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
0.439277	3b87ae925ad94ca8b1383b3a870bd305	(Color online)(a) Representation of the density of states (DOS) as a function of energy. (b) Dependence ρ_xx(V_g) at various temperatures in the absence of a magnetic field (vertical arrows indicate the gate voltage values that correspond to the valence band ceiling E_v and the conduction band bottom E_c). (c) Dependence of the magnetoresistance amplitude (ρ_xx(B)/ρ_xx(B=0)) in the gate voltage range near the energy band gap indicated by the dashed line (inset shows the region 3 < V_g < 7 V on an enlarged scale).	2012.13738	f18.jpg
0.445447	850e74ef10c64ddfaf3bf7a7127b7a64	(Color online) (a) Dependences ρ_xx(V_g) measured in various magnetic fields at T = 1.5 K. (b) Dependences ρ_xy(V_g) measured under the same conditions; horizontal dashed lines show the theoretical values h/(ν e^2) of the corresponding QHE plateaus. Dependences ρ_xx(B) for fixed values of V_g for the (c) hole and (d) electron regions.	2012.13738	f19.jpg
0.470904	8321c9445e1c446bbd6fded24c8d3ae2	(Color online) (a) Subband bottom energy as a function of the QW thickness in the range 5.5â11 nm (solid curves are for surface orientation (100), dashed curves for orientation (013)). (b) Dispersion law of bulk and edge states for HgTe QWs with orientation (013). (The figure is taken from [42].	2012.13738	f2.jpg
0.516741	41dd7dbb251c41e18cc1d8790ad9495a	(Color online)Number N of the oscillation minima determined from the dependences ρ_xx(B) shown in Fig. 19c, d as a function of 1/B for (a) hole and (b) electron regions. (c) Dependences N(1/B) at V_g = 3 V; solid black lines correspond to the best fit of the strong- and weak-field parts of the dependence; the arrow at the ordinate axis indicates the point of intersection with the vertical axis; dashed lines show possible fittings subject to the condition that the vertical axis is crossed at a point with an integer value. (d) Comparison of the electron N_s and hole P_s densities determined in various ways: by fitting the dependences ρ_xx(B) and ρ_xy(B) in the Drude model and by analyzing the position of the minima of SdH oscillations in strong magnetic fields; the density of Dirac electrons on the upper surface of is determined from an analysis of SdH oscillations in weak fields, and on the lower surface,as the difference between and.	2012.13738	f20.jpg
0.382315	546c0789ca3d4a1ba0f1374ad8648870	(Color online)Schematic representation of the band diagram of the system under study at various gate voltages (the voltage drop in the dielectric is shown not to scale). Dirac points for both surfaces denoted as and are located in the valence band. Flat bands correspond to a zero voltage at the gate. Application of a gate voltage leads to the bending of the bands; the distant surface (bottom) is partially screened by the nearer one (top) and bulk carriers. In the range 2 < V_g < 4V, the Fermi level is located in the band gap (e, f), while for V_g < 2 V, the system contains bulk holes (aâd), and for V_g > 4 V, bulk electrons (g)	2012.13738	f21.jpg
0.468638	65be107999dc4a14a0f885d2c724be6f	(Color online) (a) Section of the structure under study. (b) Electric circuit for connecting layers to a gate voltage source and the pattern of electric field lines. (c) Equivalent circuit of structure capacitance. VAC is the measuring variable signal, is the dielectric constant of the insulator layer between the gate and the upper surface of the HgTe film, is the thickness of this layer, ϵ_HgTe is the dielectric constant of HgTe, dHgTe is the thickness of the HgTe film, is the capacitance between the gate and the upper surface of the HgTe film, C_tb is the capacitance between the top and bottom surfaces of the HgTe film, D_t is the density of DF states on the upper surface, D_b is the density of DF states on the bottom surface.	2012.13738	f22.jpg
0.412251	97b018936e0c4d7cae8220f1d48b6c6c	(Color online)a) Dependences ρ_xx(V_g) and ρ_xy(V_g). (b) Capacitance as a function of the gate voltage in a zero magnetic field and in a 2 T field.	2012.13738	f23.jpg
0.408055	ae24a4f2afcb452380428e641cd823e3	(Color online)(a) Positions of the minima of magnetocapacitance and magnetoconductivity oscillations (dependence with a kink) in the inverse magnetic field, measured at fixed gate voltages V_g = 2, 3, ..., 6 V as a function of oscillation number n. Drawn through the experimental dependences are fitting straight lines extrapolated to the intersection with the horizontal axis. The point where the lines cross the axis corresponds to the oscillation phase δ associated with the Berry phase of Dirac electrons. (b) Phase δ obtained from the data displayed in (a) as a function of the gate voltage.	2012.13738	f24.jpg
0.388377	9cb2c47cddbb43d0bdf0b66dd092e2b4	(Color online) (a) Transistor Hall structure based on HgTe QW and HgTe films and (b) dependences, typical of this structure, of the dissipative and Hall resistance components on the gate voltage for a QW 8 nm in thickness.	2012.13738	f3.jpg
0.500639	c2baf540e4034b8d94d121fbf2d4b0f4	(Color online)(a) Two-dimensional topological insulator. (b) Two-dimensional conductor in the QHE regime when the ground Landau level alone is filled.	2012.13738	f4.jpg
0.39582	4cb8fcde1cf34b3faf5b2748ea81f28f	(Color online)(a) Two-dimensional conductor and distribution of currents in it.	2012.13738	f5.jpg
0.44984	3ba1eec11ed04280b23a57ea67636910	(Color online)(a) Local and nonlocal resistance in the diffusive regime.	2012.13738	f6.jpg
0.485754	5c1ee86419434443971d4f924d23c7e4	(Color online)(a)Equivalent circuit of a 2D TI in the edge transport regime in the case of a standard Hall bridge with two current and four potentiometric contacts. The edge channel is replaced with the equivalent resistance R_k = (h/e^2)(L_k/l), where L_k and l are the channel length and mean free path. Current inflows between the resistances R1 and R2 and outflows between R4 and R5; the voltage drop is measured on the resistance R6.	2012.13738	f7.jpg
0.437785	275a82bf5e154114beb029bae85c2d41	(Color online)Comparison of the dependences of (a) local and (b) nonlocal resistances on the density of charge carriers with the results of a model that takes edge and bulk scattering into account for various configurations of applied current and measured voltage. The calculated dependences are shown by thinner lines. The calculation parameters are specified in [46].	2012.13738	f8.jpg
0.424017	0d4fb568c3074a698e6f307174353f16	(Color online)Temperature dependence of the resistance of a sample based on an HgTe QW 8 nm in thickness at (a) high (100 > T > 4 K) and (b) low (4 > T > 0.2 K) temperatures.	2012.13738	f9.jpg
0.433388	41b367bcf76840c880c46a38094fc5b5	Lattice relaxation effects in twisted trilayer graphene. (a) and (d) the out-of-plane displacement, (b) and (e) the band structure with the inverse participation ratio (IPR) represented by colors, (c) and (f) the layer-projected weights of band eigenstates represented by the thickness of the lines in tTLG-AÃA-1.35 and tTLG-ÃAB-1.05, respectively. The hopping parameters are γ_0=3.2 eV and γ_1=0.48 eV.	2012.13741	Figure2_relax.jpg
0.417388	a2b9462f19f54003aabebfdb2c52f100	Calculated local density of states (LDOS) mappings of different energies in the tTLG. (a) The density of states (DOS) of tTLG-AÃA-1.35. (b) and (c) LDOS mappings of tTLG-AÃA-1.35 without and with lattice relaxations, respectively. Energies of the LDOS are selected near the Fermi energy in (a). (e) and (f) The LDOS mappings of tTLG-ÃAB-1.05 without and with relaxations, respectively. Energies of the LDOS pick up from the three peaks near the Fermi energy in (d). The hopping parameters are γ_0=3.2 eV and γ_1=0.48 eV.	2012.13741	figure3_map.jpg
