Search NASA⌕ Search

SEARCH · Search NASA

Results for “distributed clustering methods”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 757 records · Page 42

Critical magnetic Reynolds number of the turbulent dynamo in collisionless plasmas

The intracluster medium of galaxy clusters is an extremely hot and diffuse, nearly collisionless plasma, which hosts dynamically important magnetic fields of ~μG strength. Seed magnetic fields of much weaker strength of astrophysical or primordial origin can be present in the intracluster medium. In collisional plasmas, which can be approximated in the magnetohydrodynamical (MHD) limit, the turbulent dynamo mechanism can amplify weak seed fields to strong dynamical levels efficiently by converting turbulent kinetic energy into magnetic energy. However, the viability of this mechanism in weakly collisional or completely collisionless plasma is much less understood. In this study, we explore the properties of the collisionless turbulent dynamo using three-dimensional hybrid-kinetic particle-in-cell simulations. We explore the properties of the collisionless turbulent dynamo in the kinematic regime for different values of the magnetic Reynolds number, Rm, initial magnetic-to-kinetic energy ratio, (E mag /E kin ) i , and initial Larmor ratio, (r Larmor /L box ) i , i.e. the ratio of the Larmor radius to the size of the turbulent system. We find that in the ‘un-magnetized’ regime, (r Larmor /L box ) i > 1, the critical magnetic Reynolds number for the dynamo action Rm crit ≈ 107 ± 3. In the ‘magnetized’ regime, (r Larmor /L box ) i ≲ 1, we find a marginally higher Rm crit = 124 ± 8. We find that the growth rate of the magnetic energy does not depend on the strength of the seed magnetic field when the initial magnetization is fixed. We also study the distribution and evolution of the pressure anisotropy in the collisionless plasma and compare our results with the MHD turbulent dynamo.

79 ASTRONOMY AND ASTROPHYSICS↗

Parallel multiple-chain DRAM MCMC for large-scale geosteering inversion and uncertainty quantification

Geosteering is the proactive control of a wellbore placement based on the downhole measurements, aiming at maximizing economic production from the well. We report the recent development of azimuthal resistivity logging-whiledrilling (LWD) tools have a much larger depth of detection, thus sets higher demands for geosteering inversion as more unknown parameters need to be taken into account in the earth model to be inverted. For complicated nonlinear problems, traditional deterministic inversion methods are more likely to be trapped near local optima. In general, Markov chain Monte Carlo (MCMC) inversion methods are more capable of finding the global optimal solution and providing additional uncertainty analyses by sampling from the target distribution. However, MCMC methods usually have the problem of slow convergence. Even though optimization methods like delayed rejection (DR) and adaptive Metropolis (AM) were proposed to speed up the convergence, using MCMC methods to solve geosteering inversion problems may still incur unacceptable time cost. To reduce the sampling cost of MCMC methods, we use parallel multiple-chain DRAM MCMC methods to solve geosteering inverse problems and estimate the corresponding uncertainty. A clustering method, density-based spatial clustering of applications with noise (DBSCAN), is applied to select the best solution among many results. Lastly, the simulation results show that running many relatively short MCMC chains for one problem can obtain a similar result as running a long single chain. Besides, avoiding communications between multiple Markov chains during the sampling process yields almost linear scalability.

