Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “explainable”

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 397 records · Page 22
All sourcesNASA NTRSDOE OSTI
DOE OSTI2021

It’s Dust: Solving the Mysteries of the Intrinsic Scatter and Host-galaxy Dependence of Standardized Type Ia Supernova Brightnesses

The use of Type Ia supernovae (SNe Ia) as cosmological tools has motivated significant effort to understand what drives the intrinsic scatter of SN Ia distance modulus residuals after standardization, characterize the distribution of SN Ia colors, and explain why properties of the host galaxies of the SNe correlate with SN Ia distance modulus residuals. We use a compiled sample of ~1450 spectroscopically confirmed photometric light curves of SNe Ia and propose a solution to these three problems simultaneously that also explains an empirical 11 σ detection of the dependence of Hubble residual scatter on SN Ia color. We introduce a physical model of color where intrinsic SN Ia colors with a relatively weak correlation with luminosity are combined with extrinsic dust-like colors ( E ( B - V )) with a wide range of extinction parameter values ( R V ). This model captures the observed trends of Hubble residual scatter and indicates that the dominant component of SN Ia intrinsic scatter is variation in R V . We also find that the recovered E ( B - V ) and R V distributions differ based on global host-galaxy stellar mass, and this explains the observed correlation ( γ ) between mass and Hubble residuals seen in past analyses, as well as an observed 4.5 σ dependence of γ on SN Ia color. This finding removes any need to ascribe different intrinsic luminosities to different progenitor systems. Finally, we measure biases in the equation of state of dark energy ( w ) up to |Δ w | = 0.04 by replacing previous models of SN color with our dust-based model; this bias is larger than any systematic uncertainty in previous SN Ia cosmological analyses.

Astronomy & Astrophysics↗
DOE OSTI2023

The Early Light Curve of the Type Ia Supernova 2021hpr in NGC 3147: Progenitor Constraints with the Companion Interaction Model

The progenitor system of Type Ia supernovae (SNe Ia) is expected to be a close binary system consisting of a carbon/oxygen white dwarf (WD) and a nondegenerate star or another WD. Here, we present results from high-cadence monitoring observations of SN 2021hpr in a spiral galaxy, NGC 3147, and constraints on the progenitor system based on its early multicolor light-curve data. First, we classify SN 2021hpr as a normal SN Ia from its long-term photometric and spectroscopic data. More interestingly, we found a significant "early excess" in the light curve over a simple power-law ~t 2 evolution. The early light curve evolves from blue to red to blue during the first week. To explain this, we fitted the early part of the BVRI-band light curves with a two-component model consisting of ejecta–companion interaction and a simple power-law model. The early excess and its color can be explained by shock-cooling emission due to a companion star having a radius of 8.84 ± 0.58 R ⊙ . We also examined Hubble Space Telescope preexplosion images, finding no detection of a progenitor candidate, consistent with the above result. However, we could not detect signs of a significant amount of stripped mass from a nondegenerate companion star (≲0.003 M ⊙ for Hα emission). The early excess light in the multiband light curve supports a nondegenerate companion in the progenitor system of SN 2021hpr. At the same time, the nondetection of emission lines opens the door for other methods to explain this event.

79 ASTRONOMY AND ASTROPHYSICS↗
DOE OSTI2024

The Fuzzball Paradigm

We describe the puzzles that arise in the quantum theory of black holes, and explain how they are resolved in string theory. We review how the Bekenstein entropy is obtained through the count of brane bound states. We describe the fuzzball construction of black hole microstates. These states have no horizon and radiate from their surface like a normal body, so there is no information puzzle. We explain how the semiclassical approximation is violated in gravitational collapse even though curvatures are low at the classical horizon. This violation happens because the collapse leads to a stretching of space that is {\it fast}: light does not have time to travel across the collapsing region to establish the `vecro' correlations needed in the quantum gravitational vacuum. These vecro correlations arise from the existence of virtual fuzzball fluctuations in the gravitational vacuum, and are significant because of the large degeneracy of fuzzball states implied by the Bekenstein entropy. It is plausible that similar effects of fast expansion may be responsible for effects like dark energy and the Early Dark Energy postulated to explain the Hubble tension.

FOS: Physical sciences↗
DOE OSTI2020

Critical Infrastructure Classification via CNN-based Modeling and Image Analysis

With recent advances in the fields of satellite imagery and machine learning we now have the ability to develop explainable deep learning models that enhance critical infrastructure analysis. Funded through Idaho National Laboratory’s (INL) Laboratory Directed Research and Development (LDRD) office we are in the process of developing a deep learning model capable of identifying critical infrastructure facilities and embedded features within those facilities. Utilizing current limit of practice techniques in the machine learning areas of explainability and transfer learning our model, once complete, will have the capacity to be used on multiple different imagery data sets and produce results that not only classify critical infrastructure facilities, but also explain why a critical infrastructure facility was classified as a certain type of facility. These advancements eliminate the ‘black box’ approach deep learning models have had in the past, where a user will have to trust the conclusion of a model without understanding what reasoning when into the model’s classification process. They also expand a model’s usefulness, traditionally a deep learning model will have to use the same data set it was originally trained on. Given the long training times of deep learning models this is impractical in a number of scenarios. By utilizing transfer learning advancements, we are eliminating the need to train our model on the same data set it is then run on to classify critical infrastructure facilities. We are also enabling the analysis and classification of data sets that are potentially too small to be divided into a training and testing data set. Once completed this model can provide a foundation to enhanced critical infrastructure analysis, dependency analysis, and potential disaster relief efforts.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2023

