Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “dl”

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 55 records · Page 3
All sourcesNASA NTRSDOE OSTI
DOE OSTI2014

Elastic properties, sp³ fraction, and Raman scattering in low and high pressure synthesized diamond-like boron rich carbides

Dense BC{sub x} phases with high boron concentration are predicted to be metastable, superhard, and conductors or superconductors depending on boron concentration. However, up to this point, diamond-like boron rich carbides BC{sub x} (dl-BC{sub x}) phases have been thought obtainable only through high pressure and high temperature treatment, necessitating small specimen volume. Here, we use electron energy loss spectroscopy combined with transmission electron microscopy, Raman spectroscopy, surface Brillouin scattering, laser ultrasonics (LU) technique, and analysis of elastic properties to demonstrate that low pressure synthesis (chemical vapor deposition) of BC{sub x} phases may also lead to the creation of diamond-like boron rich carbides. The elastic properties of the dl-BC{sub x} phases depend on the carbon sp²versus sp³ content, which decreases with increasing boron concentration, while the boron bonds determine the shape of the Raman spectra of the dl-BC{sub x} after high pressure-high temperature treatment. Using the estimation of the density value based on the sp³ fraction, the shear modulus μ of dl-BC₄, containing 10% carbon atoms with sp³ bonds, and dl-B₃C₂, containing 38% carbon atoms with sp³ bonds, were found to be μ = 19.3 GPa and μ = 170 GPa, respectively. The presented experimental data also imply that boron atoms lead to a creation of sp³ bonds during the deposition processes.

75 CONDENSED MATTER PHYSICS, SUPERCONDUCTIVITY AND↗
DOE OSTI2016

Effect of heavy metal layer thickness on spin-orbit torque and current-induced switching in Hf|CoFeB|MgO structures

We study the heavy metal layer thickness dependence of the current-induced spin-orbit torque (SOT) in perpendicularly magnetized Hf|CoFeB|MgO multilayer structures. The damping-like (DL) current-induced SOT is determined by vector anomalous Hall effect measurements. A non-monotonic behavior in the DL-SOT is found as a function of the thickness of the heavy-metal layer. The sign of the DL-SOT changes with increasing the thickness of the Hf layer in the trilayer structure. As a result, in the current-driven magnetization switching, the preferred direction of switching for a given current direction changes when the Hf thickness is increased above ∼7 nm. Although there might be a couple of reasons for this unexpected behavior in DL-SOT, such as the roughness in the interfaces and/or impurity based electric potential in the heavy metal, one can deduce a roughness dependence sign reversal in DL-SOT in our trilayer structure.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗
DOE OSTI2015

TU-F-CAMPUS-J-03: Evaluation of a New GE Device-Less Cine 4D-CT

Purpose: Standard cine 4D-CT (S-4DCT) is the cine CT scan of the thorax followed by image sorting with the respiratory signal recorded by the RPM. Although the feasibility of cine 4D-CT without RPM or device-less 4DCT (DL-4DCT) has been reported in a laboratory setting, the only commercial implementation of DL-4DCT was made recently by GE based on the measurements of the lung, body and air area and density. We report the initial results of this new DL-4DCT on its determination of gross tumor volume (GTV). Methods: 30 stereotactic body radiation therapy (SBRT) patients with NSCLC were included in the study. All patients received the S-4DCT for their treatment planning. Their cine CT data without the respiratory signal from RPM were submitted to the DL-4DCT. The DL-4DCT image quality was assessed in reference to S-4DCT. Using maximum intensity projection (MIP) images, the GTVs of the S-4DCT and DL-4DCT were compared on a subset of 9 patients whose tumors in the low density lung regions could be contoured using a region growing algorithm in MIM without contouring bias from the user. A lower threshold of −424 HU was used for all patients and other algorithm parameters were held constant for each patient. Results: The DL-4DCT was able to produce the 4DCT images on 29 out of the 30 SBRT cases. One case failed due to the enhanced calcification surrounding both the breast implants. The GTVs determined on the 9 patients with DL-4DCT were 4.2 ± 4.8% smaller than the GTVs with S-4DCT. However, this was statistically insignificant (p=0.15). The Dice similarity coefficients were 95.1 ± 1.8%. The image quality of DL-4DCT and S-4DCT was similar on the 29 cases. Conclusion: The first commercial DL-4DCT was promising in generating 4D-CT images without a respiratory monitoring device in this preliminary study of 30 patients.