97 MATHEMATICS AND COMPUTING↗

Understanding ligand effects in gold clusters using mass spectrometry

This review summarizes recent research on the influence of phosphine ligands on the size, stability, and reactivity of gold clusters synthesized in solution. Sub-nanometer clusters exhibit size- and composition-dependent properties that are unique from those of larger nanoparticles. The highly tunable properties of clusters and their high surface-to-volume ratio make them promising candidates for a variety of technological applications. However, because “each-atom-counts” toward defining cluster properties it is critically important to develop robust synthesis methods to efficiently prepare clusters of predetermined size. For decades phosphines have been known to direct the size-selected synthesis of gold clusters. Despite the preparation of numerous species it is still not understood how different functional groups at phosphine centers affect the size and properties of gold clusters. Using electrospray ionization mass spectrometry (ESI-MS) it is possible to characterize the effect of ligand substitution on the distribution of clusters formed in solution at defined reaction conditions. In addition, ligand exchange reactions on preformed clusters may be monitored using ESI-MS. Collision induced dissociation (CID) may also be employed to obtain qualitative insight into the fragmentation of mixed ligand clusters and the relative binding energies of differently substituted phosphines. Quantitative ligand binding energies and cluster stability may be determined employing surface induced dissociation (SID) in a custom-built Fourier transform ion cyclotron resonance mass spectrometer (FT-ICR-MS). Rice-Ramsperger-Kassel-Marcus (RRKM) based modeling of the SID data allows dissociation energies and entropy values to be extracted that may be compared with the results of high-level theoretical calculations. The charge reduction and reactivity of atomically precise gold clusters, including partially ligated species generated in the gas-phase by in source CID, on well-defined surfaces may be explored using ion soft landing (SL) in a custom-built instrument combined with in situ time of flight secondary ion mass spectrometry (TOF-SIMS). Jointly, this multipronged experimental approach allows characterization of the full spectrum of relevant phenomena including cluster synthesis, ligand exchange, thermochemistry, surface immobilization, and reactivity. The fundamental insights obtained from this work will facilitate the directed synthesis of gold clusters with predetermined size and properties for specific applications.

Environmental Molecular Sciences Laboratory↗

A Selective Overview and Comparison of Robust Mixture Regression Estimators

Summary Mixture regression models have been widely used in business, marketing and social sciences to model mixed regression relationships arising from a clustered and thus heterogeneous population. The unknown mixture regression parameters are usually estimated by maximum likelihood estimators using the expectation–maximisation algorithm based on the normality assumption of component error density. However, it is well known that the normality‐based maximum likelihood estimation is very sensitive to outliers or heavy‐tailed error distributions. This paper aims to give a selective overview of the recently proposed robust mixture regression methods and compare their performance using simulation studies.

Yu, Chun↗

Q-BEEP: Quantum Bayesian Error Mitigation Employing Poisson Modeling over the Hamming Spectrum

Quantum computing technology has grown rapidly in recent years, with new technologies being explored, error rates being reduced, and quantum processor’s qubit capacity growing. However, near-term quantum algorithms are still unable to be induced without compounding consequential levels of noise, leading to non-trivial erroneous results. Quantum Error Correction (in-situ error mitigation) and Quantum Error Mitigation (post-induction error mitigation) are promising fields of research within the quantum algorithm scene, aiming to alleviate quantum errors, increasing the overall fidelity and hence the overall quality of circuit induction. Earlier this year, a pioneering work, namely HAMMER, published in ASPLOS-22 demonstrated the existence of a latent structure regarding post-circuit induction errors when mapping to the Hamming spectrum. However, they intuitively assumed that errors occur in local clusters, and that at higher average Hamming distances this structure falls away. In this work, we show that such a correlation structure is not only local but extends certain non-local clustering patterns which can be precisely described by a Poisson distribution model taking the input circuit, the device run time status (i.e., calibration statistics) and qubit topology into consideration. Using this quantum error characterizing model, we developed an iterative algorithm over the generated Bayesian network state-graph for post-induction error mitigation. Thanks to more precise modeling of the error distribution latent structure and the new iterative method, our Q-Beep approach provides state of the art performance and can boost circuit execution fidelity by up to 234.6% on Bernstein-Vazirani circuits and on average 71.0% on QAOA solution quality, using 16 practical IBMQ quantum processors. For other benchmarks such as those in QASMBench, the fidelity improvement is up to 17.8%. Q-Beep is a light-weight post-processing technique that can be performed offline and remotely, making it a useful tool for quantum vendors to integrate and provide more reliable circuit induction results.

