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 451 records · Page 25

Contributions to the NUCLEI Project for Years 1-5(6)

The NUCLEI project, as defined by the scope of work, will develop, implement and run codes for large-scale computations of many topics in low-energy nuclear physics. Physics to be studied include the properties of nuclei and nuclear decays, nuclear structure and reactions, and the properties of nuclear matter. The computational techniques to be used include Quantum Monte Carlo, Configuration Interaction, Coupled Cluster, and Density Functional methods. The research program will emphasize areas of high interest to current and possible future DOE nuclear physics facilities, including ATLAS and FRIB (nuclear structure and reactions, and nuclear astrophysics), TJNAF (neutron distributions in nuclei, few body systems, and electroweak processes), NIF (thermonuclear reactions), MAJORANA and FNPB (neutrinoless double-beta decay and physics beyond the Standard Model), and LANSCE (fission studies).

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗

DISENTANGLING THE ICL WITH THE CHEFs: ABELL 2744 AS A CASE STUDY

Measurements of the intracluster light (ICL) are still prone to methodological ambiguities, and there are multiple techniques in the literature to address them, mostly based on the binding energy, the local density distribution, or the surface brightness. A common issue with these methods is the a priori assumption of a number of hypotheses on either the ICL morphology, its surface brightness level, or some properties of the brightest cluster galaxy (BCG). The discrepancy in the results is high, and numerical simulations just place a boundary on the ICL fraction in present-day galaxy clusters in the range 10%–50%. We developed a new algorithm based on the Chebyshev–Fourier functions to estimate the ICL fraction without relying on any a priori assumption about the physical or geometrical characteristics of the ICL. We are able to not only disentangle the ICL from the galactic luminosity but mark out the limits of the BCG from the ICL in a natural way. We test our technique with the recently released data of the cluster Abell 2744, observed by the Frontier Fields program. The complexity of this multiple merging cluster system and the formidable depth of these images make it a challenging test case to prove the efficiency of our algorithm. We found a final ICL fraction of 19.17 ± 2.87%, which is very consistent with numerical simulations.

79 ASTRONOMY AND ASTROPHYSICS↗

Physical properties of star clusters in the outer LMC as observed by the DES

The Large Magellanic Cloud (LMC) harbors a rich and diverse system of star clusters, whose ages, chemical abundances, and positions provide information about the LMC history of star formation. We use Science Verification imaging data from the Dark Energy Survey to increase the census of known star clusters in the outer LMC and to derive physical parameters for a large sample of such objects using a spatially and photometrically homogeneous data set. Our sample contains 255 visually identified cluster candidates, of which 109 were not listed in any previous catalog. We quantify the crowding effect for the stellar sample produced by the DES Data Management pipeline and conclude that the stellar completeness is < 10% inside typical LMC cluster cores. We therefore develop a pipeline to sample and measure stellar magnitudes and positions around the cluster candidates using DAOPHOT. We also implement a maximum-likelihood method to fit individual density profiles and colour-magnitude diagrams. For 117 (from a total of 255) of the cluster candidates (28 uncatalogued clusters), we obtain reliable ages, metallicities, distance moduli and structural parameters, confirming their nature as physical systems. The distribution of cluster metallicities shows a radial dependence, with no clusters more metal-rich than [Fe/H] ~ -0.7 beyond 8 kpc from the LMC center. Furthermore, the age distribution has two peaks at ≃ 1.2 Gyr and ≃ 2.7 Gyr.

79 ASTRONOMY AND ASTROPHYSICS↗

PRE-MAIN-SEQUENCE TURN-ON AS A CHRONOMETER FOR YOUNG CLUSTERS: NGC 346 AS A BENCHMARK

We present a novel approach to deriving the age of very young star clusters, by using the Turn-On (TOn). The TOn is the point in the color-magnitude diagram (CMD) where the pre-main sequence (PMS) joins the main sequence (MS). In the MS luminosity function (LF) of the cluster, the TOn is identified as a peak followed by a dip. We propose that by combining the CMD analysis with the monitoring of the spatial distribution of MS stars it is possible to reliably identify the TOn in extragalactic star-forming regions. Compared to alternative methods, this technique is complementary to the turnoff dating and avoids the systematic biases affecting the PMS phase. We describe the method and its uncertainties and apply it to the star-forming region NGC 346, which has been extensively imaged with the Hubble Space Telescope (HST). This study extends the LF approach in crowded extragalactic regions and opens the way for future studies with HST/WFC3, the James Webb Space Telescope and from the ground with adaptive optics.