60 APPLIED LIFE SCIENCES↗
DOE OSTI2018

A Transdisciplinary Review of Deep Learning Research and Its Relevance for Water Resources Scientists

Deep learning (DL), a new generation of artificial neural network research, has transformed industries, daily lives, and various scientific disciplines in recent years. DL represents significant progress in the ability of neural networks to automatically engineer problem-relevant features and capture highly complex data distributions. I argue that DL can help address several major new and old challenges facing research in water sciences such as interdisciplinarity, data discoverability, hydrologic scaling, equifinality, and needs for parameter regionalization. This review paper is intended to provide water resources scientists and hydrologists in particular with a simple technical overview, transdisciplinary progress update, and a source of inspiration about the relevance of DL to water. The review reveals that various physical and geoscientific disciplines have utilized DL to address data challenges, improve efficiency, and gain scientific insights. DL is especially suited for information extraction from image-like data and sequential data. Techniques and experiences presented in other disciplines are of high relevance to water research. Meanwhile, less noticed is that DL may also serve as a scientific exploratory tool. A new area termed AI neuroscience , where scientists interpret the decision process of deep networks and derive insights, has been born. This budding subdiscipline has demonstrated methods including correlation-based analysis, inversion of network-extracted features, reduced-order approximations by interpretable models, and attribution of network decisions to inputs. Moreover, DL can also use data to condition neurons that mimic problem-specific fundamental organizing units, thus revealing emergent behaviors of these units. Vast opportunities exist for DL to propel advances in water sciences.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2020

MARBLE: A Multi-GPU Aware Job Scheduler for Deep Learning on HPC Systems

Deep learning (DL) has become a key tool for solving complex scientific problems. However, managing the multi-dimensional large-scale data associated with DL, especially atop extant multiple graphics processing units (GPUs) in modern supercomputers poses significant challenges. Moreover, the latest high-performance computing (HPC) architectures bring different performance trends in training throughput compared to the existing studies. Existing DL optimizations such as larger batch size and GPU locality-aware scheduling have little effect on improving DL training throughput performance due to fast CPU-to-GPU connections. Additionally, DL training on multiple GPUs scales sublinearly. Thus, simply adding more GPUs to a system is ineffective. To this end, we design MARBLE, a first-of-its-kind job scheduler, which considers the non-linear scalability of GPUs at the intra-node level to schedule an appropriate number of GPUs per node for a job. By sharing the GPU resources on a node with multiple DL jobs, MARBLE avoids low GPU utilization in current multi-GPU DL training on HPC systems. Our comprehensive evaluation in the Summit supercomputer shows that MARBLE is able to improve DL training performance by up to 48.3% compared to the popular Platform Load Sharing Facility (LSF) scheduler. Compared to the state-of-the-art of DL scheduler, Optimus, MARBLE reduces the job completion time by up to 47%.

Han, Jingoo↗
DOE OSTI2019

A Quantitative Study of Deep Learning Training on Heterogeneous Supercomputers

Deep learning (DL) has become a key technique for solving complex problems in scientific research and discovery. DL training for science is substantially challenging because it has to deal with massive quantities of multi-dimensional data. High- performance computing (HPC) supercomputers are increasingly being employed for meeting the exponentially growing demand for DL. Multiple GPUs and high-speed interconnect network are needed for supporting DL on HPC systems. However, the excessive use of GPUs without considering effective benefits leads to inefficient resource utilization of these expensive setups. In this paper, we conduct a quantitative analysis to gauge the efficacy of DL workloads on the latest HPC system and identify viability of next-generation DL-optimized heterogeneous supercomputers for enabling researchers to develop more efficient resource management and distributed DL middleware. We evaluate well- known DL models with large-scale datasets using the popular TensorFlow framework, and provide a thorough evaluation in- cluding scalability, accuracy, variability, storage resource, GPU- GPU/GPU-CPU data transfer, and GPU utilization. Our analysis reveals that the latest heterogeneous supercomputing cluster shows varying performance trend as compared to the existing literature for single- and multi-node training. To the best of our knowledge, this is the first work to conduct such a quantitative and comprehensive study of DL training on a supercomputing system with multiple GPUs.

Han, Jingoo↗
DOE OSTI2017

What does fault tolerant Deep Learning need from MPI?