Stein, Samuel A.↗

Imaging the Moon's Core with Seismology

Constraining the structure of the lunar core is necessary to improve our understanding of the present-day thermal structure of the interior and the history of a lunar dynamo, as well as the origin and thermal and compositional evolution of the Moon. We analyze Apollo deep moonquake seismograms using terrestrial array processing methods to search for the presence of reflected and converted energy from the lunar core. Although moonquake fault parameters are not constrained, we first explore a suite of theoretical focal spheres to verify that fault planes exist that can produce favorable core reflection amplitudes relative to direct up-going energy at the Apollo stations. Beginning with stacks of event seismograms from the known distribution of deep moonquake clusters, we apply a polarization filter to account for the effects of seismic scattering that (a) partitions energy away from expected components of ground motion, and (b) obscures all but the main P- and S-wave arrivals. The filtered traces are then shifted to the predicted arrival time of a core phase (e.g. PcP) and stacked to enhance subtle arrivals associated with the Moon s core. This combination of filtering and array processing is well suited for detecting deep lunar seismic reflections, since we do not expect scattered wave energy from near surface (or deeper) structure recorded at varying epicentral distances and stations from varying moonquakes at varying depths to stack coherently. Our results indicate the presence of a solid inner and fluid outer core, overlain by a partial-melt-containing boundary layer (Table 1). These layers are consistently observed among stacks from four classes of reflections: P-to-P, S-to-P, P-to-S, and S-to-S, and are consistent with current indirect geophysical estimates of core and deep mantle properties, including mass, moment of inertia, lunar laser ranging, and electromagnetic induction. Future refinements are expected following the successful launch of the GRAIL lunar orbiter and SELENE 2 lunar lander missions.

Weber, Renee C.↗

Vibrational spectra and structures of bare and Xe-tagged cationic Si{sub n}O{sub m}{sup +} clusters

Vibrational spectra of Xe-tagged cationic silicon oxide clusters Si{sub n}O{sub m}{sup +} with n = 3–5 and m = n, n ± 1 in the gas phase are obtained by resonant infrared multiple photon dissociation (IRMPD) spectroscopy and density functional theory calculations. The Si{sub n}O{sub m}{sup +} clusters are produced in a laser vaporization ion source and Xe complexes are formed after thermalization to 100 K. The clusters are subsequently irradiated with tunable light from an IR free electron laser and changes in the mass distribution yield size-specific IR spectra. The measured IRMPD spectra are compared to calculated linear IR absorption spectra leading to structural assignments. For several clusters, Xe complexation alters the energetic order of the Si{sub n}O{sub m}{sup +} isomers. Common structural motifs include the Si{sub 2}O{sub 2} rhombus, the Si{sub 3}O{sub 2} pentagon, and the Si{sub 3}O{sub 3} hexagon.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Introducing the SLICE Method for estimating pebble-bed reactor inventories at equilibrium operation with SCALE

This paper introduces the SCALE Leap-In method for Cores at Equilibrium (SLICE) for estimating pebble-bed reactor equilibrium core isotopic inventories using capabilities in the SCALE code system, requiring only a small computational cluster and a few days of computation. This method uses an iterative approach that relies on (1) a surrogate spectrum model that captures spatial and time-dependent spectral conditions, (2) a multi-pass model that captures the pebble’s evolving nuclide inventory as a function of location and time in the core, and (3) a full-core model that captures the core’s spatial neutron flux distribution. The SLICE approach is applied to a generic fluoride salt–cooled high-temperature reactor, demonstrating fuel inventory convergence through nuclide concentration inspection across iterations and comparisons for core realizations with varying discretizations. Results agree within ~5% with another state-of-the-art code, with differences attributed to input parameter or modeling assumption variations in the equilibrium generation methods.

22 GENERAL STUDIES OF NUCLEAR REACTORS↗

Distributed-Memory Parallel Symmetric Nonnegative Matrix Factorization

We develop the first distributed-memory parallel implementation of Symmetric Nonnegative Matrix Factorization (SymNMF), a key data analytics kernel for clustering and dimensionality reduction. Our implementation includes two different algorithms for SymNMF, which give comparable results in terms of time and accuracy. The first algorithm is a parallelization of an existing sequential approach that uses solvers for non symmetric NMF. The second algorithm is a novel approach based on the Gauss-Newton method. It exploits second-order information without incurring large computational and memory costs. We evaluate the scalability of our algorithms on the Summit system at Oak Ridge National Laboratory, scaling up to 128 nodes (4,096 cores) with 70% efficiency. Additionally, we demonstrate our software on an image segmentation task.