79 ASTRONOMY AND ASTROPHYSICS↗

WIYN IMAGING OF THE GLOBULAR CLUSTER SYSTEMS OF THE SPIRAL GALAXIES NGC 891 AND NGC 4013

We present results from a WIYN 3.5 m telescope imaging study of the globular cluster (GC) systems of the edge-on spiral galaxies NGC 891 and NGC 4013. We used the 10' x 10' Minimosaic Imager to observe the galaxies in BVR filters to projected radii of {approx}20 kpc from the galaxy centers. We combined the WIYN data with archival and published data from the WFPC2 and AdvancedCameraforSurveys on the Hubble Space Telescope (HST) to assess the contamination level of the WIYN GC candidate sample and to follow the GC systems further in toward the galaxies' centers. We constructed radial distributions for the GC systems using both the WIYN and HST data. The GC systems of NGC 891 and NGC 4013 extend to 9 {+-} 3 kpc and 14 {+-} 5 kpc, respectively, before falling off to undetectable levels in our images. We use the radial distributions to calculate global values for the total number (N{sub GC}) and specific frequencies (S{sub N} and T) of GCs. NGC 4013 has N{sub GC} = 140 {+-} 20, S{sub N} = 1.0 {+-} 0.2, and T = 1.9 {+-} 0.5; our N{sub GC} value is {approx}40% smaller than a previous determination from the literature. The HST data were especially useful for NGC 891 because the GC system is concentrated toward the plane of the galaxy and was only weakly detected in our WIYN images. Although NGC 891 is thought to resemble the Milky Way in terms of its overall properties, it has only half as many GCs, with N{sub GC}= 70 {+-} 20, S{sub N} = 0.3 {+-} 0.1, and T = 0.6 {+-} 0.3. We also calculate the galaxy-mass-normalized number of blue (metal-poor) GCs in NGC 891 and NGC 4013 and find that they fall along a general trend of increasing specific frequency of blue GCs with increasing galaxy mass. Given currently available resources, the optimal method for studying the global properties of extragalactic GC systems is to combine HST data with wide-field, ground-based imaging with good resolution. The results here demonstrate the advantage gained by using both methods when possible.

79 ASTRONOMY AND ASTROPHYSICS↗

THE CLUSTER LENSING AND SUPERNOVA SURVEY WITH HUBBLE (CLASH): STRONG-LENSING ANALYSIS OF A383 FROM 16-BAND HST/WFC3/ACS IMAGING

We examine the inner mass distribution of the relaxed galaxy cluster A383 (z = 0.189), in deep 16 band Hubble Space Telescope/ACS+WFC3 imaging taken as part of the Cluster Lensing And Supernova survey with Hubble (CLASH) multi-cycle treasury program. Our program is designed to study the dark matter distribution in 25 massive clusters, and balances depth with a wide wavelength coverage, 2000-16000 A, to better identify lensed systems and generate precise photometric redshifts. This photometric information together with the predictive strength of our strong-lensing analysis method identifies 13 new multiply lensed images and candidates, so that a total of 27 multiple images of nine systems are used to tightly constrain the inner mass profile gradient, dlog {Sigma}/dlog r {approx_equal} -0.6 {+-} 0.1 (r < 160 kpc). We find consistency with the standard distance-redshift relation for the full range spanned by the lensed images, 1.01 < z < 6.03, with the higher-redshift sources deflected through larger angles as expected. The inner mass profile derived here is consistent with the results of our independent weak-lensing analysis of wide-field Subaru images, with good agreement in the region of overlap ({approx}0.7-1 arcmin). Combining weak and strong lensing, the overall mass profile is well fitted by a Navarro-Frenk-White profile with M{sub vir} = (5.37{sup +0.70}{sub -0.63} {+-} 0.26) Multiplication-Sign 10{sup 14} M{sub Sun} h{sup -1} and a relatively high concentration, c{sub vir} = 8.77{sup +0.44}{sub -0.42} {+-} 0.23, which lies above the standard c-M relation similar to other well-studied clusters. The critical radius of A383 is modest by the standards of other lensing clusters, r{sub E} {approx_equal} 16 {+-} 2'' (for z{sub s} = 2.55), so the relatively large number of lensed images uncovered here with precise photometric redshifts validates our imaging strategy for the CLASH survey. In total we aim to provide similarly high-quality lensing data for 25 clusters, 20 of which are X-ray-selected relaxed clusters, enabling a precise determination of the representative mass profile free from lensing bias.

