40908614543892b6a917		2012.13097	_phase_classes_phase692.jpg
0.446422	a0fcaf00f7b7424a960d16e46be8b396		2012.13097	_phase_imag.jpg
0.446648	86a054eee73b40d2842dd77163bc35dd		2012.13097	_phase_mediumgaussian_phase9.jpg
0.433092	4bda0be657a34161bcf9ab4a3cb0d281		2012.13097	_phase_noise_ring9.jpg
0.45917	0b1f9e1d1fa54a479094329d01ff5876		2012.13097	_phase_real.jpg
0.448182	254dcaaf2a9e4e4f9f56893069c0e1ad		2012.13097	_phase_ring14.jpg
0.485854	6f06dd65c684426083deffd94011e38d		2012.13097	_phase_stripes_ring8.jpg
0.450424	dee42a01bb924f12bc9bcff1ca0c91e4		2012.13097	_phase_stronggaussian_phase185.jpg
0.413351	212a0493a8dc4e48aa461039ddeed80c		2012.13097	_phase_weakgaussian_phase180.jpg
0.445397	689f9dd687c045c5828f4b065538d850	The network takes as input images of dimension 256× 256 with the condensate density alone (a), or the density and phase profile as two separate channels (b). The images are fed through 7 convolutional and 3 maxpool layers until the final layer outputs 3 matrices of dimension 64× 64. Each entry of the 64× 64 matrices is associated with a distinct 4× 4 cell in the original image and represents the probability of a vortex core being present inside the cell (I), and the scaled x (II), and y (III) position of the vortex within that cell.	2012.13097	architecture-07-07-07-07.jpg
0.527334	10d18a2a6f264e3cae9a112fa5e37423		2012.13097	density_roc.jpg
0.623289	faff50e6d6fb4fdc918ff5ce4e52593f		2012.13097	phase_roc2.jpg
0.434098	ef66dffc676b45fdb3fdd2754fe04662	The figure shows the first six iterates of f acting on the wandering domain, which lies inside the black disk. There are three points around this disk which are part of the sequence of points that accumulates everywhere on the boundary of Ω. Points x_1, x_2^N_1+1 and x_3^N_2+2, that belong to the orbit of points x_1, x_2 and x_3 respectively, and are mapped to the attracting fixed point 1.	2012.13284	Oscillating.jpg
0.480819	6d4f4cadaecb42c2957ff8b121c63729	The figure shows the first three iterates of f acting on the wandering domain Ω, which lies inside the black disk. There are four points around this disk which are part of the sequence of points that accumulates everywhere on the boundary of Ω. Points x_1, x_2^1, x_3^2 and x_4^3, that belong to the orbit of a point x_1, x_2, x_3 and x_4 respectively, are mapped to the attracting fixed point at the origin. The action of f on the orbit of Ω is approximately a translation by 4.	2012.13284	escaping3.jpg
0.598449	583798d28c9f4d7e91e2aa42b111f1b2	π̃_n(-1) - V^*_n(-1).	2012.13288	fig1.jpg
0.502239	d8bb78b870b143138195288191d33ca5	Scale dependence of all non-vanishing TripoSH coefficients for the ellipticity correlation functions, _0 ± 2Ξ_ℓℓ_1 ℓ_2^δγ (left panel for ℓ = even and center panel for ℓ = odd), _0 ± 2Ξ_ℓℓ_1 ℓ_2^u γ (center panel), _± 2 ± 2Ξ_ℓℓ_1 ℓ_2^γγ and _± 2 ∓ 2Ξ_ℓℓ_1 ℓ_2^γγ (right panel). Taking their absolute values, the results multiplied by x_12^2 or x_12/(aH) are plotted as function of separation, x_12. Here, we specifically consider the case with z_1 = z_2 = 0.3 and b_ g = b_ K = 1. For the δγ correlation, distinct results are obtained in real and redshift space, and their non-vanishing coefficients are depicted as dashed and solid lines, respectively. Note that all the coefficients shown here are invariant under the sign inversion of helicities [see Eq. (<ref>)]	2012.13290	xi_TripoSHcoeff_per_x12.jpg
0.466278	4eb7c529cc944517a755c1883ee08db2	Ellipticity correlation functions, ξ_0 ± 2^δγ, ξ_0 ± 2^u γ, ξ_± 2 ± 2^γγ and ξ_± 2 ∓ 2^γγ (top), and the ratios of these to the corresponding PP-limit results (bottom), plotted against the separation x_12 or the opening angle Θ for a specific case with z_1 = z_2 = 0.3 and b_ g = b_ K = 1. Note that the u γ correlation in the PP limit strictly vanishes. Hence, in the bottom panel, instead of plotting the u γ result, we show the δδ case for a reference. For the δγ correlation, the results in real and redshift space are shown in dashed and solid lines, respectively.	2012.13290	xi_xiPP_per_x12.jpg
0.455716	e20db4eef5314abf9aabe7ceba722f90	Percentage of active while infected population.	2012.13296	active.jpg
0.535955	46cc90a1ce56488193e2f511e6709823	Patient count heatmap.	2012.13296	date_location_heatmap.jpg
0.42166	c9860247ebf94822b371119d66e94d58	Percentage of infected population when simulating patient isolation.	2012.13296	final.jpg
0.451291	5ff81b12198b48549209735d4c0fc17d	Mobility of Seoul population over time by age group according to cell-phone data provided by SK telecom.	2012.13296	floating_age_label.jpg
0.47275	36a3da3642364908a1e9b2667bf8c407	Outbreak associated with one particular hotspot location in Gangwon-do.	2012.13296	gangwon-do_outbreak.jpg
0.425831	bb6def4d9e46406d8ab51a789d947cc3	Subset of known infection spread between patients. Red nodes indicate patients with route information who infected others. Green nodes indicate patients who infected others but do not have any route information. Blue nodes indicate patients who did not infect anyone else.	2012.13296	infect_by_gt.jpg
0.417719	6729f0a8be1e428a994106c0d852a435	Percentage of isolated population.	2012.13296	isolated.jpg
0.525616	8bd91bc89ff943d281bb99215a7101cf	(a) The structure of Pt_2HgSe_3/CrI_3 heterostructure. The pink, silvery, green, blue and purple balls are for Hg, Pt, Se, Cr and I atoms, respectively. d_0 represents the optimized vdW gap and the two buckled Hg sites are respectively denoted as Hg_1, Hg_2. (b) Band structure of monolayer Pt_2HgSe_3 without and with SOC, respectively.	2012.13298	Fig1.jpg
0.496868	134e9e0a9ef949049fc5ebabd4f00966	(a) The influence of symmetric and anti-symmetric exchange field on the band gaps in K and K' valleys. (b)-(c) The band structure and orbital magnetic moment around K and K' valleys for (b) m_1=0.6λ and (c) m_1=1.5λ, respectively. The inset of (c) shows the band near Fermi energy with inclusion of small Rashba spin-orbit coupling. The red (blue) color denotes spin up (down) state. (d)-(e) The corresponding orbital magnetization in an unit cell and anomalous Hall conductance. The hollow (solid) square represent the system without (with) Rashba spin-orbit coupling.	2012.13298	Fig2.jpg
0.407897	7364553f250c4c8f83bf3adc4b417d67	(a)-(b) Band structure of Pt_2HgSe_3/CrI_3 without (a) and with (b) SOC. In (a), the red (blue) represents spin up (down) state. In (b), Δ_K=-105.5meV and Δ_K'=167.1meV with valley splitting Δ=(Δ_K'-Δ_K)/2=136.3meV. The gap with negative sign indicates the presence of band inversion. (c) The Berry curvature along the high symmetry lines. The inset shows the Berry curvature in the first Brillouin zone. (d) The Fermi energy dependence of the anomalous Hall conductivity. (e) The local density of states of a semi-infinite zigzag ribbon of Pt_2HgSe_3/CrI_3 heterostructure.	2012.13298	Fig3.jpg