Eswar, Srinivas↗

Panels of HIV-1 Subtype C Env Reference Strains for Standardized Neutralization Assessments

In the search for effective immunologic interventions to prevent and treat HIV-1 infection, standardized reference reagents are a cost-effective way to maintain robustness and reproducibility among immunological assays. To support planned and ongoing studies where clade C predominates, here we describe three virus panels, chosen from 200 well-characterized clade C envelope (Env)-pseudotyped viruses from early infection. All 200 Envs were expressed as a single round of replication pseudoviruses and were tested to quantify neutralization titers by 16 broadly neutralizing antibodies (bnAbs) and sera from 30 subjects with chronic clade C infections. We selected large panels of 50 and 100 Envs either to characterize cross-reactive breadth for sera identified as having potent neutralization activity based on initial screening or to evaluate neutralization magnitude-breadth distributions of newly isolated antibodies. We identified these panels by downselection after hierarchical clustering of bnAb neutralization titers. The resulting panels represent the diversity of neutralization profiles throughout the range of virus sensitivities identified in the original panel of 200 viruses. A small 12-Env panel was chosen to screen sera from vaccine trials or natural-infection studies for neutralization responses. We considered panels selected by previously described methods but favored a computationally informed method that enabled selection of viruses representing diverse neutralization sensitivity patterns, given that we do not a priori know what the neutralization-response profile of vaccine sera will be relative to that of sera from infected individuals. The resulting 12-Env panel complements existing panels. Use of standardized panels enables direct comparisons of data from different trials and study sites testing HIV-1 clade C-specific products. HIV-1 group M includes nine clades and many recombinants. Clade C is the most common lineage, responsible for roughly half of current HIV-1 infections, and is a focus for vaccine design and testing. Standard reference reagents, particularly virus panels to study neutralization by antibodies, are crucial for developing cost-effective and yet rigorous and reproducible assays against diverse examples of this variable virus. We developed clade C-specific panels for use as standardized reagents to monitor complex polyclonal sera for neutralization activity and to characterize the potency and breadth of cross-reactive neutralization by monoclonal antibodies, whether engineered or isolated from infected individuals. We chose from 200 southern African, clade C envelope-pseudotyped viruses with neutralization titers against 16 broadly neutralizing antibodies and 30 sera from chronic clade C infections. We selected panels to represent the diversity of bnAb neutralization profiles and Env neutralization sensitivities. Finally, use of standard virus panels can facilitate comparison of results across studies and sites.

59 BASIC BIOLOGICAL SCIENCES↗

The interaction of RNA helicase DDX3 with HIV-1 Rev-CRM1-RanGTP complex during the HIV replication cycle

Molecular traffic between the nucleus and the cytoplasm is regulated by the nuclear pore complex (NPC), which acts as a highly selective channel perforating the nuclear envelope in eukaryotic cells. The human immunodeficiency virus (HIV) exploits the nucleocytoplasmic pathway to export its RNA transcripts across the NPC to the cytoplasm. Despite extensive study on the HIV life cycle and the many drugs developed to target this cycle, no current drugs have been successful in targeting the critical process of viral nuclear export, even though HIV’s reliance on a single host protein, CRM1, to export its unspliced and partially spliced RNA transcripts makes it a tempting target. Due to recent findings implicating a DEAD-box helicase, DDX3, in HIV replication and a member of the export complex, it has become an appealing target for anti-HIV drug inhibition. In the present research, we have applied a hybrid computational protocol to analyze protein-protein interactions in the HIV mRNA export cycle. This method is based on molecular docking followed by molecular dynamics simulation and accompanied by approximate free energy calculation (MM/GBSA), computational alanine scanning, clustering, and evolutionary analysis. We highlight here some of the most likely binding modes and interfacial residues between DDX3 and CRM1 both in the absence and presence of RanGTP. This work shows that although DDX3 can bind to free CRM1, addition of RanGTP leads to more concentrated distribution of binding modes and stronger binding between CRM1 and RanGTP.

59 BASIC BIOLOGICAL SCIENCES↗

Progressive CPU Volume Rendering with Sample Accumulation

Here, we present a new method for progressive volume rendering by accumulating object-space samples over successively rendered frames. Existing methods for progressive refinement either use image space methods or average pixels over frames, which can blur features or integrate incorrectly with respect to depth. Our approach stores samples along each ray, accumulates new samples each frame into a buffer, and progressively interleaves and integrates these samples. Though this process requires additional memory, it ensures interactivity and is well suited for CPU architectures with large memory and cache. This approach also extends well to distributed rendering in cluster environments. We implement this technique in Intel’s open source OSPRay CPU ray tracing framework and demonstrate that it is particularly useful for rendering volumetric data with costly sampling functions.