Deep Learning (DL) algorithms have become the {\em de facto} Machine Learning (ML) algorithm for large scale data analysis. DL algorithms are computationally expensive -- even distributed DL implementations which use MPI require days of training (model learning) time on commonly studied datasets. Long running DL applications become susceptible to faults -- requiring development of a fault tolerant system infrastructure, in addition to fault tolerant DL algorithms. This raises an important question: {\em What is needed from MPI for designing fault tolerant DL implementations?} In this paper, we address this problem for permanent faults. We motivate the need for a fault tolerant MPI specification by an in-depth consideration of recent innovations in DL algorithms and their properties, which drive the need for specific fault tolerance features. We present an in-depth discussion on the suitability of different parallelism types (model, data and hybrid); a need (or lack thereof) for check-pointing of any critical data structures; and most importantly, consideration for several fault tolerance proposals (user-level fault mitigation (ULFM), Reinit) in MPI and their applicability to fault tolerant DL implementations. We leverage a distributed memory implementation of Caffe, currently available under the Machine Learning Toolkit for Extreme Scale (MaTEx). We implement our approaches by extending MaTEx-Caffe for using ULFM-based implementation. Our evaluation using the ImageNet dataset and AlexNet neural network topology demonstrates the effectiveness of the proposed fault tolerant DL implementation using OpenMPI based ULFM.

Amatya, Vinay C.↗
DOE OSTI2022

Understanding the Physics Representation of Deep Learning Models in Environmental Applications

Deep learning (DL) models have been popular in earth and environmental modeling and analysis, which exhibit huge potential in capturing and reconstructing the non-linearity of relevant environmental processes. They are extensively used as analytical tools or emulators for multiple domains (atmosphere, land surface, ocean, and biogeochemistry). Despite their success, their internal working mechanism remains largely unknown. Such a lack of knowledge hinders the identification of physically consistent models that are fully adaptive to non-stationary climate, as well as the development of physics-informed machine learning such as physics-informed neural network (PINN). To establish preliminary knowledge and framework of such physics representation evaluation, this project focuses on an improved understanding of DL models in the environmental applications. DL models are increasingly applied to environmental modeling and prediction. However, they have been evaluated mostly from a performance perspective, and there is a gap in understanding how they represent the known physics internally. Such knowledge is especially critical when applying DL models under climate change conditions, where new inputs are likely outside the ranges of the training datasets. In this project, we reveal how the known physical processes are represented within DL models from both statistical and mechanistic perspectives. Leveraging the traditional model evaluations that focus more on the accuracies of predictions, we establish a framework that examines both the accuracy and physics representation of DL models. This analysis framework can identify DL models that make the correct predictions based on correct physics, thus enhancing the existing explainable artificial intelligence (explainable-AI) portfolio. It lays a foundation for developing novel metrics to evaluate the emerging DL models in environmental applications. This knowledge also informs the development of physics-informed DL models by revealing the direct connections between the known physical processes and specific model components or structures.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2021

ClimateNet: an expert-labeled open dataset and deep learning architecture for enabling high-precision analyses of extreme weather

Abstract. Identifying, detecting, and localizing extreme weather events is a crucial first step in understanding how they may vary under different climate change scenarios. Pattern recognition tasks such as classification, object detection, and segmentation (i.e., pixel-level classification) have remained challenging problems in the weather and climate sciences. While there exist many empirical heuristics for detecting extreme events, the disparities between the output of these different methods even for a single event are large and often difficult to reconcile. Given the success of deep learning (DL) in tackling similar problems in computer vision, we advocate a DL-based approach. DL, however, works best in the context of supervised learning – when labeled datasets are readily available. Reliable labeled training data for extreme weather and climate events is scarce. We create “ClimateNet” – an open, community-sourced human-expert-labeled curated dataset that captures tropical cyclones (TCs) and atmospheric rivers (ARs) in high-resolution climate model output from a simulation of a recent historical period. We use the curated ClimateNet dataset to train a state-of-the-art DL model for pixel-level identification – i.e., segmentation – of TCs and ARs. We then apply the trained DL model to historical and climate change scenarios simulated by the Community Atmospheric Model (CAM5.1) and show that the DL model accurately segments the data into TCs, ARs, or “the background” at a pixel level. Further, we show how the segmentation results can be used to conduct spatially and temporally precise analytics by quantifying distributions of extreme precipitation conditioned on event types (TC or AR) at regional scales. The key contribution of this work is that it paves the way for DL-based automated, high-fidelity, and highly precise analytics of climate data using a curated expert-labeled dataset – ClimateNet. ClimateNet and the DL-based segmentation method provide several unique capabilities: (i) they can be used to calculate a variety of TC and AR statistics at a fine-grained level; (ii) they can be applied to different climate scenarios and different datasets without tuning as they do not rely on threshold conditions; and (iii) the proposed DL method is suitable for rapidly analyzing large amounts of climate model output. While our study has been conducted for two important extreme weather patterns (TCs and ARs) in simulation datasets, we believe that this methodology can be applied to a much broader class of patterns and applied to observational and reanalysis data products via transfer learning.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2018