0.414297	54db9bd74cb54d0f81f0d844ee8fb4d2	(a) and (b) Hofstadter's butterflies of relaxed tTLG-AÃA-1.35 and tTLG-ÃAB-1.05 with magnetic field less than 20 T, respectively.	2012.13741	figure5_B.jpg
0.444792	c09199b4a2b84e0cabcfeaccbca95b15	Variations of the offset energy ΔE in the relaxed tTLG-AÃA-1.35. (a) The band structure obtained with γ_0=3.2 eV and γ_1=0.32 eV. (b) The evolution of the ΔE with different γ_0 (fixed γ_1) and different γ_1 (fixed γ_0) illustrated in the blue and red lines, respectively. (c) The band structure calculated with only the nearest-neighbor intralayer hoppings. (d) The band structure calculated by assuming that the interlayer hoppings of A and B sublattices between two layers are reduced to zero.	2012.13741	figure_4_Dirac_point.jpg
0.449525	ea9d841ead494d499041153d7b8721e5	The side (upper panel) and top (lower panel) views of tTLG (a) tTLG-AÃA-6.01 and (b) tTLG-ÃAB-6.01. The high-symmetry stackings AAA, ABA, BAB, AAB, and ABC are highlighted by circles of red, black, blue, purple, and green colors, respectively. The number 6.01 stands for the twist angle 6.01^∘.	2012.13741	tTLG-structure.jpg
0.444071	dcb67876b2254d64a54f4f1dbc0e23b0	(a) The example of the Raman absorption measurements with amplitude-modulated Raman pump field, showing the signal peak at the Raman pump modulation frequency (300 kHz), when the Raman laser is in resonance with the two-photon transition to the 5S_1/2,F=2→5D_3/2,F”=1 transition, on top of the reduced noise floor due to balanced detection using intensity squeezed twin beams. (b) Measured absorption spectra mapping the 5D_3/2 structure as the Raman pump field frequency is tuned through the corresponding two-photon resonances for various powers of the coupling laser. The cell temperature 25^oC (atomic density 9.6· 10^9cm^-3).	2012.13745	FIG3-4zn.jpg
0.450375	1f76969b6b16420c84430672a2d7638a	Measured amplitude of the Raman absorption at the 5S_1/2,F=2 → 5D_3/2,F"=1 resonance peak as a function of the Raman pump power for different ^87Rb cell temperatures. The inset plot, showing the measured spectrum analyzer signal for different Raman pump intensities for the same cell temperature, indicated the location of the measured peak.	2012.13745	FIG4-4.jpg
0.403632	213b19a02f11480abfd6c6b084201d5b	Conceptual experimental setup of the quantum-enhanced spectroscopy experiments using two-mode intensity squeezing generated in ^85Rb vapor via four-wave mixing (FWM). An additional strong Raman pump provides access to the levels of interest in a sample either via ladder (a) or Λ (b) two-photon Raman transitions, and the resulting absorption resonances are detected in differential intensity of the squeezed twin beams. See text for the abbreviations	2012.13745	combinedsetup.jpg
0.444866	a4198f88972045f2bf7f1cb83ae6a9ff	(a) The two-photon transition (5S_1/2,F=2→ 5D_3/2,F”=1,2,3) of the probe field coupled to the Raman pump field. (b) Differential intensity noise trace from spectrum analyzer mapping the 5D_3/2 structure as the Raman pump field frequency is tuned through the corresponding two-photon resonances. The Raman pump power is 5 mW, the cell temperature 65^oC (atomic density 3.7· 10^11cm^-3).	2012.13745	figure2-3zn.jpg
0.399094	169165c3d7c24ddb9f7ad587383f5be0	Circuit QED analogue of our model with the energy levels of the atom (transmon) and the first two levels of the resonator modes denoted. The resonator mode at ω_c is two-photon detuned from ω_01 as shown. Interaction with another far-detuned readout mode is indicated at the detuning δ_r.	2012.13749	cQED.jpg
0.45749	57584bea58044e8fb9be852969573fd0	Deterministic Dicke state preparation involves the action of the unitary operators U_j,m_1 and U_j,m_2 on the computational state prepare Dicke states. The state after passing through the control-SWAP gate is projected onto \|μ⟩⟨μ\|⊗ I ⊗\|ζ⟩⟨ζ\|. The non-zero overlaps ⟨j,m_1\|ζ\|$⟩ and⟨j,m_2\|ζ\|$⟩ are used to define the state \|μ⟩ which produces the desired initial state as shown.	2012.13749	input.jpg
0.481964	16bf914d648143a88b9c8bdd643c9ab2	The time-evolved system-meter state \|Ψ(t)⟩ enters the circuit where \|μ⟩ and \|ζ⟩ are ancilla and reference states defined in the text. The final state is projected onto the joint quantum state \|ν⟩⊗\|j,m_1⟩⊗\|j,m_2⟩ which results in the postselection of the system in the state \|ψ_f⟩.	2012.13749	measurement.jpg
0.490894	ebd73a4acdac479f8183abf43203fef0	The 6 types of vertices in the six-vertex model together with the corresponding height function values on the four squares touching this vertex (we set h=0 on the upper-left square). Each type comes with a weight 𝐚_1,𝐚_2,𝐛_1,𝐛_2,𝐜_1,𝐜_2.	2012.13750	6vertex-config_with_height.jpg
0.52574	7cbba067e2434c0a8a9cedbbd1cc0c8b	A deterministic map from local 6-vertex configuration to oriented loops and paths. Vertices of types 5 and 6 could be split into non-crossing paths in two ways, but we always choose the left-turning splitting.	2012.13750	8-oriented_loops.jpg
0.482439	075c29606ea14d9c8c9a4185c7d548fd	An illustration for the proof of Lemma <ref>. In every domain of Q_x, y, exactly two boundary segments in γ_L^x ∪γ_R^x (in black) cross _y vertically; the left- and right-most such crossings are γ_L^x,y and γ_R^x,y, respectively. The boundary condition ξ on Q_x,y is {1, 2} on the black parts of ∂ Q_x, y, and their maximal extension which is smaller or equal to k on the gray parts. The blue paths form a fence: they separate Top_x,y from Bottom_x,y inside Q_x,y and have h ≤ (1-c_0)k.	2012.13750	Q_low.jpg
0.469832	7ad77df3de85421b939a68b92b411c63	Left: V_x ⋂ (_y - 1∪_y ∪_y + 1 ) may consist of several domains; U_x, y is the bottom-most traversing one. The curves ϑ_L^x,y and ϑ_R^x,y are highlighted by doubled lines. Right: U_x, y⋂_y-1 may consist of several domains; those that cross the slice vertically form Q_x, y-1. For the middle slice, Q_x,y is defined in the same way as U_x,y. It is separated inside V_x from the bottom and top of the cylinder by Q_x,y-1 and Q_x,y+1, respectively. The curves γ_L^x,y-1, γ_R^x,y-1, γ_L^x,y and γ_R^x,y are highlighted.	2012.13750	UQ.jpg
0.448012	99fd89eb07eb4a339a82a438cc5a4eaa	The last step in the proof of Theorem <ref> is, roughly speaking, based on the Bayes formula for the events of the horizontal crossing _h ≥ c k(R) and the two vertical crossings _h ≤ℓ+2(S_L) ∩_h ≤ℓ+2(S_R).	2012.13750	fig10.jpg
0.436945	d82315b446ca4813ab48d2bd0f88c4a4	An illustration of the setup of the proof of Theorem <ref>.	2012.13750	fig13.jpg
0.4908	ec5ca1ccc47b4f14a4ecad92af9d66aa	The event = Q is determined by the value of h on D∖ Q. For z as above, any circuit disconnecting ∂ D from Λ_n must cross the annulus A(n, 2n) + z from inside to outside. By duality, any circuit in _2n(z) surrounding _n(z) induces a crossing between ∂ Q and _n.	2012.13750	fig18.jpg
0.478495	1d28d084dd604435bcb1ababa61adb97	A geometric argument in the end of the proof of Proposition <ref>. The light-blue and dashed boundary segments are reflection symmetric. The small-height crossing between the light-blue boundary segments (in blue) excludes the large-height crossing between the dashed boundary segments (in dashed red). Solid-line boundaries represent here boundary conditions 0,1 and dashed boundaries represent their maximal extensions smaller or equal to k.	2012.13750	fig6_alternative.jpg
0.515443	03b3e607e80e4d2381c912b3ab59986d	Illustration for the proof of Lemma <ref>. The curves γ_L and γ_R (blue) and the boundary segments of the strip between them define the quad D. Solid-line boundaries represent here boundary conditions ζ∈{ 0,1 }, and dashed lines represent maximal extensions of boundary conditions remaining at most k.	2012.13750	fig7_alternative.jpg
0.468081	8388a75ff9be482b809107a572786a9a	An illustration for the proof of Lemma <ref>. The red paths γ_L and γ_R have \|h\|≥ k and are measurable in terms of the value of \|h\| on them and on the faces to their left and right, respectively.	2012.13750	fig9_alternative.jpg