A Real-Time ANPC Inverter Digital Twin with Integrated Design-For-Trust

The demand for renewable energy has increased over the last few years, and so has the demand for greater expectations within the energy market. This increasing trend has been accompanied by more significant usage of internet-connected devices (IoT), leading to critical electrical infrastructure being connected to the internet. Implementing internet connectivity with such devices and systems provides benefits such as improving the system's performance, facilitating irregularity and anomaly mitigation, and providing additional situational awareness for enhanced decision-making. However, enhancing the connected system with IoT introduces a drawback – a greater vulnerability to cyber-attacks. Cyber-attacks targeting critical infrastructure in the electrical sector have occurred in the United States and Ukraine. These cyber-attacks highlight and expose vulnerabilities that a system inherits when connecting to the internet. These attacks left thousands of customers without electricity for hours until operators could regain control of the electric utility grid. Therefore, to address the vulnerabilities of an internet-connected power electronic device, this work focused on the hardware layer of the system. Implementing a cyber-control system inside the hardware layer can significantly reduce the possibility of an attacker patching malicious controller firmware into a photovoltaic grid-connected inverter, thus mitigating the likelihood that the inverter becomes inactive a cyber-attack scenario. With this mitigation technique, if a cyberattack is successful and an attacker gains control of the network, a cyber-defense technique is in place to mitigate the impact of the cyber-attack. This additional protection layer was developed based on an innovative concept known as Digital Twin (DT). A DT, in this case, replicates an Active-Neutral Point Clamped (ANPC) inverter and was designed using a hardware language known as VHDL (Very High-Speed Integrated Circuit Hardware Description Language) and applied to Field-Programmable-GateArray (FPGA). The DT is embedded within the FPGA and contained in a controller board, the UCB (Unified Controller Board), developed by the University of Arkansas electrical engineering team. This UCB also contains two Digital Signal Processors (DSPs) responsible for generating associated signals to control an authentic physical inverter. These DSP signals are received and processed by the FPGA that implements the DT of an ANPC; in other words, it simulates in realtime the expected output of an actual ANPC inverter using the signals from the DSP. When a new firmware is ready to be patched, the DT provides output signals simulating behavior that a real ANPC inverter would generate with the new firmware. The new firmware is tested to check if it meets all the operational requirements established using a Design-For-Trust technique (DFTr). If the new firmware fails in at least one of the DFT tests, it is considered malicious and must be rejected. This work is divided into sections, such as Background, which explains the pieces that were used and the strategy behind this work; Process and Procedure, which explains the methodology that was adopted to prove the reliability and effectiveness of this work; Results and Discussion, where the simulations and results are described and explained; followed by Conclusion and Future work section, which concludes this work and adds possible future projects to continue this work further.

do Amaral Custodio, Paulo Vitor↗
DOE OSTI2023

A detailed study of interpretability of deep neural network based top taggers

Abstract Recent developments in the methods of explainable artificial intelligence (XAI) allow researchers to explore the inner workings of deep neural networks (DNNs), revealing crucial information about input–output relationships and realizing how data connects with machine learning models. In this paper we explore interpretability of DNN models designed to identify jets coming from top quark decay in high energy proton–proton collisions at the Large Hadron Collider. We review a subset of existing top tagger models and explore different quantitative methods to identify which features play the most important roles in identifying the top jets. We also investigate how and why feature importance varies across different XAI metrics, how correlations among features impact their explainability, and how latent space representations encode information as well as correlate with physically meaningful quantities. Our studies uncover some major pitfalls of existing XAI methods and illustrate how they can be overcome to obtain consistent and meaningful interpretation of these models. We additionally illustrate the activity of hidden layers as neural activation pattern diagrams and demonstrate how they can be used to understand how DNNs relay information across the layers and how this understanding can help to make such models significantly simpler by allowing effective model reoptimization and hyperparameter tuning. These studies not only facilitate a methodological approach to interpreting models but also unveil new insights about what these models learn. Incorporating these observations into augmented model design, we propose the particle flow interaction network model and demonstrate how interpretability-inspired model augmentation can improve top tagging performance.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2020

Direct Observation of Bandgap Oscillations Induced by Optical Phonons in Hybrid Lead Iodi.de Perovskites