0.421983	d846d4ebbecd48109476acfbef0676f0	(a) Lattice and (b) electronic structures of h-BN/Pt_2HgSe_3/CrI_3 with SOC. Here, Δ_K=-113.4meV and Δ_K'=137.4meV with valley splitting Δ=125.4meV. The gap with negative sign indicates the presence of band inversion. (c) vdW superlattice consisting of h-BN/Pt_2HgSe_3/CrI_3 multilayer as building blocks to realize abundant topological phases coupled spin, valley and layer degree of freedoms with Chern numbers.	2012.13298	Fig4.jpg
0.46467	706db258b281430b8e64b9d53173991b	Spectrum of approaches to solve a dynamic resource allocation problem under uncertainty. A completely centralized approach deals with a monolithic state representation. In a completely decentralized approach, each agent simulates what other agents do and performs its own action estimation. Our hierarchical approach segments the planning problem into sub-problems to improve scalability without agents estimating other agents' actions.	2012.13300	PlanningParadigm_v2.jpg
0.464261	322e3d94804d401fabbe2b0134b5feae	Subfigure (a) – The various spatial regions under consideration. Pins on the map represent depot locations, and different colors represent different spatial regions. Subfigure (b) – Nashville's historic incident density from January 2018 to May 2019 overlaid on the spatial grid environment.	2012.13300	combined_regions_heatmap.jpg
0.422612	cc0ca21990f34e21a15f2b4fc4e2ef15	Different approaches to solve a dynamic resource allocation problem under uncertainty. A completely centralized approach (leftmost) deals with a monolithic state representation. Using such a representation, the decision-maker could estimate state transition probabilites using a simulator <cit.> or use an online solution approach <cit.>. A completely decentralized approach involves each agent simulating what other agents do and performing its own action estimation (rightmost). Our approach (shown in the middle) segments the original planning problem into multiple sub-problems to improve scalability without agents estimating other agents' actions.	2012.13300	planning_horizon.jpg
0.478809	35d3a36983a0444d8df37347957902ce	Solution Approach	2012.13300	software_framework_v2.jpg
0.429604	17e6c888adc64528a913959da63b75cc	System State and Actions.	2012.13300	state.jpg
0.419844	a3d0d1860daf4294b21513a556584aad	Overview of the hierarchical planning approach. We first divide the problem into regions which induce smaller sub-problems. The high-level planner optimizes the distribution of agents among the regions periodically. The low-level planner optimizes the distribution of agents within each region and dispatches agents to incidents as they occur.	2012.13300	systemModel.jpg
0.413494	9b3516bcaf2a4d4a9b9da19868e0ca5f	Blind demixing in the karate network. Demixing Error (DE) as a function of the noise level σ_n for different number of filters () and seeding nodes (). The DE decreases as the sum + also decreases. When + is fixed, smaller results in higher demixing accuracy. In the three cases with +≥5 where the error does not converge to zero as the noise power goes to zero, demixing is too hard given the size of the karate network (=34) to achieve successful recovery.	2012.13301	karate.jpg
0.564764	63179daed9d34f00ae07b0eb3dc5f349		2012.13301	predictor.jpg
0.437234	5cce343acf184e6cb35d7f425a49a9b5		2012.13301	predictor_explanation.jpg
0.531651	2a30a510b47e49ccb77b58a84e1522e3	Illustration of the state-of-art and our new way of redistributing the share of weights in online learning.	2012.13307	fig-directionalExpert.jpg
0.381966	68744797ae2e4c868b87f6126b5758fa	Illustration of a typical learning procedure, where each candidate model _j is trained in terms of [_j] in its parameter space ℋ_n[_j], and then used to evaluate future data under some loss function ℓ(·).	2012.13307	fig-learning.jpg
0.442195	d0b31306bf1e42f2b13ad0c11ab87b2d	(a) Crystal structure of GCO. Dashed black lines mark the boundaries of the structural primitive cell. The primitive cell consists of two GeO4 tetrahedra and one Co4 tetrahedral unit. Here Co, Ge, and O ions are presented by the blue, gray, and red color, respectively. (b) Schematic representation of q = (1/2, 1/2, 1/2) AFM ordering in GCO. Magnetic moments at Co sublattice are shown using arrows in a collinear setting, i.e., majority or up (+) and minority or down (-) spin states are denoted using up and down arrows, respectively. Alternating kagomé (KGM) and triangular (TRI) planes are highlighted using dashed horizontal lines. A T_+ K_+ T_- K_- T_+ K_+ T_- K_-⋯ type AFM spin configuration of TRI (T_±) and KGM (K_±) layers can be noticed along the [111] direction.	2012.13310	Figure1_GCO_struct.jpg
0.453039	3daad286025e4a5498eb886be4cdc917	ßComparison of the DFT+U predicted phonon frequencies (ωtheory) for the simulated paramagnetic phase with the experimentally measured frequencies at 300 K (ωexpt) for the IR and Raman-active modes. The data plotted in this figure were obtained from Table <ref> and <ref>.	2012.13310	compare_ir_raman_fig.jpg
0.449178	4a7e53a523e04c09a5345cae4720162c	Raman spectra of GCO recorded at temperatures T = 5, 10, 12, 14, 18, 20, 21, 22, 24, 26, 30, 40, 60, 80, 100, and 300 K.	2012.13310	fig-6.jpg
0.553343	b6bcecd4545946a19847d9539b2f1f78	Fourier-transform infrared spectrum of GCO polycrystalline sample recorded at room temperature. ßThe DFT+U simulated IR spectrum is given in the SM <cit.>.	2012.13310	fig-IR.jpg
0.511384	6f13c2b04a814e08b7751fd6521d0e27	The temperature dependence of the line separation Δω of the low-frequency shoulder from the position of the T2g(1) peak. Inset marks the shoulder appearing on the low-frequency side of the T2g(1) peak.	2012.13310	fig-line-seperation.jpg
0.45594	09e1763f4f264e148e6e666bb4446dde	Atomic vibration patterns for all four IR-active phonon modes: (i) T1u(1), (ii) T1u(2), (iii) T1u(3), and (iv) T1u(4). The color coding of atoms is the same as in Fig. <ref>(a). These modes are listed here in the order of decreasing frequency (see Table <ref>).	2012.13310	figure_IR_eigenmodes_b.jpg
0.44845	84865a9ad3cf4a7d97d24771e671d49d	(a) Definition of all four magnetic-exchange interactions, first (J_1), second (J_2), and third (J_3 and J_3^') NN, considered in this work. Co atoms are shown in blue color. Ge and O atoms are omitted for clarity. Note that J_3^' passes through an intermediate Co atom (see text). (b) Fitting of the DFT (PBEsol+U) energy values computed for various different spin configurations in a doubled primitive cell, as shown in (a), with our model spin Hamiltonian. Here, we decide to choose the PBEsol+U method since it predicts better lattice parameters compared to the PBE+U predictions.	2012.13310	figure_Jij.jpg
0.441759	8e1cf7038be0464580323e9a6ce8f888	Atomic vibration patterns for all five Raman-active phonon modes: A1g, T2g(1), T2g(2), Eg, and T2g(3). These modes are listed here in the order of decreasing frequency (see Table <ref>).	2012.13310	figure_Raman_eigenmodes_b.jpg