HESS Opinions: Incubating deep-learning-powered hydrologic science advances as a community

Recently, deep learning (DL) has emerged as a revolutionary and versatile tool transforming industry applications and generating new and improved capabilities for scientific discovery and model building. The adoption of DL in hydrology has so far been gradual, but the field is now ripe for breakthroughs. This paper suggests that DL-based methods can open up a complementary avenue toward knowledge discovery in hydrologic sciences. In the new avenue, machine-learning algorithms present competing hypotheses that are consistent with data. Interrogative methods are then invoked to interpret DL models for scientists to further evaluate. However, hydrology presents many challenges for DL methods, such as data limitations, heterogeneity and co-evolution, and the general inexperience of the hydrologic field with DL. The roadmap toward DL-powered scientific advances will require the coordinated effort from a large community involving scientists and citizens.Integrating process-based models with DL models will help alleviate data limitations. The sharing of data and baseline models will improve the efficiency of the community as a whole. Open competitions could serve as the organizing events to greatly propel growth and nurture data science education in hydrology, which demands a grassroots collaboration. The area of hydrologic DL presents numerous research opportunities that could, in turn,stimulate advances in machine learning as well.

58 GEOSCIENCES↗
NASA NTRS2009

Reliability of Maximal Strength Testing in Novice Weightlifters

The one repetition maximum (1RM) is a criterion measure of muscle strength. However, the reliability of 1RM testing in novice subjects has received little attention. Understanding this information is crucial to accurately interpret changes in muscle strength. To evaluate the test-retest reliability of a squat (SQ), heel raise (HR), and deadlift (DL) 1RM in novice subjects. Twenty healthy males (31 plus or minus 5 y, 179.1 plus or minus 6.1 cm, 81.4 plus or minus 10.6 kg) with no weight training experience in the previous six months participated in four 1RM testing sessions, with each session separated by 5-7 days. SQ and HR 1RM were conducted using a smith machine; DL 1RM was assessed using free weights. Session 1 was considered a familiarization and was not included in the statistical analyses. Repeated measures analysis of variance with Tukey fs post-hoc tests were used to detect between-session differences in 1RM (p.0.05). Test-retest reliability was evaluated by intraclass correlation coefficients (ICC). During Session 2, the SQ and DL 1RM (SQ: 90.2 }4.3, DL: 75.9 }3.3 kg) were less than Session 3 (SQ: 95.3 }4.1, DL: 81.5 plus or minus 3.5 kg) and Session 4 (SQ: 96.6 }4.0, DL: 82.4 }3.9 kg), but there were no differences between Session 3 and Session 4. HR 1RM measured during Session 2 (150.1 }3.7 kg) and Session 3 (152.5 }3.9 kg) were not different from one another, but both were less than Session 4 (157.5 }3.8 kg). The reliability (ICC) of 1RM measures for Sessions 2-4 were 0.88, 0.83, and 0.87, for SQ, HR, and DL, respectively. When considering only Sessions 3 and 4, the reliability was 0.93, 0.91, and 0.86 for SQ, HR, and DL, respectively. One familiarization session and 2 test sessions (for SQ and DL) were required to obtain excellent reliability (ICC greater than or equal to 0.90) in 1RM values with novice subjects. We were unable to attain this level of reliability following 3 HR testing sessions therefore additional sessions may be required to obtain an ICC of greater than or equal to 0.90. Future resistive exercise studies should consider the reliability of specific measures to ensure that changes in strength with training are attributable to training and not learning effects associated with 1RM testing.