0.49844	04c4716d217c4393970606b6bc6a6161	A discrete domain D generated by the marked vertices. The boundary ∂ D is represented in grey and some of the arrows of the associated 6-vertex configuration are shown. The boundary condition is such that there exists a unique height function inside D consistent with it (we say that the height function is frozen). If the red square were to contain a 9 instead of a 7, the boundary condition would become non-admissible.	2012.13750	frozen.jpg
0.487001	ca448aa8494b491fbad93cc00265a84c	The construction of 𝐓 by reversing the orientation of all the loops and paths (and hence the arrows of 6-vertex configuration) that are between the paths γ_i^* and γ_L+i^* (grey region) or on the former.	2012.13750	loop-reversal.jpg
0.42637	6e579a80fbe14cb78d88de86ef6f9041	A quad (with marked faces at its corners) which is symmetric with respect to the reflection σ along the diagonal. The boundary condition ξ is such that -ξ≼σξ, as required in Lemma <ref>. This domain contains no path of non-negative height from top to bottom, but contains a negative ×-crossing from left to right. Such ×-crossings are not necessarily unique; one may be found by following the contour of the connected component of non-negative faces of the bottom arc.	2012.13750	symmetric_dom.jpg
0.46111	167e3e21c0b3476195e1863b79ea559b	The dashed lines represent boundary conditions ck, ck-1, while the thick ones are boundary conditions k,k-1 (up to interpolations at the ends of the intervals). The segments at the bottom and top have length δ n. Left: the domain D formed of faces contained between γ_L and γ_R. Middle: the domain D is extended into D” by pushing away the (small) boundary conditions ck,ck-1. The probability of existence of a crossing from I_0 to Ĩ_0 of large height increases. Right: the domain D” is shrunk to D' by pulling closer the (large) boundary conditions k,k-1. The probability of existence of a crossing from γ_1 to B' of low height decreases.	2012.13750	tricky_2.jpg
0.463393	9cac04e4fc6f48fe85c777f5da53b416	Minimal equilibrium price and the intrinsic value as a function of D for the initial values in (<ref>)	2012.13753	Figure1_re.jpg
0.466269	09bb8814b226480fa93f0b0a8124c757	Relative size of the bubble for the parameters in (<ref>)	2012.13753	Figure2_re.jpg
0.511236	bff1cb0bb2814bf4bd36948355e3ee6c	Relative size of the bubble for the parameters in (<ref>)	2012.13753	Rexample1.jpg
0.505272	a7750fd8cf99402b8b8f2009fb7397eb	Relative size of the bubble for the parameters in (<ref>)	2012.13753	Rsametheta.jpg
0.470363	09f286496c2c4556bc4c5ff5759689c6	Minimal equilibrium price and the intrinsic value as a function of D for the initial values in (<ref>)	2012.13753	example1.jpg
0.462394	e5386340b9a049dfadbc73c72b9b0c4a	Minimal equilibrium price and the intrinsic value as a function of D for the parameters in (<ref>)	2012.13753	sametheta.jpg
0.471827	07580e2e709148ceaebf63e5863fda6a	Two different ways of identifying points across a superposition of classical manifolds. Using projector Π̂_1 = \|x_0^(1)⟩⟨x_0^(1)\| + \|y_1^(2)⟩⟨y_1^(2)\| + \|z_1^(3)⟩⟨z_1^(3)\|, point x_0 in the first manifold ℳ_1 is identified with point y_1 in the manifold ℳ_2 and with point z_1 in the manifold ℳ_3. With the second projector Π̂_2 = \|x_0^(1)⟩⟨x_0^(1)\| + \|y_2^(2)⟩⟨y_2^(2)\| + \|z_2^(3)⟩⟨z_2^(3)\|, point x_0 in the first manifold ℳ_1 is identified with point y_2 in the manifold ℳ_2 and with point z_2 in the manifold ℳ_3. The different physical identification of points corresponds to a different choice of measurement apparatus, where the projector is a mathematical abstraction representing the position of a measurement apparatus.	2012.13754	projectors.jpg
0.468977	15357086bb50460cb7b607acd03221d1	We formally describe the nonclassical structure of spacetime as a superposition of classical backgrounds enumerated as a set of manifolds ℳ = {ℳ_i}_i=1,⋯, N (with N=3 in the figure). On each classical manifold ℳ_i, we choose a coordinate system. Different points X^(i), ∀ i=1, ⋯, N belonging to different manifolds are operationally identified with a “physical point” in the superposition of manifolds if a quantum system living in a superposition of those manifolds can be localised at those points X^(i). In the figure, we represent with the shaded area the subregion of each classical manifold where the wavefunction of such quantum system has support. In the idealised case, the support of the quantum system reduces to a single point (red point in the figure) in every classical manifold. In the case considered, to each classical manifold ℳ_i there corresponds a classical state of the gravitational field. In general, we do not know how to assign values to such gravitational field on a superposition of manifolds. We describe the gravitational field by correlating the spacetime point at which the field is defined with a quantum system, and we write this operation as ∑_i \|g_i▹ϕ_i⟩. This identification is natural, because tests of the gravitational field can be performed either directly, by measuring the gravitational radiation, or indirectly, as we choose here, via a probe particle. In the latter case, we can perform a measurement on the physical system to learn information about the gravitational field in the superposition of manifolds.	2012.13754	superposition_of_manifolds.jpg
0.440387	56c4f0b5c3e747c693313440f2e92c2b	Algorithm Flowchart. The left part shows the hight level overview of our proposed architecture, and the right part illustrates the details of each neural network module. At frame t, We use a 3D object detector and extract the LiDAR and image features for each detected object. These features are fused by the Feature Fusion Module. The fused features of detections from time t and tracks from time t-1 are used in the Trainable Distance Combination Module to learn the combination of the deep feature distance and the Mahalanobis distance. We apply a greedy match algorithm on the combined distance for data association. Matched pairs are further processed by the Kalman Filter to refine the final object states. The Track Initialization Module determines whether to initialize a new track for each unmatched detection.	2012.13755	architecture_2020_1031_0749.jpg
0.418527	4f9c659ba02b4b08869617544f11be9c		2012.13755	center_point_201_motorcycle_random_new_15_70_upper_right_crop.jpg
0.510605	8e686a18783041039ba50ab1e59ede59		2012.13755	detection_201_motorcycle_upper_right_crop.jpg
0.404204	2d9b8e3ec04841eea1b50d2843266ad0		2012.13755	gt_201_motorcycle_random_new_15_70_upper_right_crop.jpg
0.43341	75fa476d040d42a6b0bdbcd7c26716aa		2012.13755	ours_201_motorcycle_random_new_15_70_upper_right_crop.jpg
0.444715	67f967eea8954667af06e10c494998e0		2012.13755	scene_201b7c65a61f4bc1a2333ea90ba9a932_frame_1_channel_CAM_BACK_RIGHT_visual_debug.jpg
0.393954	56b2aad9ddfc42398ccc183d6c22b2cc		2012.13755	scene_201b7c65a61f4bc1a2333ea90ba9a932_frame_28_channel_CAM_BACK_visual_debug.jpg
0.414431	bf624c579c5e4c26b895a0dbc67f87b4		2012.13755	scene_201b7c65a61f4bc1a2333ea90ba9a932_frame_29_channel_CAM_BACK_visual_debug.jpg
0.423696	bd838a77e78c4c2d84d12006f321e8b8		2012.13755	scene_201b7c65a61f4bc1a2333ea90ba9a932_frame_2_channel_CAM_BACK_RIGHT_visual_debug.jpg
0.464004	fab7b4f09c184d6ab4508ef9ed34ff18	2-nd quant. level	2012.13762	2nd_quant.jpg
0.437907	516037fd5e5840468fd565b6e24802ea	3-rd quant. level	2012.13762	3rd_quant.jpg
0.412434	0023b68a18c341e189891a513754bf3e	4-th quant. level	2012.13762	4th_quant.jpg
0.424918	3051dfe32349419fb625fe32a887c63c	MSE relative changes	2012.13762	MSE_relative_change.jpg
0.408883	7ed5c93a9b7e4fc9888ce57df89862d8	Test error	2012.13762	error_resnet_cifar100.jpg
0.465555	bb15d1f2774142f388c6c0aeb3b3bb40	Cross-entropy loss	2012.13762	loss_resnet_cifar100.jpg