79 ASTRONOMY AND ASTROPHYSICS↗

MASS AND LUMINOSITY EVOLUTION OF YOUNG STELLAR OBJECTS

A model of protostar mass and luminosity evolution in clusters gives new estimates of cluster age, protostar birthrate, accretion rate, and mean accretion time. The model assumes constant protostar birthrate, core-clump accretion, and equally likely accretion stopping. Its parameters are set to reproduce the initial mass function and to match protostar luminosity distributions in nearby star-forming regions. It obtains cluster ages and birthrates from the observed numbers of protostars and pre-main sequence (PMS) stars, and from the modal value of the protostar luminosity. In 31 embedded clusters and complexes, the global cluster age is 1-3 Myr, matching available estimates based on optical spectroscopy (OS) and evolutionary tracks. This method of age estimation is simpler than OS and is more useful for young embedded clusters where optical spectroscopy is not possible. In the youngest clusters, the protostar fraction decreases outward from the densest gas, indicating that the local star-forming age increases outward from a few 0.1 Myr in small protostar-dominated zones to a few Myr in large PMS-dominated zones.

79 ASTRONOMY AND ASTROPHYSICS↗

Signatures of Light Massive Relics on non-linear structure formation

ABSTRACT Cosmologies with Light Massive Relics (LiMRs) as a subdominant component of the dark sector are well-motivated from a particle physics perspective, and can also have implications for the σ8 tension between early and late time probes of clustering. The effects of LiMRs on the cosmic microwave background (CMB) and structure formation on large (linear) scales have been investigated extensively. In this paper, we initiate a systematic study of the effects of LiMRs on smaller, non-linear scales using cosmological N-body simulations; focusing on quantities relevant for photometric galaxy surveys. For most of our study, we use a particular model of non-thermal LiMRs but the methods developed generalizing to a large class of LiMR models – we explicitly demonstrate this by considering the Dodelson–Widrow velocity distribution. We find that, in general, the effects of LiMR on small scales are distinct from those of a ΛCDM universe, even when the value of σ8 is matched between the models. We show that weak lensing measurements around massive clusters, between ∼0.1 h−1Mpc and ∼10 h−1Mpc, should have sufficient signal-to-noise in future surveys to distinguish between ΛCDM and LiMR models that are tuned to fit both CMB data and linear scale clustering data at late times. Furthermore, we find that different LiMR cosmologies indistinguishable by conventional linear probes can be distinguished by non-linear probes if their velocity distributions are sufficiently different. LiMR models can, therefore, be best tested by jointly analyzing the CMB and late-time structure formation on both large and small scales.

79 ASTRONOMY AND ASTROPHYSICS↗

Measuring metallicities with Hubble space telescope/wide-field camera 3 photometry

We quantified and calibrated the metallicity and temperature sensitivities of colors derived from nine Wide-Field Camera 3 filters on board the Hubble Space Telescope using Dartmouth isochrones and Kurucz atmosphere models. The theoretical isochrone colors were tested and calibrated against observations of five well studied galactic clusters, M92, NGC 6752, NGC 104, NGC 5927, and NGC 6791, all of which have spectroscopically determined metallicities spanning –2.30 < [Fe/H] <+0.4. We found empirical corrections to the Dartmouth isochrone grid for each of the following color-magnitude diagrams (CMDs): (F555W-F814W, F814W), (F336W-F555W, F814W), (F390M-F555W, F814W), and (F390W-F555W, F814W). Using empirical corrections, we tested the accuracy and spread of the photometric metallicities assigned from CMDs and color-color diagrams (which are necessary to break the age-metallicity degeneracy). Testing three color-color diagrams [(F336W-F555W),(F390M-F555W),(F390W-F555W), versus (F555W-F814W)], we found the colors (F390M-F555W) and (F390W-F555W) to be the best suited to measure photometric metallicities. The color (F390W-F555W) requires much less integration time, but generally produces wider metallicity distributions and, at very low metallicity, the metallicity distribution function (MDF) from (F390W-F555W) is ∼60% wider than that from (F390M-F555W). Using the calibrated isochrones, we recovered the overall cluster metallicity to within ∼0.1 dex in [Fe/H] when using CMDs (i.e., when the distance, reddening, and ages are approximately known). The measured MDF from color-color diagrams shows that this method measures metallicities of stellar clusters of unknown age and metallicity with an accuracy of ∼0.2-0.5 dex using F336W-F555W, ∼0.15-0.25 dex using F390M-F555W, and ∼0.2-0.4 dex with F390W-F555W, with the larger uncertainty pertaining to the lowest metallicity range.