0.425033	b936e9f6961143ed8a582b6d5820088c	The temperature dependence of Raman intensity (I×10^4), fullwidth at half maximum (Δ F), and Raman-peak position (RPP) for the A1g, T2g(1), T2g(2), and Eg Raman active modes. The lines connecting the data points are visual guides. The TN and TS mark the transition temperatures corresponding to the antiferromagnetic and the cubic-to-tetragonal structural phase transitions, respectively.	2012.13310	temp-dept-Raman-modes-21-Dec-2020.jpg
0.451178	2f89b841117745caa84f5583d6c0603e	Optimal proposal distribution corresponding to A.	2012.13311	opt_proposal_1.jpg
0.476322	8abdfde3886745699408cb6c279e3ec4	Overview of variational determinant estimation. A spherical proposal distribution q(s) is trained to estimate the absolute determinant of a matrix A.	2012.13311	overview.jpg
0.406881	d2c3520a3fd34dffab0c1203a8c6b3fc	Linear worst-case response to spanwise optimal perturbations with (α,β)=(0,2) and Re=3000.	2012.13314	a0b2_TEG.jpg
0.493095	573bc47d8f0d43f6a307b54e803e38fb	(α,β) = (0,2) transient energy growth of optimal disturbance in uncontrolled flow with various amplitudes of E_0.	2012.13314	a0b2_base_E.jpg
0.429239	c51d9b55912b4734900a5d5436f1cf5d	Transient energy growth of uncontrolled and controlled flows due to linear optimal perturbations of E_0 =1× 10^-4. Inserts correspond to iso-surfaces of 𝒬-criterion <cit.> (𝒬(h/u̅_c)^2=4× 10^-4) colored by streamwise vorticity ω_x h/u̅_c.	2012.13314	a0b2_flowfield.jpg
0.458152	d7d93c49a03b4753bbefdc4a87842108	DNS of the uncontrolled flow and the full-information LQR controlled flow with (α,β)=(0,2), Re=3000, and E_0=5× 10 ^-5. (Top) Friction velocity and (Bottom) normalized wall-normal velocity v(-h)/√(E_0) at the lower channel wall with x=0, z=0.	2012.13314	a0b2_lqr_op_friction_v.jpg
0.507097	2422c56174d74b7c82f0c881c1868bca	Comparison between uncontrolled flow (left) and LQR controlled flow (right) with (α,β) = (0,2), Re=3000. Modification of instantaneous streamwise velocity gradient in spanwise direction ∂u/∂z at tu̅_c/h=49, x/h=0.m	2012.13314	a0b2_op_lqr_shear49_update.jpg
0.365394	28fe49bb45474136b4a8b8addbe998e5	Robustness of LQR and SOF-LQR-s controllers to off-design Reynolds numbers. Both controllers are designed for Re=3000, but robustly reduce TEG from spanwise optimal perturbations over a range of off-design Reynolds numbers.	2012.13314	a0b2_robustness_toRe.jpg
0.418594	33023ae46d97481f80d1cf5badd6ec89	DNS of the uncontrolled flow and the SOF-LQR-s controlled flow with (α,β)=(0,2), Re=3000, and various perturbation amplitudes E_0. (Top) Friction velocity and (Bottom) normalized wall-normal velocity v(-h)/√(E_0) at the lower channel wall x=0, z=0.	2012.13314	a0b2_sof_op_friction_v.jpg
0.528878	715a882406b84a22adc12389a6f43388	Linear worst-case response to streamwise optimal perturbations with (α,β)=(1,0) and Re=3000.	2012.13314	a1b0_TEG.jpg
0.482637	cb174982d9ee46829ea2e087a86651b5	(α,β) = (1,0) transient energy growth of optimal disturbance in uncontrolled flow with various amplitudes of E_0.	2012.13314	a1b0_base_E.jpg
0.405074	a6c9fc0f46ea4f0cae60f60fb28dee22	Transient energy growth energy and friction velocity on walls of uncontrolled flow (E_0 =1× 10^-4). Inserts are corresponding iso-surface of 𝒬-criterion <cit.> (𝒬(h/u̅_c)^2=1× 10^-4) colored by streamwise vorticity ω_x h/u̅_c.	2012.13314	a1b0_flowfield.jpg
0.39393	0c72c54aed8a4cc9b6933bec57bf6e33	Robustness of LQR and SOF-LQR controllers to off-design Reynolds numbers. Both controllers are designed for Re=3000, but robustly reduce TEG from streamwise optimal perturbations over a range of off-design Reynolds numbers.	2012.13314	a1b0_robustness_toRe.jpg
0.481674	4b305b2aa28443a8a21b2e193d8edb22	Illustration of wall-based sensing and actuation in channel flow.	2012.13314	flow_configuration.jpg
0.387115	ab9d910bdce645569c76bd7c2e1f1b73	Plot with portfolio sizes 1 through 20.	2012.13315	big_portfolio.jpg
0.443927	a0ba4305988c4787b8020bfcb60332b5	Three partnership networks that have the same aggregate network. (a) Dynamic partnership network without concurrency. Each node is either isolated (k=0) or involved in a non-concurrent relationship (k=1) at each time. (b) Dynamic partnership network with concurrency. At two out of five times, a cycle graph is formed such that all nodes have degree 2. To compensate for a relatively large mean degree at these times, the network at the other three discrete times are the empty network, where every node is isolated. (c) Static partnership network with concurrency. The complete graph is presented. The weight of each edge is 1/5 of that of the edge shown in (a) and (b). The time-averaged network of the dynamic partnership networks shown in (a) and (b) is the complete network shown in (c).	2012.13317	concurrency-comparison.jpg
0.433534	20ec65f3b01f4be685a154f9743a890b	An illustration of the time course of the measures κ_1 and κ_3 (shown in the bottom panel) in a temporal network (shown by a timeline plot of the events in the top panel). Here we assume that the first event of a pair of nodes is the beginning of a “partnership” and that the last event is the end of it, i.e., we consider the momentary network at the given time.	2012.13317	kappas.jpg
0.486104	9ec6a679b708433a93a1a79e887c1b1d	A motivating example for the concurrency measure κ_3 (adapted from Ref. <cit.>). Panels (a) and (b) show two networks of currently active partnerships. From a disease spreading point of view, the situation in (a) is worse than the one in (b). This is more evident if one considers the corresponding line graphs in panels (c) and (d). The concurrency measure κ_1—basically the average degree of the networks—is the same between (a) and (b). However, the concurrency measure κ_3—basically the average degree of the line graphs shown in (c) and (d), which correspond to (a) and (b), respectively—correctly identifies the first situation as being more densely connected (and thus worse).	2012.13317	line_graph.jpg
0.431075	02e58960338d4110a34164c76b14f9ab	Schematic of temporal networks with three nodes that are different in the degree of concurrency. (a) Serial monogamy. (b) Concurrent partnerships. A vertical bar represents a time-stamped interaction event between a pair of nodes, neglecting the duration of the event. The thicker, shaded lines are examples of time-respecting paths from 1 to 3 and, only in (b), from 3 to 1. This figure is inspired by Fig. 1 of Ref. <cit.>.	2012.13317	schem-miller-concurrency.jpg
0.481954	65110d2b334d46d5a8c0295907ce58c9	Two ways of projecting time out of a temporal network. Panel (a) shows a temporal network. Panel (b) shows the network of nodes that have an event within the time window [t_0,t_1]. Panel (c) shows a network of concurrent partnerships—pairs of nodes that have an event before and after [t_0,t_1].	2012.13317	window_concurrent.jpg