0.485123	d80bcfe6e69f442e81c8436f0eb86039	CBLAS (8 threads)	2012.13762	speedup_blas.jpg
0.44153	2f3dc7df66d24500b31f41ec27d7d33b	1 thread w/o SIMD	2012.13762	speedup_cpp.jpg
0.476919	a6c5a0cad9b348c99edbec5c3e502e57	1 thread with SIMD	2012.13762	speedup_simd.jpg
0.448006	4c3b203b9448421ab257bece4ce44178	Visualization for the effect of re-weighting without (dashed line) and with (solid line) linear scheduling (LS) from the perspective of the optimizer. It is shown by multiplying a learning rate by re-weighting factors that purple, red, green, yellow and blue stand for 2.0, 1.5, 1.0, 0.5 and 0.0, respectively. For simplicity, cosine annealing with initial value = 1 is used as a learning rate scheduler.	2012.13763	Figure_1.jpg
0.463195	5cf51369a0a04eb5bc47227b2d43a7a5	The different schedules for re-weighting when weighting factor is 2.	2012.13763	Figure_2.jpg
0.417831	970ca64b75f1451ca2721bac6bc2db36	Reconstruction of a line segment from pairwise distances. The average distance error is 9 · 10^-4.	2012.13770	segment_n=100_d=2_eps=02.jpg
0.38484	7d02c7c70cfe47bc9d17eda1ffa1a294	Examples of 2×2 squares of the covering. On the left the case j_1,j_2>0, on the right j_1>0, j_2<0.	2012.13772	Fig7.jpg
0.440781	4078a9e9589e469eb0371babafa6d957	The point (a_h,2h+1)∈ A(ℓ) (black circle on the left). The point (b_h+1,2h)∈ A(ℓ_m+1) (black circle on the right). The gray points represent ×{2h}.	2012.13772	a_e_b.jpg
0.465989	0ca73f4df6ca438d86af847d0c062c5f	The point (a_h,2h+1)∈ A(ℓ_m) (black circle on the left). The point (b_h+1,2h)∈ A(ℓ_m) (black circle on the right).	2012.13772	a_e_b_slant.jpg
0.44045	3ff53ee6675041e6ae66d083f96beeda	Matching of the coverings outside a pair of adjacent discrete edges.	2012.13772	compatibilita.jpg
0.565755	7ec620c583254701a5664ee575d8e2d5	The lattice points involved in (<ref>). The black dots are points of , circles are points of . The energy contribution of the even checkerboard in the white regions is negative.	2012.13772	energia-top.jpg
0.515825	0b0f09f5afd1409b8e5781fc1738d027	On the left the set E_, on the right we exhibit a set E_^δ satisfying (i) and (ii).	2012.13772	fig-thm5.jpg
0.521152	0103cdb80db6409c92684d884d0c6967	A discrete vertex of E may be a boundary point (not a vertex) of (Z(E)).	2012.13772	lati_discreti_poligono1.jpg
0.47784	43642cdd7caa45d380cb7df6722bcd35	An example of a discrete vertex of E contained in the interior of (Z(E)).	2012.13772	lati_discreti_poligono2.jpg
0.442906	634444aea5d646cfa09fe969df350214	The discrete edge ℓ represented with black dots does not satisfy the compatibility condition (<ref>). The red dots denote the points of A(ℓ) which does not comply with (<ref>).	2012.13772	non-monotone.jpg
0.41062	50e6a44794104a31959ce577b8825c73	An example of the situation described in Remark <ref> in the case s(ℓ)≤1/3. The lighter dots represent the points of inside A(ℓ^-), A(ℓ) and A(ℓ^+).	2012.13772	remk-tech-ass-extra.jpg
0.48867	2974e1610a2a4cd882f6d60884cae183	An example of the situation described in Remark <ref> in the case s(ℓ)<1/3. The lighter dots represent those points of lying inside A(ℓ^-), A(ℓ) and A(ℓ^+).	2012.13772	remk-tech-ass2-extra.jpg
0.490367	a223a8de7bdb4d8fbd250968860a6dc3	Example of covering outside ℓ_m_1 in the case (i).	2012.13772	ricoprimenti-m1-i.jpg
0.520687	7fe88a37e4d241eba3a097e1cfe6b03a	The covering considered in (iii) in the case L even.	2012.13772	ricoprimenti-m1-iii-ev.jpg
0.49591	f2c89591322d40a486985c613db89644	Example of C(), ∈(ℓ_0). The black dots represent the lattice points (± b_h,2h).	2012.13772	ricoprimenti-top-punta.jpg
0.398964	6ec2529f5e6244c38dbef420ec52b08a	Maximum eigenvalue λ_ max=λ^(M)_ max of the M-body density matrix ρ^(M) for M≤ 4 in the state (<ref>), as a function of the number of pairs k, for sp space dimension D=30 (all labels and quantities plotted are dimensionless).	2012.13785	fig1.jpg
0.385162	6b5e8dc92580465490e941005680d069	Illustration of the IRS-assisted K-user interference channel. Direct links are shown by dashed black arrows, and links between the IRS and transmitters or receivers are shown by continuous black arrows. The black rectangle represents a blockage of the direct links.	2012.13787	Fig1.jpg
0.414349	96c6131130704f208639451b51338445	Comparison of the lower and upper bounds on the sum DoF of a 3-user interference channel in the presence of a passive IRS and the sum DoF without an IRS.	2012.13787	Fig3.jpg
0.399818	de0dd119198b40f3ace68f40701fb2f1	Comparison of the lower bound on the sum DoF of a 3-user interference channel in the presence of passive and 0.9-relaxed passive lossless IRS, the sum DoF without an IRS, and the maximum DoF K.	2012.13787	Kcomparison3.jpg
0.415474	6d1b47cb71e54061a79f609d8f1146e5	Comparison of bounds on the sum DoF of a 3-user interference channel in the presence of an active IRS and IR, respectively, and the sum DoF without an IRS.	2012.13787	k=3_active3.jpg
0.529811	4275b8f9e0514024992c84dd6e5518b5	Comparison of the lower bound on the sum DoF of a 3-user interference channel in the presence of 0.9,0.8, and 0.7-relaxed passive lossless IRSs and the sum DoF without an IRS.	2012.13787	k=3_e_passive4.jpg
0.507431	9c3c56307daf43ca9f284eb3c49dea63	Reference frame Eq. (<ref>) in de-Sitter's universe. Chainlike lines show apparent horizons, solid lines correspond to world lines (g_22=const). R- and T_+-regions are shown.	2012.13788	wh3_fig2.jpg
0.386834	f955176817cc4385815e18ed6e6329e6	Evolution of I(t) and D(t) with r=3 (red line), r=1 after t=20 days (green line) and r=0.8 after t=30 days (blue line). Initial conditions: E(0)=10. I(0)=0, N(0)=6.e7; p=0.01.	2012.13792	fig1.jpg
0.417861	b5b7874af63648029f60144ea6db18e4	Simulation with r=1.5 and p=0.01. Initial conditions: E(0)=10, I(0)=0, N=6· 10^7. I/N (a) and C/N (b) vs. time from the numerical solution (solid black line) and from the analytic expressions, Eqs. (<ref>) and (<ref>) (dashed blue line). The time t is in units of days and τ_E=3.69 days, τ_I=3.48 days.	2012.13792	fig2.jpg
0.50127	6eb980faefe047eb92090c2c26337472	Plot of s vs. z for r=1.5,2,2.5,3.0, from Eq.(<ref>).	2012.13792	fig3.jpg
0.420796	02a57e3a130346d4b2f8249dfbfe514b	(a): Plot of C_∞/N vs. r, from the numerical solution of Eqs.(<ref>)-(<ref>) (dashed line) and from the analytical result of Eq.(<ref>) (continuous line). The dotted red line is the threshold value k=1-1/r. (b) Peak value of I/N vs. r from the numerical solution of Eqs.(<ref>)-(<ref>) (dashed line) and from Eq.(<ref>) (continuous line)	2012.13792	fig4.jpg
0.498079	a0ebeb5cd922446a97a634fb2607549f	Plot of peak time t_peak (in units of days) vs. r for initial values of E(0)=10 and I(0)=0, N(0)=6· 10^7 and τ_E=3.69 days, τ_I=3.48 days.	2012.13792	fig5.jpg
0.447295	b9cf6a48ce484aa89f40d42507766add	The correspondence between faces at the white vertex w (left) and ovals bounded by Γ_w (right) under the assumption that the curve is constructed reflecting the graph w.r.t. a vertical ray. Objects paired by the duality relation between the graph and the curve share the same color: vertices correspond to rational components, edges to double points, faces to ovals.	2012.13797	Corr_V_G.jpg
0.443164	914d8b21c8a04a31875af4bc5a8775aa	The local winding number at an ordered pair of oriented edges and the intersection number of gauge rays starting at the boundary sources with an oriented edge depend on the choice of gauge ray direction 𝔩.	2012.13797	Example_Gr_26_wind_int.jpg
0.469591	dc7e1ef4e237491f88de94b12c564557	A Kasteleyn signature for the reduced planar bipartite graph in the disk of Figure <ref>. Labels of the boundary vertices have been changed to be consistent with the notation of Definition <ref>.	2012.13797	Example_Gr_36.jpg
0.507995	d5f4e5660297431bb6382c13e8abd495	Strands at internal and at boundary edges.	2012.13797	Strand.jpg