Loehr, James A.↗
DOE OSTI2019

Strategies to Deploy and Scale Deep Learning on the Summit Supercomputer

The rapid growth and wide applicability of Deep Learning (DL) frameworks poses challenges to computing centers which need to deploy and support the software, and also to domain scientists who have to keep up with the system environment and scale up scientific exploration through DL. We offer recommendations for deploying and scaling DL frameworks on the Summit supercomputer, currently atop the Top500 list, at the Oak Ridge National Laboratory Leadership Computing Facility (OLCF). We discuss DL software deployment in the form of containers, and compare performance of native-built frameworks and containerized deployment. Software containers show no noticeable negative performance impact and exhibit faster Python loading times and promise easier maintenance. To explore strategies for scaling up DL model training campaigns, we assess DL compute kernel performance, discuss and recommend I/O data formats and staging, and identify communication needs for scalable message exchange for DL runs at scale. We recommend that users take a step-wise tuning approach beginning with algorithmic kernel choice, node I/O configuration, and communications tuning as best-practice. We present baseline examples of scaling efficiency 87% for a DL run of ResNet50 running on 1024 nodes (6144 V100 GPUs).

Yin, Junqi↗
DOE OSTI2017

Scaling deep learning workloads: NVIDIA DGX-1/Pascal and Intel Knights Landing

Deep Learning (DL) algorithms have become ubiquitous in data analytics. As a result, major computing vendors --- including NVIDIA, Intel, AMD, and IBM --- have architectural road-maps influenced by DL workloads. Furthermore, several vendors have recently advertised new computing products as accelerating large DL workloads. Unfortunately, it is difficult for data scientists to quantify the potential of these different products. This paper provides a performance and power analysis of important DL workloads on two major parallel architectures: NVIDIA DGX-1 (eight Pascal P100 GPUs interconnected with NVLink) and Intel Knights Landing (KNL) CPUs interconnected with Intel Omni-Path or Cray Aries. Our evaluation consists of a cross section of convolutional neural net workloads: CifarNet, AlexNet, GoogLeNet, and ResNet50 topologies using the Cifar10 and ImageNet datasets. The workloads are vendor-optimized for each architecture. Our analysis indicates that although GPUs provide the highest overall performance, the gap can close for some convolutional networks; and the KNL can be competitive in performance/watt. We find that NVLink facilitates scaling efficiency on GPUs. However, its importance is heavily dependent on neural network architecture. Furthermore, for weak-scaling --- sometimes encouraged by restricted GPU memory --- NVLink is less important.

Gawande, Nitin A.↗
DOE OSTI2019

Troubleshooting deep-learner training data problems using an evolutionary algorithm on Summit

Architectural and hyper-parameter design choices can influence deep-learner (DL) model fidelity but can also be affected by malformed training and validation data. However, practitioners may spend significant time refining layers and hyper-parameters before discovering that distorted training data was impeding training progress. We found that an evolutionary algorithm (EA) can be used to diagnose this kind of DL problem. An EA evaluated thousands of DL configurations on Summit that yielded no overall improvement in DL performance, which suggested problems with the training and validation data. We suspected that Contrast Limited Adaptive Histogram Equalization (CLAHE) enhancement that was applied to previously generated digital surface models (DSMs), for which we were training DLs to find errors, had damaged the training data. Subsequent runs with an alternative global normalization yielded significantly improved DL performance. However, the DL Intersection Over Union (IOU) still exhibited consistent sub-par performance, which suggested further problems with the training data and DL approach. Nonetheless, we were able to diagnose this problem within a 12-hour span via Summit runs, which prevented several weeks of unproductive trial-and-error DL configuration refinement, and allowed for a more timely convergence on an ultimately viable solution.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2020

Reification of latent microstructures: On supervised unsupervised and semi-supervised deep learning applications for microstructures in materials informatics

Machine learning (ML), including deep learning (DL), has become increasingly popular in the last few years due to its continually outstanding performance. In this context, we apply machine learning techniques to "learn" the microstructure using both supervised and unsupervised DL techniques. In particular, we focus (1) on the localization problem bridging (micro)structure (localized) property using supervised DL and (2) on the microstructure reconstruction problem in latent space using unsupervised DL. The goal of supervised and semi-supervised DL is to replace crystal plasticity finite element model (CPFEM) that maps from (micro)structure (localized) property, and implicitly the (micro)structure (homogenized) property relationships, while the goal of unsupervised DL is (1) to represent high-dimensional microstructure images in a non-linear low-dimensional manifold, and (2) to discover a way to interpolate microstructures via latent space associating with latent microstructure variables. At the heart of this report is the applications of several common DL architectures, including convolutional neural networks (CNN), autoencoder (AE), and generative adversarial network (GAN), to multiple microstructure datasets, and the quest of neural architecture search for optimal DL architectures.