0.442695

4c28ced282fd47a6b7badee0d50d1c6a

GK with momentum (sketch Sample size, τ = 5): comparison among momentum variants on LIBSVM data, residual error and relative error vs CPU time and No. of iterations.

2012.13087

gk_tau_5.jpg

0.460661

19eb7be47d1a4deda3bf911acdaf36f8

GK with momentum (sketch Sample size, τ = 50): comparison among momentum variants on LIBSVM data, residual error and relative error vs CPU time and No. of iterations.

2012.13087

gk_tau_50.jpg

0.445708

e7576a4eba164c7fa1e633974146a9b8

GK with momentum (sketch Sample size, τ = m): comparison among momentum variants on LIBSVM data, residual error and relative error vs CPU time and No. of iterations.

2012.13087

gk_tau_max.jpg

0.567204

a758ad61be834bd8b8f676876aa2e241

GK momentum (effect of sketch sample size τ and capped parameter θ).

2012.13087

gk_tau_theta_.jpg

0.414506

120b1ec4858d479b99b050167d9cd386

Overall pipeline of our model. We fuse RGB and depth information only at the coarsest level and then efficiently do the multi-scale aggregation with CPR module. The IDR branch is to strengthen the less powerful features learned by the mobile networks in a computationally free manner.

2012.13095

newpipeline_crop.jpg

0.572956

4560410bcaf64c75bac78769f1401138

Architecture differences between existing RGB-D SOD methods (a)-(g) and ours (h). (a): late fusion. (b)-(d): early fusion. (e)-(h): multi-scale fusion. For efficiency, we only utilize the highest-level depth features on cross-modal fusion. Compared with (a), we leverage multi-scale RGB features. Best viewed in color.