79 ASTRONOMY AND ASTROPHYSICS↗

Quantitative Imaging of Cobalt Phthalocyanine Distribution on Carbon Nanotubes: A Deep Learning Approach to Catalyst Characterization

Electrochemical reduction of carbon dioxide (CO 2 ) offers a pathway to valuable products, with catalysts playing a crucial role. This study investigates the distribution of cobalt tetraaminophthalocyanine (CoPc-NH 2 ) immobilized on carbon nanotubes (CNTs), utilizing high-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) to characterize CoPc-NH 2 distribution. A challenge in the quantitative HAADF-STEM analysis is the introduction of bias from manual Co atom identification. To address this, we developed and trained a convolutional neural network (CNN) using a data set generated from images of CoPc-NH 2 /CNT samples with varying Co loadings. The CNN, implemented in TensorFlow and Keras, facilitated Co atom detections. Analysis of the CNN-generated data confirmed a correlation between Co loading and surface density, consistent with findings from UV–vis spectroscopy. Furthermore, the application of Ripley’s L(d) function highlighted the presence of slight Co atom clustering. Furthermore, this work demonstrates the utility of the combined HAADF-STEM and CNN approach for providing spatially resolved information about catalyst distribution on nonplanar supports, revealing structural details that are typically lost through other characterization methods.

HAADF-STEM↗

Phosphorus doping of Si nanocrystals embedded in silicon oxynitride determined by atom probe tomography

Silicon nanocrystals (SiNCs) embedded in a silicon oxide matrix were studied by 3D atom probe tomography (APT). The distribution of the SiNC diameter was found to have a mean value of 3.7 ± 0.8 nm. The elemental composition of these particles was determined by employing two different approaches: (i) The proximity histogram method and (ii) a cluster identification algorithm based on maximum-atom separations. Both approaches give very similar values in terms of the amount of P, O, and Si within the SiNCs: the mean atomic concentrations are c{sub P} = 0.77% ± 0.4%, c{sub O} = 12.3% ± 2.1%, and c{sub Si} = 85.3% ± 2.1%. A detailed cluster analysis implies that, on average, a 4.5-nm SiNC would contain around 30 P atoms, whereas a 2.0-nm SiNC would contain only around 3 P atoms. Radial concentration profiles obtained for these SiNCs indicate that the P content is inhomogeneous and possibly enhanced at the boundary as compared to the interior of the NCs. About 20% of the P atoms are found to be incorporated into the SiNCs, whereas roughly 30% are trapped within the interfacial layer (with a thickness of ∼ 0.8 nm); the remainder resides in the surrounding matrix. Cluster-size dependent P concentrations support the view of self-purification in the Si nanostructures.

75 CONDENSED MATTER PHYSICS, SUPERCONDUCTIVITY AND↗

Conformational sampling enhancement of replica exchange molecular dynamics simulations using swarm particle intelligence

In this paper, we present a new method based on swarm particle social intelligence for use in replica exchange molecular dynamics simulations. In this method, the replicas (representing the different system configurations) are allowed communicating with each other through the individual and social knowledge, in additional to considering them as a collection of real particles interacting through the Newtonian forces. The new method is based on the modification of the equations of motion in such way that the replicas are driven towards the global energy minimum. The method was tested for the Lennard-Jones clusters of N = 4, 5, and 6 atoms. Our results showed that the new method is more efficient than the conventional replica exchange method under the same practical conditions. In particular, the new method performed better on optimizing the distribution of the replicas among the thermostats with time and, in addition, ergodic convergence is observed to be faster. We also introduce a weighted histogram analysis method allowing analyzing the data from simulations by combining data from all of the replicas and rigorously removing the inserted bias.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Nuclear Computational Low Energy Initiative (NUCLEI)