0.439277

3b87ae925ad94ca8b1383b3a870bd305

(Color online)(a) Representation of the density of states (DOS) as a function of energy. (b) Dependence ρ_xx(V_g) at various temperatures in the absence of a magnetic field (vertical arrows indicate the gate voltage values that correspond to the valence band ceiling E_v and the conduction band bottom E_c). (c) Dependence of the magnetoresistance amplitude (ρ_xx(B)/ρ_xx(B=0)) in the gate voltage range near the energy band gap indicated by the dashed line (inset shows the region 3 < V_g < 7 V on an enlarged scale).

2012.13738

f18.jpg

0.445447

850e74ef10c64ddfaf3bf7a7127b7a64

(Color online) (a) Dependences ρ_xx(V_g) measured in various magnetic fields at T = 1.5 K. (b) Dependences ρ_xy(V_g) measured under the same conditions; horizontal dashed lines show the theoretical values h/(ν e^2) of the corresponding QHE plateaus. Dependences ρ_xx(B) for fixed values of V_g for the (c) hole and (d) electron regions.

2012.13738

f19.jpg

0.470904

8321c9445e1c446bbd6fded24c8d3ae2

(Color online) (a) Subband bottom energy as a function of the QW thickness in the range 5.5â11 nm (solid curves are for surface orientation (100), dashed curves for orientation (013)). (b) Dispersion law of bulk and edge states for HgTe QWs with orientation (013). (The figure is taken from [42].

2012.13738

f2.jpg

0.516741

41dd7dbb251c41e18cc1d8790ad9495a

(Color online)Number N of the oscillation minima determined from the dependences ρ_xx(B) shown in Fig. 19c, d as a function of 1/B for (a) hole and (b) electron regions. (c) Dependences N(1/B) at V_g = 3 V; solid black lines correspond to the best fit of the strong- and weak-field parts of the dependence; the arrow at the ordinate axis indicates the point of intersection with the vertical axis; dashed lines show possible fittings subject to the condition that the vertical axis is crossed at a point with an integer value. (d) Comparison of the electron N_s and hole P_s densities determined in various ways: by fitting the dependences ρ_xx(B) and ρ_xy(B) in the Drude model and by analyzing the position of the minima of SdH oscillations in strong magnetic fields; the density of Dirac electrons on the upper surface of is determined from an analysis of SdH oscillations in weak fields, and on the lower surface,as the difference between and.

2012.13738

f20.jpg

0.382315

546c0789ca3d4a1ba0f1374ad8648870

(Color online)Schematic representation of the band diagram of the system under study at various gate voltages (the voltage drop in the dielectric is shown not to scale). Dirac points for both surfaces denoted as and are located in the valence band. Flat bands correspond to a zero voltage at the gate. Application of a gate voltage leads to the bending of the bands; the distant surface (bottom) is partially screened by the nearer one (top) and bulk carriers. In the range 2 < V_g < 4V, the Fermi level is located in the band gap (e, f), while for V_g < 2 V, the system contains bulk holes (aâd), and for V_g > 4 V, bulk electrons (g)

2012.13738

f21.jpg

0.468638

65be107999dc4a14a0f885d2c724be6f

(Color online) (a) Section of the structure under study. (b) Electric circuit for connecting layers to a gate voltage source and the pattern of electric field lines. (c) Equivalent circuit of structure capacitance. VAC is the measuring variable signal, is the dielectric constant of the insulator layer between the gate and the upper surface of the HgTe film, is the thickness of this layer, ϵ_HgTe is the dielectric constant of HgTe, dHgTe is the thickness of the HgTe film, is the capacitance between the gate and the upper surface of the HgTe film, C_tb is the capacitance between the top and bottom surfaces of the HgTe film, D_t is the density of DF states on the upper surface, D_b is the density of DF states on the bottom surface.

2012.13738

f22.jpg

0.412251

97b018936e0c4d7cae8220f1d48b6c6c

(Color online)a) Dependences ρ_xx(V_g) and ρ_xy(V_g). (b) Capacitance as a function of the gate voltage in a zero magnetic field and in a 2 T field.

2012.13738

f23.jpg

0.408055

ae24a4f2afcb452380428e641cd823e3

(Color online)(a) Positions of the minima of magnetocapacitance and magnetoconductivity oscillations (dependence with a kink) in the inverse magnetic field, measured at fixed gate voltages V_g = 2, 3, ..., 6 V as a function of oscillation number n. Drawn through the experimental dependences are fitting straight lines extrapolated to the intersection with the horizontal axis. The point where the lines cross the axis corresponds to the oscillation phase δ associated with the Berry phase of Dirac electrons. (b) Phase δ obtained from the data displayed in (a) as a function of the gate voltage.

2012.13738

f24.jpg

0.388377

9cb2c47cddbb43d0bdf0b66dd092e2b4

(Color online) (a) Transistor Hall structure based on HgTe QW and HgTe films and (b) dependences, typical of this structure, of the dissipative and Hall resistance components on the gate voltage for a QW 8 nm in thickness.

2012.13738

f3.jpg

0.500639

c2baf540e4034b8d94d121fbf2d4b0f4

(Color online)(a) Two-dimensional topological insulator. (b) Two-dimensional conductor in the QHE regime when the ground Landau level alone is filled.

2012.13738

f4.jpg

0.39582

4cb8fcde1cf34b3faf5b2748ea81f28f

(Color online)(a) Two-dimensional conductor and distribution of currents in it.

2012.13738

f5.jpg

0.44984

3ba1eec11ed04280b23a57ea67636910

(Color online)(a) Local and nonlocal resistance in the diffusive regime.

2012.13738

f6.jpg

0.485754

5c1ee86419434443971d4f924d23c7e4

(Color online)(a)Equivalent circuit of a 2D TI in the edge transport regime in the case of a standard Hall bridge with two current and four potentiometric contacts. The edge channel is replaced with the equivalent resistance R_k = (h/e^2)(L_k/l), where L_k and l are the channel length and mean free path. Current inflows between the resistances R1 and R2 and outflows between R4 and R5; the voltage drop is measured on the resistance R6.

2012.13738

f7.jpg

0.437785

275a82bf5e154114beb029bae85c2d41

(Color online)Comparison of the dependences of (a) local and (b) nonlocal resistances on the density of charge carriers with the results of a model that takes edge and bulk scattering into account for various configurations of applied current and measured voltage. The calculated dependences are shown by thinner lines. The calculation parameters are specified in [46].

2012.13738

f8.jpg

0.424017

0d4fb568c3074a698e6f307174353f16

(Color online)Temperature dependence of the resistance of a sample based on an HgTe QW 8 nm in thickness at (a) high (100 > T > 4 K) and (b) low (4 > T > 0.2 K) temperatures.

2012.13738

f9.jpg

0.433388

41b367bcf76840c880c46a38094fc5b5

Lattice relaxation effects in twisted trilayer graphene. (a) and (d) the out-of-plane displacement, (b) and (e) the band structure with the inverse participation ratio (IPR) represented by colors, (c) and (f) the layer-projected weights of band eigenstates represented by the thickness of the lines in tTLG-AÃA-1.35 and tTLG-ÃAB-1.05, respectively. The hopping parameters are γ_0=3.2 eV and γ_1=0.48 eV.

2012.13741

Figure2_relax.jpg

0.417388

a2b9462f19f54003aabebfdb2c52f100

Calculated local density of states (LDOS) mappings of different energies in the tTLG. (a) The density of states (DOS) of tTLG-AÃA-1.35. (b) and (c) LDOS mappings of tTLG-AÃA-1.35 without and with lattice relaxations, respectively. Energies of the LDOS are selected near the Fermi energy in (a). (e) and (f) The LDOS mappings of tTLG-ÃAB-1.05 without and with relaxations, respectively. Energies of the LDOS pick up from the three peaks near the Fermi energy in (d). The hopping parameters are γ_0=3.2 eV and γ_1=0.48 eV.

2012.13741

figure3_map.jpg

0.414297

54db9bd74cb54d0f81f0d844ee8fb4d2

(a) and (b) Hofstadter's butterflies of relaxed tTLG-AÃA-1.35 and tTLG-ÃAB-1.05 with magnetic field less than 20 T, respectively.