2012.13095

rgbd-archs.jpg

0.405502

eb56886ab2de45d69c0304805223fd08

2012.13097

_density104.jpg

0.441134

d9171b86bb15491c9656336d694bdb55

2012.13097

_density13.jpg

0.350979

758eeda2340f43c19945a6acc15674d8

2012.13097

_density_7.jpg

0.46399

d1a958c8d9444a558c604ae8723bd692

2012.13097

_density_classes_phase126.jpg

0.447822

71d2ccc3aefd4ce9a0da5dd617037b54

2012.13097

_density_classes_phase35.jpg

0.455027

a904b7aeaf2c41a7974a09eb9e261f7b

2012.13097

_density_classes_phase692.jpg

0.440951

a730b2e9cc1d45f480b6a821975eacfb

2012.13097

_density_imag.jpg

0.389825

530e0fdae7ab4baf827d2400c4b6c10e

2012.13097

_density_mediumgaussian160.jpg

0.397739

567ab7995c28472dbe42fa631ba359b0

2012.13097

_density_mediumgaussian_phase9.jpg

0.60074

8207ecafd20e4ee0a6a20ec7eebd23f5

2012.13097

_density_noise_ring9.jpg

0.449095

db1ac191d69446bf8109990c4408630b

2012.13097

_density_real.jpg

0.480973

57c9cde3bfc24390b790a2fb6667c8f7

2012.13097

_density_ring14.jpg

0.515482

5fda00d5294248109c55c207fb4abef1

2012.13097

_density_stripes_ring8.jpg

0.41341

4150ff8f003c47dc9bae22de0364256f

2012.13097

_density_stronggaussian120.jpg

0.425356

1f79464ac35b47c697be1d3d07b75d54

2012.13097

_density_stronggaussian_phase185.jpg

0.447878

cd0dbbe83ef54370b76efcb9df8518d5

2012.13097

_density_strongstripes112.jpg

0.348393

8057975903844d6f820b2fe3bef5b547

2012.13097

_density_superstrongstripes50.jpg

0.445392

1b68927b37964d0aa16ae00504c25a17

2012.13097

_density_weakgaussian142.jpg

0.389511

e2f6305773284effb56ade4ab165dde7

2012.13097

_density_weakgaussian_phase180.jpg

0.398996

e6d7d095f413497db7113bc494bb7e93

2012.13097

_density_weakstripes54.jpg

0.446794

04b3898d17c644089a203f57a2c412a5

2012.13097

_histo_imag.jpg

0.447968

88406995c9e54d1c943f53c9dc6beece

2012.13097

_histo_real.jpg

0.431441

0da7eb6e3c5a4fcebea49a76df55da9a

2012.13097

_phase13.jpg

0.423471

36ccffdda73b432e8385718979fad1eb

2012.13097

_phase_7.jpg

0.442628

e7bd37d3428b4fdca4297c72c636d9eb

2012.13097

_phase_classes_phase126.jpg

0.457503

b2b3b0f778164ff5a96412c1ca940f05

2012.13097

_phase_classes_phase35.jpg

0.441587

02fbb09f3a7b40908614543892b6a917

2012.13097

_phase_classes_phase692.jpg

0.446422

a0fcaf00f7b7424a960d16e46be8b396

2012.13097

_phase_imag.jpg

0.446648

86a054eee73b40d2842dd77163bc35dd

2012.13097

_phase_mediumgaussian_phase9.jpg

0.433092

4bda0be657a34161bcf9ab4a3cb0d281

2012.13097

_phase_noise_ring9.jpg

0.45917

0b1f9e1d1fa54a479094329d01ff5876

2012.13097

_phase_real.jpg

0.448182

254dcaaf2a9e4e4f9f56893069c0e1ad

2012.13097

_phase_ring14.jpg

0.485854

6f06dd65c684426083deffd94011e38d

2012.13097

_phase_stripes_ring8.jpg

0.450424

dee42a01bb924f12bc9bcff1ca0c91e4

2012.13097

_phase_stronggaussian_phase185.jpg

0.413351

212a0493a8dc4e48aa461039ddeed80c

2012.13097

_phase_weakgaussian_phase180.jpg

0.445397

689f9dd687c045c5828f4b065538d850

The network takes as input images of dimension 256× 256 with the condensate density alone (a), or the density and phase profile as two separate channels (b). The images are fed through 7 convolutional and 3 maxpool layers until the final layer outputs 3 matrices of dimension 64× 64. Each entry of the 64× 64 matrices is associated with a distinct 4× 4 cell in the original image and represents the probability of a vortex core being present inside the cell (I), and the scaled x (II), and y (III) position of the vortex within that cell.

2012.13097

architecture-07-07-07-07.jpg

0.527334

10d18a2a6f264e3cae9a112fa5e37423

2012.13097

density_roc.jpg

0.623289

faff50e6d6fb4fdc918ff5ce4e52593f

2012.13097

phase_roc2.jpg

0.434098

ef66dffc676b45fdb3fdd2754fe04662

The figure shows the first six iterates of f acting on the wandering domain, which lies inside the black disk. There are three points around this disk which are part of the sequence of points that accumulates everywhere on the boundary of Ω. Points x_1, x_2^N_1+1 and x_3^N_2+2, that belong to the orbit of points x_1, x_2 and x_3 respectively, and are mapped to the attracting fixed point 1.

2012.13284

Oscillating.jpg

0.480819

6d4f4cadaecb42c2957ff8b121c63729

The figure shows the first three iterates of f acting on the wandering domain Ω, which lies inside the black disk. There are four points around this disk which are part of the sequence of points that accumulates everywhere on the boundary of Ω. Points x_1, x_2^1, x_3^2 and x_4^3, that belong to the orbit of a point x_1, x_2, x_3 and x_4 respectively, are mapped to the attracting fixed point at the origin. The action of f on the orbit of Ω is approximately a translation by 4.

2012.13284

escaping3.jpg

0.598449

583798d28c9f4d7e91e2aa42b111f1b2

π̃_n(-1) - V^*_n(-1).

2012.13288

fig1.jpg

0.502239

d8bb78b870b143138195288191d33ca5

Scale dependence of all non-vanishing TripoSH coefficients for the ellipticity correlation functions, _0 ± 2Ξ_ℓℓ_1 ℓ_2^δγ (left panel for ℓ = even and center panel for ℓ = odd), _0 ± 2Ξ_ℓℓ_1 ℓ_2^u γ (center panel), _± 2 ± 2Ξ_ℓℓ_1 ℓ_2^γγ and _± 2 ∓ 2Ξ_ℓℓ_1 ℓ_2^γγ (right panel). Taking their absolute values, the results multiplied by x_12^2 or x_12/(aH) are plotted as function of separation, x_12. Here, we specifically consider the case with z_1 = z_2 = 0.3 and b_ g = b_ K = 1. For the δγ correlation, distinct results are obtained in real and redshift space, and their non-vanishing coefficients are depicted as dashed and solid lines, respectively. Note that all the coefficients shown here are invariant under the sign inversion of helicities [see Eq. (<ref>)]

2012.13290

xi_TripoSHcoeff_per_x12.jpg

0.466278

4eb7c529cc944517a755c1883ee08db2

Ellipticity correlation functions, ξ_0 ± 2^δγ, ξ_0 ± 2^u γ, ξ_± 2 ± 2^γγ and ξ_± 2 ∓ 2^γγ (top), and the ratios of these to the corresponding PP-limit results (bottom), plotted against the separation x_12 or the opening angle Θ for a specific case with z_1 = z_2 = 0.3 and b_ g = b_ K = 1. Note that the u γ correlation in the PP limit strictly vanishes. Hence, in the bottom panel, instead of plotting the u γ result, we show the δδ case for a reference. For the δγ correlation, the results in real and redshift space are shown in dashed and solid lines, respectively.