This is the Final Report for Ohio State University for the NUCLEI SciDAC­3 project. In addition to the local OSU summary given here, we have attached a global NUCLEI summary and a list of papers and talks by the full collaboration, with OSU contributions marked in red. The NUCLEI project, as defined by the scope of work, developed, implemented and ran codes for large­scale computations of many topics in low­energy nuclear physics. Physics studied included the properties of nuclei and nuclear decays, nuclear structure and reactions, and the properties of nuclear matter. The computational techniques used included Quantum Monte Carlo, Configuration Interaction, Coupled Cluster, In­Medium SRG, and Density Functional methods. The research program emphasized areas of high interest to current and possible future DOE nuclear physics facilities, including ATLAS and FRIB (nuclear structure and reactions, and nuclear astrophysics), TJNAF (neutron distributions in nuclei, few body systems, and electroweak processes), NIF (thermonuclear reactions), MAJORANA and FNPB (neutrinoless double­beta decay and physics beyond the Standard Model), and LANSCE (fission studies). Research at Ohio State University played an integral role in NUCLEI by providing input microscopic interactions and operators needed for ab initio structure and reaction calculations including uncertainty quantification, independent ab initio calculations using the in­medium similarity renormalization group (IM­SRG), and developing microscopic constraints for the next­generation energy density functionals (EDF). Much of the theoretical work at OSU was closely coordinated with the theory team at Michigan State University. Although the OSU group did not usually directly develop codes for the highest performance computing components of NUCLEI, we provided needed inputs for much of the effort. In particular, our work contributes to the “Validated Nuclear Interactions” task as well as to both of the “Structure and Reactions” tasks.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗

Porting a Hall MHD Code to a Graphic Processing Unit

We present our experience porting a Hall MHD code to a Graphics Processing Unit (GPU). The code is a 2nd order accurate MUSCL-Hancock scheme which makes use of an HLL Riemann solver to compute numerical fluxes and second-order finite differences to compute the Hall contribution to the electric field. The divergence of the magnetic field is controlled with Dedner?s hyperbolic divergence cleaning method. Preliminary benchmark tests indicate a speedup (relative to a single Nehalem core) of 58x for a double precision calculation. We discuss scaling issues which arise when distributing work across multiple GPUs in a CPU-GPU cluster.

Dorelli, John C.↗

The Gas Distribution in Galaxy Cluster Outer Regions

Aims. We present the analysis of a local (z = 0.04 - 0.2) sample of 31 galaxy clusters with the aim of measuring the density of the X-ray emitting gas in cluster outskirts. We compare our results with numerical simulations to set constraints on the azimuthal symmetry and gas clumping in the outer regions of galaxy clusters. Methods. We exploit the large field-of-view and low instrumental background of ROSAT/PSPC to trace the density of the intracluster gas out to the virial radius. We perform a stacking of the density profiles to detect a signal beyond r200 and measure the typical density and scatter in cluster outskirts. We also compute the azimuthal scatter of the profiles with respect to the mean value to look for deviations from spherical symmetry. Finally, we compare our average density and scatter profiles with the results of numerical simulations. Results. As opposed to some recent Suzaku results, and confirming previous evidence from ROSAT and Chandra, we observe a steepening of the density profiles beyond approximately r(sub 500). Comparing our density profiles with simulations, we find that non-radiative runs predict too steep density profiles, whereas runs including additional physics and/or treating gas clumping are in better agreement with the observed gas distribution. We report for the first time the high-confidence detection of a systematic difference between cool-core and non-cool core clusters beyond 0.3r(sub 200), which we explain by a different distribution of the gas in the two classes. Beyond r(sub 500), galaxy clusters deviate significantly from spherical symmetry, with only little differences between relaxed and disturbed systems. We find good agreement between the observed and predicted scatter profiles, but only when the 1% densest clumps are filtered out in the simulations. Conclusions. Comparing our results with numerical simulations, we find that non-radiative simulations fail to reproduce the gas distribution, even well outside cluster cores. Although their general behavior is in better agreement with the observations, simulations including cooling and star formation convert a large amount of gas into stars, which results in a low gas fraction with respect to the observations. Consequently, a detailed treatment of gas cooling, star formation, AGN feedback, and taking into account gas clumping is required to construct realistic models of cluster outer regions.