2012.13741

figure5_B.jpg

0.444792

c09199b4a2b84e0cabcfeaccbca95b15

Variations of the offset energy ΔE in the relaxed tTLG-AÃA-1.35. (a) The band structure obtained with γ_0=3.2 eV and γ_1=0.32 eV. (b) The evolution of the ΔE with different γ_0 (fixed γ_1) and different γ_1 (fixed γ_0) illustrated in the blue and red lines, respectively. (c) The band structure calculated with only the nearest-neighbor intralayer hoppings. (d) The band structure calculated by assuming that the interlayer hoppings of A and B sublattices between two layers are reduced to zero.

2012.13741

figure_4_Dirac_point.jpg

0.449525

ea9d841ead494d499041153d7b8721e5

The side (upper panel) and top (lower panel) views of tTLG (a) tTLG-AÃA-6.01 and (b) tTLG-ÃAB-6.01. The high-symmetry stackings AAA, ABA, BAB, AAB, and ABC are highlighted by circles of red, black, blue, purple, and green colors, respectively. The number 6.01 stands for the twist angle 6.01^∘.

2012.13741

tTLG-structure.jpg

0.444071

dcb67876b2254d64a54f4f1dbc0e23b0

(a) The example of the Raman absorption measurements with amplitude-modulated Raman pump field, showing the signal peak at the Raman pump modulation frequency (300 kHz), when the Raman laser is in resonance with the two-photon transition to the 5S_1/2,F=2→5D_3/2,F”=1 transition, on top of the reduced noise floor due to balanced detection using intensity squeezed twin beams. (b) Measured absorption spectra mapping the 5D_3/2 structure as the Raman pump field frequency is tuned through the corresponding two-photon resonances for various powers of the coupling laser. The cell temperature 25^oC (atomic density 9.6· 10^9cm^-3).

2012.13745

FIG3-4zn.jpg

0.450375

1f76969b6b16420c84430672a2d7638a

Measured amplitude of the Raman absorption at the 5S_1/2,F=2 → 5D_3/2,F"=1 resonance peak as a function of the Raman pump power for different ^87Rb cell temperatures. The inset plot, showing the measured spectrum analyzer signal for different Raman pump intensities for the same cell temperature, indicated the location of the measured peak.

2012.13745

FIG4-4.jpg

0.403632

213b19a02f11480abfd6c6b084201d5b

Conceptual experimental setup of the quantum-enhanced spectroscopy experiments using two-mode intensity squeezing generated in ^85Rb vapor via four-wave mixing (FWM). An additional strong Raman pump provides access to the levels of interest in a sample either via ladder (a) or Λ (b) two-photon Raman transitions, and the resulting absorption resonances are detected in differential intensity of the squeezed twin beams. See text for the abbreviations

2012.13745

combinedsetup.jpg

0.444866

a4198f88972045f2bf7f1cb83ae6a9ff

(a) The two-photon transition (5S_1/2,F=2→ 5D_3/2,F”=1,2,3) of the probe field coupled to the Raman pump field. (b) Differential intensity noise trace from spectrum analyzer mapping the 5D_3/2 structure as the Raman pump field frequency is tuned through the corresponding two-photon resonances. The Raman pump power is 5 mW, the cell temperature 65^oC (atomic density 3.7· 10^11cm^-3).

2012.13745

figure2-3zn.jpg

0.399094

169165c3d7c24ddb9f7ad587383f5be0

Circuit QED analogue of our model with the energy levels of the atom (transmon) and the first two levels of the resonator modes denoted. The resonator mode at ω_c is two-photon detuned from ω_01 as shown. Interaction with another far-detuned readout mode is indicated at the detuning δ_r.

2012.13749

cQED.jpg

0.45749

57584bea58044e8fb9be852969573fd0

Deterministic Dicke state preparation involves the action of the unitary operators U_j,m_1 and U_j,m_2 on the computational state prepare Dicke states. The state after passing through the control-SWAP gate is projected onto |μ⟩⟨μ|⊗ I ⊗|ζ⟩⟨ζ|. The non-zero overlaps ⟨j,m_1|ζ|$⟩ and⟨j,m_2|ζ|$⟩ are used to define the state |μ⟩ which produces the desired initial state as shown.

2012.13749

input.jpg

0.481964

16bf914d648143a88b9c8bdd643c9ab2

2012.13749

measurement.jpg

0.490894

ebd73a4acdac479f8183abf43203fef0

The 6 types of vertices in the six-vertex model together with the corresponding height function values on the four squares touching this vertex (we set h=0 on the upper-left square). Each type comes with a weight 𝐚_1,𝐚_2,𝐛_1,𝐛_2,𝐜_1,𝐜_2.

2012.13750

6vertex-config_with_height.jpg

0.52574

7cbba067e2434c0a8a9cedbbd1cc0c8b

A deterministic map from local 6-vertex configuration to oriented loops and paths. Vertices of types 5 and 6 could be split into non-crossing paths in two ways, but we always choose the left-turning splitting.

2012.13750

8-oriented_loops.jpg

0.482439

075c29606ea14d9c8c9a4185c7d548fd

An illustration for the proof of Lemma <ref>. In every domain of Q_x, y, exactly two boundary segments in γ_L^x ∪γ_R^x (in black) cross _y vertically; the left- and right-most such crossings are γ_L^x,y and γ_R^x,y, respectively. The boundary condition ξ on Q_x,y is {1, 2} on the black parts of ∂ Q_x, y, and their maximal extension which is smaller or equal to k on the gray parts. The blue paths form a fence: they separate Top_x,y from Bottom_x,y inside Q_x,y and have h ≤ (1-c_0)k.

2012.13750

Q_low.jpg

0.469832

7ad77df3de85421b939a68b92b411c63

Left: V_x ⋂ (_y - 1∪_y ∪_y + 1 ) may consist of several domains; U_x, y is the bottom-most traversing one. The curves ϑ_L^x,y and ϑ_R^x,y are highlighted by doubled lines. Right: U_x, y⋂_y-1 may consist of several domains; those that cross the slice vertically form Q_x, y-1. For the middle slice, Q_x,y is defined in the same way as U_x,y. It is separated inside V_x from the bottom and top of the cylinder by Q_x,y-1 and Q_x,y+1, respectively. The curves γ_L^x,y-1, γ_R^x,y-1, γ_L^x,y and γ_R^x,y are highlighted.

2012.13750

UQ.jpg

0.448012

99fd89eb07eb4a339a82a438cc5a4eaa

The last step in the proof of Theorem <ref> is, roughly speaking, based on the Bayes formula for the events of the horizontal crossing _h ≥ c k(R) and the two vertical crossings _h ≤ℓ+2(S_L) ∩_h ≤ℓ+2(S_R).

2012.13750

fig10.jpg

0.436945

d82315b446ca4813ab48d2bd0f88c4a4

An illustration of the setup of the proof of Theorem <ref>.

2012.13750

fig13.jpg

0.4908

ec5ca1ccc47b4f14a4ecad92af9d66aa

The event = Q is determined by the value of h on D∖ Q. For z as above, any circuit disconnecting ∂ D from Λ_n must cross the annulus A(n, 2n) + z from inside to outside. By duality, any circuit in _2n(z) surrounding _n(z) induces a crossing between ∂ Q and _n.

2012.13750

fig18.jpg

0.478495

1d28d084dd604435bcb1ababa61adb97

A geometric argument in the end of the proof of Proposition <ref>. The light-blue and dashed boundary segments are reflection symmetric. The small-height crossing between the light-blue boundary segments (in blue) excludes the large-height crossing between the dashed boundary segments (in dashed red). Solid-line boundaries represent here boundary conditions 0,1 and dashed boundaries represent their maximal extensions smaller or equal to k.

2012.13750

fig6_alternative.jpg

0.515443

03b3e607e80e4d2381c912b3ab59986d

Illustration for the proof of Lemma <ref>. The curves γ_L and γ_R (blue) and the boundary segments of the strip between them define the quad D. Solid-line boundaries represent here boundary conditions ζ∈{ 0,1 }, and dashed lines represent maximal extensions of boundary conditions remaining at most k.

2012.13750

fig7_alternative.jpg

0.468081

8388a75ff9be482b809107a572786a9a

An illustration for the proof of Lemma <ref>. The red paths γ_L and γ_R have |h|≥ k and are measurable in terms of the value of |h| on them and on the faces to their left and right, respectively.

2012.13750

fig9_alternative.jpg

0.49844

04c4716d217c4393970606b6bc6a6161

A discrete domain D generated by the marked vertices. The boundary ∂ D is represented in grey and some of the arrows of the associated 6-vertex configuration are shown. The boundary condition is such that there exists a unique height function inside D consistent with it (we say that the height function is frozen). If the red square were to contain a 9 instead of a 7, the boundary condition would become non-admissible.