2012.13290

xi_xiPP_per_x12.jpg

0.455716

e20db4eef5314abf9aabe7ceba722f90

Percentage of active while infected population.

2012.13296

active.jpg

0.535955

46cc90a1ce56488193e2f511e6709823

Patient count heatmap.

2012.13296

date_location_heatmap.jpg

0.42166

c9860247ebf94822b371119d66e94d58

Percentage of infected population when simulating patient isolation.

2012.13296

final.jpg

0.451291

5ff81b12198b48549209735d4c0fc17d

Mobility of Seoul population over time by age group according to cell-phone data provided by SK telecom.

2012.13296

floating_age_label.jpg

0.47275

36a3da3642364908a1e9b2667bf8c407

Outbreak associated with one particular hotspot location in Gangwon-do.

2012.13296

gangwon-do_outbreak.jpg

0.425831

bb6def4d9e46406d8ab51a789d947cc3

Subset of known infection spread between patients. Red nodes indicate patients with route information who infected others. Green nodes indicate patients who infected others but do not have any route information. Blue nodes indicate patients who did not infect anyone else.

2012.13296

infect_by_gt.jpg

0.417719

6729f0a8be1e428a994106c0d852a435

Percentage of isolated population.

2012.13296

isolated.jpg

0.525616

8bd91bc89ff943d281bb99215a7101cf

(a) The structure of Pt_2HgSe_3/CrI_3 heterostructure. The pink, silvery, green, blue and purple balls are for Hg, Pt, Se, Cr and I atoms, respectively. d_0 represents the optimized vdW gap and the two buckled Hg sites are respectively denoted as Hg_1, Hg_2. (b) Band structure of monolayer Pt_2HgSe_3 without and with SOC, respectively.

2012.13298

Fig1.jpg

0.496868

134e9e0a9ef949049fc5ebabd4f00966

(a) The influence of symmetric and anti-symmetric exchange field on the band gaps in K and K' valleys. (b)-(c) The band structure and orbital magnetic moment around K and K' valleys for (b) m_1=0.6λ and (c) m_1=1.5λ, respectively. The inset of (c) shows the band near Fermi energy with inclusion of small Rashba spin-orbit coupling. The red (blue) color denotes spin up (down) state. (d)-(e) The corresponding orbital magnetization in an unit cell and anomalous Hall conductance. The hollow (solid) square represent the system without (with) Rashba spin-orbit coupling.

2012.13298

Fig2.jpg

0.407897

7364553f250c4c8f83bf3adc4b417d67

(a)-(b) Band structure of Pt_2HgSe_3/CrI_3 without (a) and with (b) SOC. In (a), the red (blue) represents spin up (down) state. In (b), Δ_K=-105.5meV and Δ_K'=167.1meV with valley splitting Δ=(Δ_K'-Δ_K)/2=136.3meV. The gap with negative sign indicates the presence of band inversion. (c) The Berry curvature along the high symmetry lines. The inset shows the Berry curvature in the first Brillouin zone. (d) The Fermi energy dependence of the anomalous Hall conductivity. (e) The local density of states of a semi-infinite zigzag ribbon of Pt_2HgSe_3/CrI_3 heterostructure.

2012.13298

Fig3.jpg

0.421983

d846d4ebbecd48109476acfbef0676f0

(a) Lattice and (b) electronic structures of h-BN/Pt_2HgSe_3/CrI_3 with SOC. Here, Δ_K=-113.4meV and Δ_K'=137.4meV with valley splitting Δ=125.4meV. The gap with negative sign indicates the presence of band inversion. (c) vdW superlattice consisting of h-BN/Pt_2HgSe_3/CrI_3 multilayer as building blocks to realize abundant topological phases coupled spin, valley and layer degree of freedoms with Chern numbers.

2012.13298

Fig4.jpg

0.46467

706db258b281430b8e64b9d53173991b

Spectrum of approaches to solve a dynamic resource allocation problem under uncertainty. A completely centralized approach deals with a monolithic state representation. In a completely decentralized approach, each agent simulates what other agents do and performs its own action estimation. Our hierarchical approach segments the planning problem into sub-problems to improve scalability without agents estimating other agents' actions.

2012.13300

PlanningParadigm_v2.jpg

0.464261

322e3d94804d401fabbe2b0134b5feae

Subfigure (a) – The various spatial regions under consideration. Pins on the map represent depot locations, and different colors represent different spatial regions. Subfigure (b) – Nashville's historic incident density from January 2018 to May 2019 overlaid on the spatial grid environment.

2012.13300

combined_regions_heatmap.jpg

0.422612

cc0ca21990f34e21a15f2b4fc4e2ef15

Different approaches to solve a dynamic resource allocation problem under uncertainty. A completely centralized approach (leftmost) deals with a monolithic state representation. Using such a representation, the decision-maker could estimate state transition probabilites using a simulator <cit.> or use an online solution approach <cit.>. A completely decentralized approach involves each agent simulating what other agents do and performing its own action estimation (rightmost). Our approach (shown in the middle) segments the original planning problem into multiple sub-problems to improve scalability without agents estimating other agents' actions.

2012.13300

planning_horizon.jpg

0.478809

35d3a36983a0444d8df37347957902ce

Solution Approach

2012.13300

software_framework_v2.jpg

0.429604

17e6c888adc64528a913959da63b75cc

System State and Actions.