97 MATHEMATICS AND COMPUTING↗

Young Stellar Objects in the Massive Star-forming Regions W51 and W43

We present the results of our investigation of the star-forming complexes W51 and W43, two of the brightest in the first Galactic quadrant. In order to determine the young stellar object (YSO) populations in W51 and W43 we used color–magnitude relations based on Spitzer mid-infrared and 2MASS/UKIDSS near-infrared data. We identified 302 Class I YSOs and 1178 Class II/transition disk candidates in W51, and 917 Class I YSOs and 5187 Class II/transition disk candidates in W43. We also identified tens of groups of YSOs in both regions using the Minimal Spanning Tree (MST) method. We found similar cluster densities in both regions, even though Spitzer was not able to probe the densest part of W43. By using the Class II/I ratios, we traced the relative ages within the regions and, based on the morphology of the clusters, we argue that several sites of star formation are independent of one another in terms of their ages and physical conditions. We used spectral energy distribution-fitting to identify the massive YSO (MYSO) candidates since they play a vital role in the star formation process, and then examined them to see if they are related to any massive star formation tracers such as UCH ii regions, masers, or dense fragments. We identified 17 MYSO candidates in W51, and 14 in W43, respectively, and found that groups of YSOs hosting MYSO candidates are positionally associated with H ii regions in W51, though we do not see any MYSO candidates associated with previously identified massive dense fragments in W43.

79 ASTRONOMY AND ASTROPHYSICS↗

Electronic Spin Relaxation and Clustering in High-Pressure High-Temperature Synthesized Microcrystalline Diamond Particles with Reduced Nitrogen Content

The negatively charged nitrogen-vacancy (NV – ) color center in diamonds is widely studied because of numerous applications of this unique quantum system in sensing and quantum information sciences. While substitutional nitrogen is required to form the NV – centers in diamond, it also yields other paramagnetic defects─primarily the neutrally charged substitutional nitrogen centers (P1)─that decrease NV – spin coherence, which in turn degrades performance in applications. Herein, we investigate high-pressure high-temperature synthesized diamond microparticles (ca. 140–185 μm) having lower─ranging from 3 to 38 ppm─than the typical nitrogen content of type 1b diamond ( ca. 100 ppm and higher) typically used for the production of fluorescent diamond particles with NV – centers. A suite of electron paramagnetic resonance, optically detected magnetic resonance, and nuclear magnetic resonance methods are used to characterize spin properties of P1 and NV – centers in the particles. Upon decreasing the nitrogen content from 29 to 3 ppm, the ensemble NV – T 2 relaxation time increased by about 3-fold as measured directly in the Hahn Echo experiment at magnetic field of 1.2 T. Analysis of electronic relaxation of P1 centers revealed the existence of at least two distinct populations of P1 centers, consisting of fast and slower relaxing spins and allowed for an estimation of local concentrations. Even with <10 ppm nitrogen contents, the analysis indicated a highly heterogeneous distribution of P1 centers, suggesting the possibility of P1 spin clustering even at low nitrogen concentrations. The combined data demonstrate that the particles prepared from HPHT diamond with a low nitrogen content offer improved spin properties that are beneficial for NV – sensing applications.

Carbon↗

The geography of solar energy in the United States: Market definition, industry structure, and choice in solar PV adoption

The solar photovoltaic (PV) installation industry comprises thousands of firms around the world who collectively installed nearly 200 million panels in 2015. Spatial analysis of the emerging industry has received considerable attention from the literature, especially on the demand side concerning peer effects and adopter clustering. However this research area does not include similarly sophisticated spatial analysis on the supply side of the installation industry. The lack of understanding of the spatial structure of the PV installation industry leaves PV market research to rely on jurisdictional lines, such as counties, to define geographic PV markets. We develop an approach that uses the spatial distribution of installers' activity to define geographic boundaries for PV markets. Our method is useful for PV market research and applicable in the contexts of other industries. We use our approach to demonstrate that the PV industry in the United States is spatially heterogeneous. Despite the emergence of some national-scale PV installers, installers are largely local and installer communities are unique from one region to the next. The social implications of the spatial heterogeneity of the emerging PV industry involve improving understanding of issues such as market power, industry consolidation, and how much choice potential adopters have.