36 MATERIALS SCIENCE↗
NASA NTRS2022

Towards an Implementation of Differential Dynamic Logic in PVS

This paper describes an ongoing effort to embed and verify differential dynamic logic (dL) in the Prototype Verification System (PVS). dL is a logic for specifying and formally reasoning about hybrid systems, which employ both continuous and discrete dynamics. There are several benefits of this effort. First, the embedding of dL in PVS offers an independent formal verification of the semantics and rules of dL. Second, the embedding is fully operational within PVS, giving PVS practitioners the ability to use dL in the formal specification and verification process. Third, the rich specification language, type system, and powerful interactive prover of PVS can be used on dL objects. In addition to the embedding and verification of dL, a custom extension for Visual Studio Code has been developed, so that a stylized dL syntax can be used to specify hybrid programs and their properties.

Differential Dynamic Logic↗
DOE OSTI2022

Comparison of Machine Learning and Deep Learning for View Identification from Cardiac Magnetic Resonance Images

Background: Artificial intelligence is increasingly utilized to aid in the interpretation of cardiac magnetic resonance (CMR) studies. One of the first steps is the identification of the imaging plane depicted, which can be achieved by both deep learning (DL) and classical machine learning (ML) techniques without user input. We aimed to compare the accuracy of ML and DL for CMR view classification and to identify potential pitfalls during training and testing of the algorithms. Methods: To train our DL and ML algorithms, we first established datasets by retrospectively selecting 200 CMR cases. The models were trained using two different cohorts (passively and actively curated) and applied data augmentation to enhance training. Once trained, the models were validated on an external dataset, consisting of 20 cases acquired at another center. We then compared accuracy metrics and applied class activation mapping (CAM) to visualize DL model performance. Results: The DL and ML models trained with the passively-curated CMR cohort were 99.1% and 99.3% accurate on the validation set, respectively. However, when tested on the CMR cases with complex anatomy, both models performed poorly. After training and testing our models again on all 200 cases (active cohort), validation on the external dataset resulted in 95% and 90% accuracy, respectively. The CAM analysis depicted heat maps that demonstrated the importance of carefully curating the datasets to be used for training. Conclusions: Both DL and ML models can accurately classify CMR images, but DL outperformed ML when classifying images with complex heart anatomy.

artificial intelligence↗
DOE OSTI2019

Direct Visualization of Drug–Polymer Phase Separation in Ritonavir–Copovidone Amorphous Solid Dispersions Using in situ Synchrotron X-ray Fluorescence Imaging of Thin Films

Amorphous solid dispersions (ASDs) are new formulations currently being used in pharmaceutical industry. The ASDs, in which amorphous drug and polymeric excipients are intimately mixed at a molecular level, exhibit dramatically enhanced solubility and dissolution characteristics relative to their crystalline drug counterparts. In the process of achieving an ever-increasing drug loading (DL), it is noticed however, that the drug release profile deteriorates significantly beyond a certain DL. As an example, a ritonavir-copovidone ASD achieves continuous and full drug release when DL ≤ 25 wt%. The release drops at 30 wt% and when DL > 30 wt% there is virtually no drug release, behaving like a pure amorphous drug. In this communication, the dissolution behavior of ASD thin films has been investigated by in situ synchrotron X-ray fluorescence (XRF) imaging to elucidate the mechanism for the unique change in extent of drug release as a function of DL. It is found that the drug release profile correlates well with the amorphous-amorphous phase separation (AAPS) onset. At a lower drug loading (up to 20 wt%), it takes more than 12 hours for AAPS to happen while in sharp contrast, it only needs less than 10 minutes for DL > 30 wt%. During AAPS amorphous drug accumulates on the surface of the film which prevents further dissolution from the interior of the ASD. The current study provides a mechanistic understanding of the confounding drug release profile of ASDs as a function of DL, and opens the door for studying drug-excipient (e.g. polymer, surfactant) interactions via XRF imaging in the future.

38 RADIATION CHEMISTRY, RADIOCHEMISTRY, AND NUCLEA↗