2012.13300

state.jpg

0.419844

a3d0d1860daf4294b21513a556584aad

Overview of the hierarchical planning approach. We first divide the problem into regions which induce smaller sub-problems. The high-level planner optimizes the distribution of agents among the regions periodically. The low-level planner optimizes the distribution of agents within each region and dispatches agents to incidents as they occur.

2012.13300

systemModel.jpg

0.413494

9b3516bcaf2a4d4a9b9da19868e0ca5f

Blind demixing in the karate network. Demixing Error (DE) as a function of the noise level σ_n for different number of filters () and seeding nodes (). The DE decreases as the sum + also decreases. When + is fixed, smaller results in higher demixing accuracy. In the three cases with +≥5 where the error does not converge to zero as the noise power goes to zero, demixing is too hard given the size of the karate network (=34) to achieve successful recovery.

2012.13301

karate.jpg

0.564764

63179daed9d34f00ae07b0eb3dc5f349

2012.13301

predictor.jpg

0.437234

5cce343acf184e6cb35d7f425a49a9b5

2012.13301

predictor_explanation.jpg

0.531651

2a30a510b47e49ccb77b58a84e1522e3

Illustration of the state-of-art and our new way of redistributing the share of weights in online learning.

2012.13307

fig-directionalExpert.jpg

0.381966

68744797ae2e4c868b87f6126b5758fa

Illustration of a typical learning procedure, where each candidate model _j is trained in terms of [_j] in its parameter space ℋ_n[_j], and then used to evaluate future data under some loss function ℓ(·).

2012.13307

fig-learning.jpg

0.442195

d0b31306bf1e42f2b13ad0c11ab87b2d

(a) Crystal structure of GCO. Dashed black lines mark the boundaries of the structural primitive cell. The primitive cell consists of two GeO4 tetrahedra and one Co4 tetrahedral unit. Here Co, Ge, and O ions are presented by the blue, gray, and red color, respectively. (b) Schematic representation of q = (1/2, 1/2, 1/2) AFM ordering in GCO. Magnetic moments at Co sublattice are shown using arrows in a collinear setting, i.e., majority or up (+) and minority or down (-) spin states are denoted using up and down arrows, respectively. Alternating kagomé (KGM) and triangular (TRI) planes are highlighted using dashed horizontal lines. A T_+ K_+ T_- K_- T_+ K_+ T_- K_-⋯ type AFM spin configuration of TRI (T_±) and KGM (K_±) layers can be noticed along the [111] direction.

2012.13310

Figure1_GCO_struct.jpg

0.453039

3daad286025e4a5498eb886be4cdc917

ßComparison of the DFT+U predicted phonon frequencies (ωtheory) for the simulated paramagnetic phase with the experimentally measured frequencies at 300 K (ωexpt) for the IR and Raman-active modes. The data plotted in this figure were obtained from Table <ref> and <ref>.

2012.13310

compare_ir_raman_fig.jpg

0.449178

4a7e53a523e04c09a5345cae4720162c

Raman spectra of GCO recorded at temperatures T = 5, 10, 12, 14, 18, 20, 21, 22, 24, 26, 30, 40, 60, 80, 100, and 300 K.

2012.13310

fig-6.jpg

0.553343

b6bcecd4545946a19847d9539b2f1f78

Fourier-transform infrared spectrum of GCO polycrystalline sample recorded at room temperature. ßThe DFT+U simulated IR spectrum is given in the SM <cit.>.

2012.13310

fig-IR.jpg

0.511384

6f13c2b04a814e08b7751fd6521d0e27

The temperature dependence of the line separation Δω of the low-frequency shoulder from the position of the T2g(1) peak. Inset marks the shoulder appearing on the low-frequency side of the T2g(1) peak.

2012.13310

fig-line-seperation.jpg

0.45594

09e1763f4f264e148e6e666bb4446dde

Atomic vibration patterns for all four IR-active phonon modes: (i) T1u(1), (ii) T1u(2), (iii) T1u(3), and (iv) T1u(4). The color coding of atoms is the same as in Fig. <ref>(a). These modes are listed here in the order of decreasing frequency (see Table <ref>).

2012.13310

figure_IR_eigenmodes_b.jpg

0.44845

84865a9ad3cf4a7d97d24771e671d49d

(a) Definition of all four magnetic-exchange interactions, first (J_1), second (J_2), and third (J_3 and J_3^') NN, considered in this work. Co atoms are shown in blue color. Ge and O atoms are omitted for clarity. Note that J_3^' passes through an intermediate Co atom (see text). (b) Fitting of the DFT (PBEsol+U) energy values computed for various different spin configurations in a doubled primitive cell, as shown in (a), with our model spin Hamiltonian. Here, we decide to choose the PBEsol+U method since it predicts better lattice parameters compared to the PBE+U predictions.

2012.13310

figure_Jij.jpg

0.441759

8e1cf7038be0464580323e9a6ce8f888

Atomic vibration patterns for all five Raman-active phonon modes: A1g, T2g(1), T2g(2), Eg, and T2g(3). These modes are listed here in the order of decreasing frequency (see Table <ref>).

2012.13310

figure_Raman_eigenmodes_b.jpg

0.425033

b936e9f6961143ed8a582b6d5820088c

The temperature dependence of Raman intensity (I×10^4), fullwidth at half maximum (Δ F), and Raman-peak position (RPP) for the A1g, T2g(1), T2g(2), and Eg Raman active modes. The lines connecting the data points are visual guides. The TN and TS mark the transition temperatures corresponding to the antiferromagnetic and the cubic-to-tetragonal structural phase transitions, respectively.

2012.13310

temp-dept-Raman-modes-21-Dec-2020.jpg

0.451178

2f89b841117745caa84f5583d6c0603e

Optimal proposal distribution corresponding to A.

2012.13311

opt_proposal_1.jpg

0.476322

8abdfde3886745699408cb6c279e3ec4

Overview of variational determinant estimation. A spherical proposal distribution q(s) is trained to estimate the absolute determinant of a matrix A.

2012.13311

overview.jpg

0.406881

d2c3520a3fd34dffab0c1203a8c6b3fc

Linear worst-case response to spanwise optimal perturbations with (α,β)=(0,2) and Re=3000.

2012.13314

a0b2_TEG.jpg

0.493095

573bc47d8f0d43f6a307b54e803e38fb

(α,β) = (0,2) transient energy growth of optimal disturbance in uncontrolled flow with various amplitudes of E_0.