Eckert, D.↗

An Iterative Local Updating Ensemble Smoother for Estimation and Uncertainty Assessment of Hydrologic Model Parameters With Multimodal Distributions

Abstract Ensemble smoother (ES) has been widely used in inverse modeling of hydrologic systems. However, for problems where the distribution of model parameters is multimodal, using ES directly would be problematic. One popular solution is to use a clustering algorithm to identify each mode and update the clusters with ES separately. However, this strategy may not be very efficient when the dimension of parameter space is high or the number of modes is large. Alternatively, we propose in this paper a very simple and efficient algorithm, i.e., the iterative local updating ensemble smoother (ILUES), to explore multimodal distributions of model parameters in nonlinear hydrologic systems. The ILUES algorithm works by updating local ensembles of each sample with ES to explore possible multimodal distributions. To achieve satisfactory data matches in nonlinear problems, we adopt an iterative form of ES to assimilate the measurements multiple times. Numerical cases involving nonlinearity and multimodality are tested to illustrate the performance of the proposed method. It is shown that overall the ILUES algorithm can well quantify the parametric uncertainties of complex hydrologic models, no matter whether the multimodal distribution exists.

Zhang, Jiangjiang↗

Image Deconvolution and Point-spread Function Reconstruction with STARRED: A Wavelet-based Two-channel Method Optimized for Light-curve Extraction

We present starred, a point-spread function (PSF) reconstruction, two-channel deconvolution, and light-curve extraction method designed for high-precision photometric measurements in imaging time series. An improved resolution of the data is targeted rather than an infinite one, thereby minimizing deconvolution artifacts. In addition, starred performs a joint deconvolution of all available data, accounting for epoch-to-epoch variations of the PSF and decomposing the resulting deconvolved image into a point source and an extended source channel. The output is a high-signal-to-noise-ratio, high-resolution frame combining all data and the photometry of all point sources in the field of view as a function of time. Of note, starred also provides exquisite PSF models for each data frame. We showcase three applications of starred in the context of the imminent LSST survey and of JWST imaging: (i) the extraction of supernovae light curves and the scene representation of their host galaxy; (ii) the extraction of lensed quasar light curves for time-delay cosmography; and (iii) the measurement of the spectral energy distribution of globular clusters in the "Sparkler," a galaxy at redshift z = 1.378 strongly lensed by the galaxy cluster SMACS J0723.3-7327. starred is implemented in jax, leveraging automatic differentiation and graphics processing unit acceleration. This enables the rapid processing of large time-domain data sets, positioning the method as a powerful tool for extracting light curves from the multitude of lensed or unlensed variable and transient objects in the Rubin-LSST data, even when blended with intervening objects.

79 ASTRONOMY AND ASTROPHYSICS↗

Respiratory motion sampling in 4DCT reconstruction for radiotherapy

Purpose: Phase-based and amplitude-based sorting techniques are commonly used in four-dimensional CT (4DCT) reconstruction. However, effect of these sorting techniques on 4D dose calculation has not been explored. In this study, the authors investigated a candidate 4DCT sorting technique by comparing its 4D dose calculation accuracy with that for phase-based and amplitude-based sorting techniques.Method: An optimization model was formed using organ motion probability density function (PDF) in the 4D dose convolution. The objective function for optimization was defined as the maximum difference between the expected 4D dose in organ of interest and the 4D dose calculated using a 4DCT sorted by a candidate sampling method. Sorting samples, as optimization variables, were selected on the respiratory motion PDF assessed during the CT scanning. Breathing curves obtained from patients' 4DCT scanning, as well as 3D dose distribution from treatment planning, were used in the study. Given the objective function, a residual error analysis was performed, and k-means clustering was found to be an effective sampling scheme to improve the 4D dose calculation accuracy and independent with the patient-specific dose distribution. Results: Patient data analysis demonstrated that the k-means sampling was superior to the conventional phase-based and amplitude-based sorting and comparable to the optimal sampling results. For phase-based sorting, the residual error in 4D dose calculations may not be further reduced to an acceptable accuracy after a certain number of phases, while for amplitude-based sorting, k-means sampling, and the optimal sampling, the residual error in 4D dose calculations decreased rapidly as the number of 4DCT phases increased to 6.Conclusion: An innovative phase sorting method (k-means method) is presented in this study. The method is dependent only on tumor motion PDF. It could provide a way to refine the phase sorting in 4DCT reconstruction and is effective for 4D dose accumulation. Optimized sorting techniques could achieve acceptable residuals (less than 0.5% of the prescription dose) using 6 sorting samples, which is much better than amplitude-based or phase-based sorting. Further increase in sorting phase number exceeding 6 or more may not be necessary when using the k-means sampling or optimal sampling points.

62 RADIOLOGY AND NUCLEAR MEDICINE↗