2012.13750

frozen.jpg

0.487001

ca448aa8494b491fbad93cc00265a84c

The construction of 𝐓 by reversing the orientation of all the loops and paths (and hence the arrows of 6-vertex configuration) that are between the paths γ_i^* and γ_L+i^* (grey region) or on the former.

2012.13750

loop-reversal.jpg

0.42637

6e579a80fbe14cb78d88de86ef6f9041

A quad (with marked faces at its corners) which is symmetric with respect to the reflection σ along the diagonal. The boundary condition ξ is such that -ξ≼σξ, as required in Lemma <ref>. This domain contains no path of non-negative height from top to bottom, but contains a negative ×-crossing from left to right. Such ×-crossings are not necessarily unique; one may be found by following the contour of the connected component of non-negative faces of the bottom arc.

2012.13750

symmetric_dom.jpg

0.46111

167e3e21c0b3476195e1863b79ea559b

The dashed lines represent boundary conditions ck, ck-1, while the thick ones are boundary conditions k,k-1 (up to interpolations at the ends of the intervals). The segments at the bottom and top have length δ n. Left: the domain D formed of faces contained between γ_L and γ_R. Middle: the domain D is extended into D” by pushing away the (small) boundary conditions ck,ck-1. The probability of existence of a crossing from I_0 to Ĩ_0 of large height increases. Right: the domain D” is shrunk to D' by pulling closer the (large) boundary conditions k,k-1. The probability of existence of a crossing from γ_1 to B' of low height decreases.

2012.13750

tricky_2.jpg

0.463393

9cac04e4fc6f48fe85c777f5da53b416

Minimal equilibrium price and the intrinsic value as a function of D for the initial values in (<ref>)

2012.13753

Figure1_re.jpg

0.466269

09bb8814b226480fa93f0b0a8124c757

Relative size of the bubble for the parameters in (<ref>)

2012.13753

Figure2_re.jpg

0.511236

bff1cb0bb2814bf4bd36948355e3ee6c

Relative size of the bubble for the parameters in (<ref>)

2012.13753

Rexample1.jpg

0.505272

a7750fd8cf99402b8b8f2009fb7397eb

Relative size of the bubble for the parameters in (<ref>)

2012.13753

Rsametheta.jpg

0.470363

09f286496c2c4556bc4c5ff5759689c6

Minimal equilibrium price and the intrinsic value as a function of D for the initial values in (<ref>)

2012.13753

example1.jpg

0.462394

e5386340b9a049dfadbc73c72b9b0c4a

Minimal equilibrium price and the intrinsic value as a function of D for the parameters in (<ref>)

2012.13753

sametheta.jpg

0.471827

07580e2e709148ceaebf63e5863fda6a

Two different ways of identifying points across a superposition of classical manifolds. Using projector Π̂_1 = |x_0^(1)⟩⟨x_0^(1)| + |y_1^(2)⟩⟨y_1^(2)| + |z_1^(3)⟩⟨z_1^(3)|, point x_0 in the first manifold ℳ_1 is identified with point y_1 in the manifold ℳ_2 and with point z_1 in the manifold ℳ_3. With the second projector Π̂_2 = |x_0^(1)⟩⟨x_0^(1)| + |y_2^(2)⟩⟨y_2^(2)| + |z_2^(3)⟩⟨z_2^(3)|, point x_0 in the first manifold ℳ_1 is identified with point y_2 in the manifold ℳ_2 and with point z_2 in the manifold ℳ_3. The different physical identification of points corresponds to a different choice of measurement apparatus, where the projector is a mathematical abstraction representing the position of a measurement apparatus.

2012.13754

projectors.jpg

0.468977

15357086bb50460cb7b607acd03221d1

We formally describe the nonclassical structure of spacetime as a superposition of classical backgrounds enumerated as a set of manifolds ℳ = {ℳ_i}_i=1,⋯, N (with N=3 in the figure). On each classical manifold ℳ_i, we choose a coordinate system. Different points X^(i), ∀ i=1, ⋯, N belonging to different manifolds are operationally identified with a “physical point” in the superposition of manifolds if a quantum system living in a superposition of those manifolds can be localised at those points X^(i). In the figure, we represent with the shaded area the subregion of each classical manifold where the wavefunction of such quantum system has support. In the idealised case, the support of the quantum system reduces to a single point (red point in the figure) in every classical manifold. In the case considered, to each classical manifold ℳ_i there corresponds a classical state of the gravitational field. In general, we do not know how to assign values to such gravitational field on a superposition of manifolds. We describe the gravitational field by correlating the spacetime point at which the field is defined with a quantum system, and we write this operation as ∑_i |g_i▹ϕ_i⟩. This identification is natural, because tests of the gravitational field can be performed either directly, by measuring the gravitational radiation, or indirectly, as we choose here, via a probe particle. In the latter case, we can perform a measurement on the physical system to learn information about the gravitational field in the superposition of manifolds.

2012.13754

superposition_of_manifolds.jpg

0.440387

56c4f0b5c3e747c693313440f2e92c2b

Algorithm Flowchart. The left part shows the hight level overview of our proposed architecture, and the right part illustrates the details of each neural network module. At frame t, We use a 3D object detector and extract the LiDAR and image features for each detected object. These features are fused by the Feature Fusion Module. The fused features of detections from time t and tracks from time t-1 are used in the Trainable Distance Combination Module to learn the combination of the deep feature distance and the Mahalanobis distance. We apply a greedy match algorithm on the combined distance for data association. Matched pairs are further processed by the Kalman Filter to refine the final object states. The Track Initialization Module determines whether to initialize a new track for each unmatched detection.

2012.13755

architecture_2020_1031_0749.jpg

0.418527

4f9c659ba02b4b08869617544f11be9c

2012.13755

center_point_201_motorcycle_random_new_15_70_upper_right_crop.jpg

0.510605

8e686a18783041039ba50ab1e59ede59

2012.13755

detection_201_motorcycle_upper_right_crop.jpg

0.404204

2d9b8e3ec04841eea1b50d2843266ad0

2012.13755

gt_201_motorcycle_random_new_15_70_upper_right_crop.jpg

0.43341

75fa476d040d42a6b0bdbcd7c26716aa

2012.13755

ours_201_motorcycle_random_new_15_70_upper_right_crop.jpg

0.444715

67f967eea8954667af06e10c494998e0

2012.13755

scene_201b7c65a61f4bc1a2333ea90ba9a932_frame_1_channel_CAM_BACK_RIGHT_visual_debug.jpg

0.393954

56b2aad9ddfc42398ccc183d6c22b2cc

2012.13755

scene_201b7c65a61f4bc1a2333ea90ba9a932_frame_28_channel_CAM_BACK_visual_debug.jpg

0.414431

bf624c579c5e4c26b895a0dbc67f87b4

2012.13755

scene_201b7c65a61f4bc1a2333ea90ba9a932_frame_29_channel_CAM_BACK_visual_debug.jpg

0.423696

bd838a77e78c4c2d84d12006f321e8b8

2012.13755

scene_201b7c65a61f4bc1a2333ea90ba9a932_frame_2_channel_CAM_BACK_RIGHT_visual_debug.jpg

0.464004

fab7b4f09c184d6ab4508ef9ed34ff18

2-nd quant. level

2012.13762

2nd_quant.jpg

0.437907

516037fd5e5840468fd565b6e24802ea

3-rd quant. level

2012.13762

3rd_quant.jpg

0.412434

0023b68a18c341e189891a513754bf3e

4-th quant. level

2012.13762

4th_quant.jpg

0.424918

3051dfe32349419fb625fe32a887c63c

MSE relative changes

2012.13762

MSE_relative_change.jpg

0.408883

7ed5c93a9b7e4fc9888ce57df89862d8

Test error

2012.13762

error_resnet_cifar100.jpg

0.465555

bb15d1f2774142f388c6c0aeb3b3bb40

Cross-entropy loss

2012.13762

loss_resnet_cifar100.jpg

0.485123

d80bcfe6e69f442e81c8436f0eb86039

CBLAS (8 threads)

2012.13762

speedup_blas.jpg

0.44153

2f3dc7df66d24500b31f41ec27d7d33b

1 thread w/o SIMD

2012.13762

speedup_cpp.jpg

0.476919

a6c5a0cad9b348c99edbec5c3e502e57

1 thread with SIMD

2012.13762

speedup_simd.jpg

0.448006

4c3b203b9448421ab257bece4ce44178

Visualization for the effect of re-weighting without (dashed line) and with (solid line) linear scheduling (LS) from the perspective of the optimizer. It is shown by multiplying a learning rate by re-weighting factors that purple, red, green, yellow and blue stand for 2.0, 1.5, 1.0, 0.5 and 0.0, respectively. For simplicity, cosine annealing with initial value = 1 is used as a learning rate scheduler.