2012.13314

a0b2_base_E.jpg

0.429239

c51d9b55912b4734900a5d5436f1cf5d

Transient energy growth of uncontrolled and controlled flows due to linear optimal perturbations of E_0 =1× 10^-4. Inserts correspond to iso-surfaces of 𝒬-criterion <cit.> (𝒬(h/u̅_c)^2=4× 10^-4) colored by streamwise vorticity ω_x h/u̅_c.

2012.13314

a0b2_flowfield.jpg

0.458152

d7d93c49a03b4753bbefdc4a87842108

DNS of the uncontrolled flow and the full-information LQR controlled flow with (α,β)=(0,2), Re=3000, and E_0=5× 10 ^-5. (Top) Friction velocity and (Bottom) normalized wall-normal velocity v(-h)/√(E_0) at the lower channel wall with x=0, z=0.

2012.13314

a0b2_lqr_op_friction_v.jpg

0.507097

2422c56174d74b7c82f0c881c1868bca

Comparison between uncontrolled flow (left) and LQR controlled flow (right) with (α,β) = (0,2), Re=3000. Modification of instantaneous streamwise velocity gradient in spanwise direction ∂u/∂z at tu̅_c/h=49, x/h=0.m

2012.13314

a0b2_op_lqr_shear49_update.jpg

0.365394

28fe49bb45474136b4a8b8addbe998e5

Robustness of LQR and SOF-LQR-s controllers to off-design Reynolds numbers. Both controllers are designed for Re=3000, but robustly reduce TEG from spanwise optimal perturbations over a range of off-design Reynolds numbers.

2012.13314

a0b2_robustness_toRe.jpg

0.418594

33023ae46d97481f80d1cf5badd6ec89

DNS of the uncontrolled flow and the SOF-LQR-s controlled flow with (α,β)=(0,2), Re=3000, and various perturbation amplitudes E_0. (Top) Friction velocity and (Bottom) normalized wall-normal velocity v(-h)/√(E_0) at the lower channel wall x=0, z=0.

2012.13314

a0b2_sof_op_friction_v.jpg

0.528878

715a882406b84a22adc12389a6f43388

Linear worst-case response to streamwise optimal perturbations with (α,β)=(1,0) and Re=3000.

2012.13314

a1b0_TEG.jpg

0.482637

cb174982d9ee46829ea2e087a86651b5

(α,β) = (1,0) transient energy growth of optimal disturbance in uncontrolled flow with various amplitudes of E_0.

2012.13314

a1b0_base_E.jpg

0.405074

a6c9fc0f46ea4f0cae60f60fb28dee22

Transient energy growth energy and friction velocity on walls of uncontrolled flow (E_0 =1× 10^-4). Inserts are corresponding iso-surface of 𝒬-criterion <cit.> (𝒬(h/u̅_c)^2=1× 10^-4) colored by streamwise vorticity ω_x h/u̅_c.

2012.13314

a1b0_flowfield.jpg

0.39393

0c72c54aed8a4cc9b6933bec57bf6e33

Robustness of LQR and SOF-LQR controllers to off-design Reynolds numbers. Both controllers are designed for Re=3000, but robustly reduce TEG from streamwise optimal perturbations over a range of off-design Reynolds numbers.

2012.13314

a1b0_robustness_toRe.jpg

0.481674

4b305b2aa28443a8a21b2e193d8edb22

Illustration of wall-based sensing and actuation in channel flow.

2012.13314

flow_configuration.jpg

0.387115

ab9d910bdce645569c76bd7c2e1f1b73

Plot with portfolio sizes 1 through 20.

2012.13315

big_portfolio.jpg

0.443927

a0ba4305988c4787b8020bfcb60332b5

Three partnership networks that have the same aggregate network. (a) Dynamic partnership network without concurrency. Each node is either isolated (k=0) or involved in a non-concurrent relationship (k=1) at each time. (b) Dynamic partnership network with concurrency. At two out of five times, a cycle graph is formed such that all nodes have degree 2. To compensate for a relatively large mean degree at these times, the network at the other three discrete times are the empty network, where every node is isolated. (c) Static partnership network with concurrency. The complete graph is presented. The weight of each edge is 1/5 of that of the edge shown in (a) and (b). The time-averaged network of the dynamic partnership networks shown in (a) and (b) is the complete network shown in (c).

2012.13317

concurrency-comparison.jpg

0.433534

20ec65f3b01f4be685a154f9743a890b

An illustration of the time course of the measures κ_1 and κ_3 (shown in the bottom panel) in a temporal network (shown by a timeline plot of the events in the top panel). Here we assume that the first event of a pair of nodes is the beginning of a “partnership” and that the last event is the end of it, i.e., we consider the momentary network at the given time.

2012.13317

kappas.jpg

0.486104

9ec6a679b708433a93a1a79e887c1b1d

A motivating example for the concurrency measure κ_3 (adapted from Ref. <cit.>). Panels (a) and (b) show two networks of currently active partnerships. From a disease spreading point of view, the situation in (a) is worse than the one in (b). This is more evident if one considers the corresponding line graphs in panels (c) and (d). The concurrency measure κ_1—basically the average degree of the networks—is the same between (a) and (b). However, the concurrency measure κ_3—basically the average degree of the line graphs shown in (c) and (d), which correspond to (a) and (b), respectively—correctly identifies the first situation as being more densely connected (and thus worse).

2012.13317

line_graph.jpg

0.431075

02e58960338d4110a34164c76b14f9ab

Schematic of temporal networks with three nodes that are different in the degree of concurrency. (a) Serial monogamy. (b) Concurrent partnerships. A vertical bar represents a time-stamped interaction event between a pair of nodes, neglecting the duration of the event. The thicker, shaded lines are examples of time-respecting paths from 1 to 3 and, only in (b), from 3 to 1. This figure is inspired by Fig. 1 of Ref. <cit.>.

2012.13317

schem-miller-concurrency.jpg

0.481954

65110d2b334d46d5a8c0295907ce58c9

Two ways of projecting time out of a temporal network. Panel (a) shows a temporal network. Panel (b) shows the network of nodes that have an event within the time window [t_0,t_1]. Panel (c) shows a network of concurrent partnerships—pairs of nodes that have an event before and after [t_0,t_1].

2012.13317

window_concurrent.jpg