14 SOLAR ENERGY↗

Modeling Short-Range Order in Disordered Rocksalt Cathodes by Pair Distribution Function Analysis

Pair distribution function (PDF) analysis is a powerful technique for the characterization of short-range order (SRO) in disordered materials. Accurate interpretation of experimental PDF data is critically reliant on the development of structural models that can account for local variations in site occupancies and bond lengths. To this end, we outline an approach to model SRO using first-principles calculations based on the cluster-expansion formalism. These methods are validated on neutron scattering data from two disordered rocksalt oxyfluorides, Li 1.3 Mn 0.4 Ti 0.3 O 1.7 F 0.3 and Li 1.3 Mn 0.4 Nb 0.2 Ti 0.1 O 1.7 F 0.3 . For each composition, we demonstrate that an average structure without any SRO fails to reproduce several key features in the experimental PDF. To pinpoint the origin of the suspected SRO in these materials, configurational and displacive effects were separately investigated using two disparate models. Special quasi-random structures were relaxed using density functional theory to account for local changes in bond lengths while maintaining a near-random ionic configuration. This leads to slightly improved accuracy but still misrepresents asymmetry in the first few peaks of the PDF. Monte Carlo simulations were performed to model configurational SRO on a fixed lattice, which by itself is shown to have a minimal influence on the PDF. Instead, we find that it is the bond length relaxations within environments created by SRO which controls the details of the PDF, thereby highlighting the subtle but important coupling between configurational and displacive SRO in disordered materials.

36 MATERIALS SCIENCE↗

On the effect of Ga and In substitutions in the Ca 11 Bi 10 and Yb 11 Bi 10 bismuthides crystallizing in the tetragonal Ho 11 Ge 10 structure type

The Ga- and In-substituted bismuthides Ca 11 Ga x Bi 10– x , Ca 11 In x Bi 10– x , Yb 11 Ga x Bi 10– x , and Yb 11 In x Bi 10– x ( x < 2) can be readily synthesized employing molten Ga or In metals as fluxes. They crystallize in the tetragonal space group I 4/ mmm and adopt the Ho 11 Ge 10 structure type (Pearson code tI 84; Wyckoff sequence n 2 m j h 2 e 2 d ). The structural response to the substitution of Bi with smaller and electron-poorer In or Ga has been studied by single-crystal X-ray diffraction methods for the case of Ca 11 In x Bi 10– x [ x = 1.73 (2); octabismuth undecacalcium diindium]. The refinements show that the In atoms substitute Bi only at the 8 h site. The refined interatomic distances show an unconventional – for this structure type – bond-length distribution within the anionic sublattice. The latter can be viewed as consisting of isolated Bi 3- anions and [In 4 Bi 8 20- ] clusters for the idealized Ca 11 In 2 Bi 8 model. Formal electron counting and first-principle calculations show that the peculiar bonding in this compound drives the system toward an electron-precise state, thereby stabilizing the observed bond-length pattern.

Chemistry↗

Mobility Pattern Analysis for Power Restoration Activities Using Geo-Tagged Tweets

In this study, we analyzed mobility patterns of at-risk populations affected by an extreme event using geotagged tweets to geo-target power restoration efforts. Unlike other studies that have used tweets to facilitate emergency management activities, we used 1.5 million geotagged tweets generated during Hurricane Sandy (2012) to determine the mobility patterns and geospatial distribution of impacted populations who experienced power outage before, during and after the hurricane. We implemented a three-step analytical framework to: (i) analyze tweet contents with visual methods, including dendrograms, word clouds to identify common keywords pertaining to power outage; (ii) identify target users whose tweets contained information about power outages; and (iii) create a user-tweet locations matrix and an origin-destination matrix to examine clusters of target users and their mobility patterns. Preliminary results indicate that potential clusters were present in and around New York city, Philadelphia, Washington D.C. and Baltimore, which were used as potential evacuation destination cities after hurricane Sandy. The travel pattern and destination information can be used to (i) mobilize restoration efforts by utility companies and (ii) address resource allocation needs both in impacted and destination cities. Future work will focus on analyzing potential destinations for different origins and travel-time to identify evacuation routing patterns.

Kar, Bandana↗