2012.13763

Figure_1.jpg

0.463195

5cf51369a0a04eb5bc47227b2d43a7a5

The different schedules for re-weighting when weighting factor is 2.

2012.13763

Figure_2.jpg

0.417831

970ca64b75f1451ca2721bac6bc2db36

Reconstruction of a line segment from pairwise distances. The average distance error is 9 · 10^-4.

2012.13770

segment_n=100_d=2_eps=02.jpg

0.38484

7d02c7c70cfe47bc9d17eda1ffa1a294

Examples of 2×2 squares of the covering. On the left the case j_1,j_2>0, on the right j_1>0, j_2<0.

2012.13772

Fig7.jpg

0.440781

4078a9e9589e469eb0371babafa6d957

The point (a_h,2h+1)∈ A(ℓ) (black circle on the left). The point (b_h+1,2h)∈ A(ℓ_m+1) (black circle on the right). The gray points represent ×{2h}.

2012.13772

a_e_b.jpg

0.465989

0ca73f4df6ca438d86af847d0c062c5f

The point (a_h,2h+1)∈ A(ℓ_m) (black circle on the left). The point (b_h+1,2h)∈ A(ℓ_m) (black circle on the right).

2012.13772

a_e_b_slant.jpg

0.44045

3ff53ee6675041e6ae66d083f96beeda

Matching of the coverings outside a pair of adjacent discrete edges.

2012.13772

compatibilita.jpg

0.565755

7ec620c583254701a5664ee575d8e2d5

The lattice points involved in (<ref>). The black dots are points of , circles are points of . The energy contribution of the even checkerboard in the white regions is negative.

2012.13772

energia-top.jpg

0.515825

0b0f09f5afd1409b8e5781fc1738d027

On the left the set E_, on the right we exhibit a set E_^δ satisfying (i) and (ii).

2012.13772

fig-thm5.jpg

0.521152

0103cdb80db6409c92684d884d0c6967

A discrete vertex of E may be a boundary point (not a vertex) of (Z(E)).

2012.13772

lati_discreti_poligono1.jpg

0.47784

43642cdd7caa45d380cb7df6722bcd35

An example of a discrete vertex of E contained in the interior of (Z(E)).

2012.13772

lati_discreti_poligono2.jpg

0.442906

634444aea5d646cfa09fe969df350214

The discrete edge ℓ represented with black dots does not satisfy the compatibility condition (<ref>). The red dots denote the points of A(ℓ) which does not comply with (<ref>).

2012.13772

non-monotone.jpg

0.41062

50e6a44794104a31959ce577b8825c73

An example of the situation described in Remark <ref> in the case s(ℓ)≤1/3. The lighter dots represent the points of inside A(ℓ^-), A(ℓ) and A(ℓ^+).

2012.13772

remk-tech-ass-extra.jpg

0.48867

2974e1610a2a4cd882f6d60884cae183

An example of the situation described in Remark <ref> in the case s(ℓ)<1/3. The lighter dots represent those points of lying inside A(ℓ^-), A(ℓ) and A(ℓ^+).

2012.13772

remk-tech-ass2-extra.jpg

0.490367

a223a8de7bdb4d8fbd250968860a6dc3

Example of covering outside ℓ_m_1 in the case (i).

2012.13772

ricoprimenti-m1-i.jpg

0.520687

7fe88a37e4d241eba3a097e1cfe6b03a

The covering considered in (iii) in the case L even.

2012.13772

ricoprimenti-m1-iii-ev.jpg

0.49591

f2c89591322d40a486985c613db89644

Example of C(), ∈(ℓ_0). The black dots represent the lattice points (± b_h,2h).

2012.13772

ricoprimenti-top-punta.jpg

0.398964

6ec2529f5e6244c38dbef420ec52b08a

Maximum eigenvalue λ_ max=λ^(M)_ max of the M-body density matrix ρ^(M) for M≤ 4 in the state (<ref>), as a function of the number of pairs k, for sp space dimension D=30 (all labels and quantities plotted are dimensionless).

2012.13785

fig1.jpg

0.385162

6b5e8dc92580465490e941005680d069

Illustration of the IRS-assisted K-user interference channel. Direct links are shown by dashed black arrows, and links between the IRS and transmitters or receivers are shown by continuous black arrows. The black rectangle represents a blockage of the direct links.

2012.13787

Fig1.jpg

0.414349

96c6131130704f208639451b51338445

Comparison of the lower and upper bounds on the sum DoF of a 3-user interference channel in the presence of a passive IRS and the sum DoF without an IRS.

2012.13787

Fig3.jpg

0.399818

de0dd119198b40f3ace68f40701fb2f1

Comparison of the lower bound on the sum DoF of a 3-user interference channel in the presence of passive and 0.9-relaxed passive lossless IRS, the sum DoF without an IRS, and the maximum DoF K.

2012.13787

Kcomparison3.jpg

0.415474

6d1b47cb71e54061a79f609d8f1146e5

Comparison of bounds on the sum DoF of a 3-user interference channel in the presence of an active IRS and IR, respectively, and the sum DoF without an IRS.

2012.13787

k=3_active3.jpg

0.529811

4275b8f9e0514024992c84dd6e5518b5

Comparison of the lower bound on the sum DoF of a 3-user interference channel in the presence of 0.9,0.8, and 0.7-relaxed passive lossless IRSs and the sum DoF without an IRS.

2012.13787

k=3_e_passive4.jpg

0.507431

9c3c56307daf43ca9f284eb3c49dea63

Reference frame Eq. (<ref>) in de-Sitter's universe. Chainlike lines show apparent horizons, solid lines correspond to world lines (g_22=const). R- and T_+-regions are shown.

2012.13788

wh3_fig2.jpg

0.386834

f955176817cc4385815e18ed6e6329e6

Evolution of I(t) and D(t) with r=3 (red line), r=1 after t=20 days (green line) and r=0.8 after t=30 days (blue line). Initial conditions: E(0)=10. I(0)=0, N(0)=6.e7; p=0.01.

2012.13792

fig1.jpg

0.417861

b5b7874af63648029f60144ea6db18e4

Simulation with r=1.5 and p=0.01. Initial conditions: E(0)=10, I(0)=0, N=6· 10^7. I/N (a) and C/N (b) vs. time from the numerical solution (solid black line) and from the analytic expressions, Eqs. (<ref>) and (<ref>) (dashed blue line). The time t is in units of days and τ_E=3.69 days, τ_I=3.48 days.

2012.13792

fig2.jpg

0.50127

6eb980faefe047eb92090c2c26337472

Plot of s vs. z for r=1.5,2,2.5,3.0, from Eq.(<ref>).

2012.13792

fig3.jpg

0.420796

02a57e3a130346d4b2f8249dfbfe514b

(a): Plot of C_∞/N vs. r, from the numerical solution of Eqs.(<ref>)-(<ref>) (dashed line) and from the analytical result of Eq.(<ref>) (continuous line). The dotted red line is the threshold value k=1-1/r. (b) Peak value of I/N vs. r from the numerical solution of Eqs.(<ref>)-(<ref>) (dashed line) and from Eq.(<ref>) (continuous line)

2012.13792

fig4.jpg

0.498079

a0ebeb5cd922446a97a634fb2607549f

Plot of peak time t_peak (in units of days) vs. r for initial values of E(0)=10 and I(0)=0, N(0)=6· 10^7 and τ_E=3.69 days, τ_I=3.48 days.

2012.13792

fig5.jpg

0.447295

b9cf6a48ce484aa89f40d42507766add

The correspondence between faces at the white vertex w (left) and ovals bounded by Γ_w (right) under the assumption that the curve is constructed reflecting the graph w.r.t. a vertical ray. Objects paired by the duality relation between the graph and the curve share the same color: vertices correspond to rational components, edges to double points, faces to ovals.

2012.13797

Corr_V_G.jpg

0.443164

914d8b21c8a04a31875af4bc5a8775aa

The local winding number at an ordered pair of oriented edges and the intersection number of gauge rays starting at the boundary sources with an oriented edge depend on the choice of gauge ray direction 𝔩.

2012.13797

Example_Gr_26_wind_int.jpg

0.469591

dc7e1ef4e237491f88de94b12c564557

A Kasteleyn signature for the reduced planar bipartite graph in the disk of Figure <ref>. Labels of the boundary vertices have been changed to be consistent with the notation of Definition <ref>.

2012.13797

Example_Gr_36.jpg

0.507995

d5f4e5660297431bb6382c13e8abd495

Strands at internal and at boundary edges.

2012.13797

Strand.jpg