Hybrid organic-inorganic perovskites such as methylammonium lead iodide have emerged as promising semiconductors for energy-relevant applications. The interactions between charge carriers and lattice vibrations, giving rise to polarons, have been invoked to explain some of their extraordinary optoelectronic properties. Here, time-resolved optical spectroscopy is performed, with off-resonant pumping and electronic probing, to examine several representative lead iodide perovskites. The temporal oscillations of electronic bandgaps induced by coherent lattice vibrations are reported, which is attributed to antiphase octahedral rotations that dominate in the examined 3D and 2D hybrid perovskites. The off-resonant pumping scheme permits a simplified observation of changes in the bandgap owing to the A(g) vibrational mode, which is qualitatively different from vibrational modes of other symmetries and without increased complexity of photogenerated electronic charges. The work demonstrates a strong correlation between the lead-iodide octahedral framework and electronic transitions, and provides further insights into the manipulation of coherent optical phonons and related properties in hybrid perovskites on ultrafast timescales.

coherent optical phonons↗
DOE OSTI2023

Counterion Control and the Spectral Signatures of Polarons, Coupled Polarons, and Bipolarons in Doped P3HT Films

When an electron is removed from a conjugated polymer, such as poly(3-hexylthiophene-2,5-diyl) (P3HT), the remaining hole and associated change in the polymer backbone structure from aromatic to quinoidal are referred to as a polaron. Bipolarons are created by removing the unpaired electron from an already-oxidized polymer segment. In electrochemically-doped P3HT films, polarons, and bipolarons are readily observed, but in chemically-doped P3HT films, bipolarons rarely form. This is explained by studying the effects of counterion position on the formation of polarons, strongly coupled polarons, and bipolarons using both spectroscopic and X-ray diffraction experiments and time-dependent density functional theory calculations. The counterion positions control whether two polarons spin-pair to form a bipolaron or whether they strongly couple without spin-pairing are found. When two counterions lie close to the same polymer segment, bipolarons can form, with an absorption spectrum that is blueshifted from that of a single polaron. Otherwise, polarons at high concentrations do not spin-pair, but instead J-couple, leading to a redshifted absorption spectrum. The counterion location needed for bipolaron formation is accompanied by a loss of polymer crystallinity. Finally, these results explain the observed formation order of single polarons, coupled single polarons, and singlet bipolarons in electrochemically- and chemically-doped conjugated polymers.

36 MATERIALS SCIENCE↗
DOE OSTI2020

Anisotropic Optical and Frictional Properties of Langmuir–Blodgett Film Consisting of Uniaxially-Aligned Rod-Shaped Cellulose Nanocrystals

Langmuir–Blodgett (LB) film deposition gives an opportunity to control the packing density and orientation of anisotropic nanoparticles at a monolayer level, allowing accurate characterization of their anisotropic material properties. Additionally, the uniaxial deposition of rod-shaped cellulose nanocrystals (CNCs) over a macroscopically large area is achieved by aligning the long axis of CNCs on the LB trough with the direction of the maximum drag force within the meniscus during the vertical pulling of the substrate from the LB trough. On the uniaxially-aligned LB films, anisotropic linear and non-linear optical properties of CNCs are obtained using Mueller matrix spectroscopy and sum frequency generation spectroscopy, respectively, and explained with time-dependent density functional theory calculations. Also, the frictional anisotropy of the LB film is measured using atomic force microscopy and explained theoretically. The findings of this study will be valuable for preparation of anisotropic nanoparticle thin films with uniform arrangements and utilization of their anisotropic material properties.

36 MATERIALS SCIENCE↗
DOE OSTI2023

Dark Exciton in 2D Hybrid Halide Perovskite Films Revealed by Magneto-Photoluminescence at High Magnetic Field

Here, a comprehensive study of the exciton fine structure (EFS) is presented in 2D-phenethylammonium lead iodide films using magnetic field-induced polarization of photoluminescence (PL) in both Faraday and Voigt configurations at fields up to 25 Tesla. Three exciton bands are identified in the PL spectrum associated with bound, dark, and bright excitons, respectively. Under a high magnetic field in Faraday/Voigt configuration, large field-induced circular/linear polarization is observed in the PL band related to the dark exciton, which is magnetically activated. Furthermore, it is found that the dark exciton has an anomalous field-induced circular polarization, which cannot be explained by the classical Boltzmann distribution of spin-polarized species. These findings are well explained by an effective mass model that includes exchange terms unique to the monoclinic symmetry as a perturbation of the EFS in the approximate tetragonal symmetry. It is also confirmed that the field-induced linear polarization is sensitive to the monoclinic exchange term, whereas the field-induced circular polarization is immune to such term.

14 SOLAR ENERGY↗
DOE OSTI2022

Nonradiative Recombination via Charge-Transfer-Exciton to Polaron Energy Transfer Limits Photocurrent in Organic Solar Cells

Here, a recombination and exciton loss mechanism is reported in organic solar cells involving energy transfer between charge transfer (CT) excitons and polarons, impacting photocurrent generation, particularly in the near-infrared where polaronic transitions typically reside. This process sets a low-energy cut-off in the external quantum efficiency spectrum of an excitonic donor/acceptor interface, determined by the low-energy polaron absorption peak and the CT state reorganization energy. Furthermore, this process explains the deviation from unity and bias dependence of the CT state's internal quantum efficiency at low photon energies. This process is demonstrated in a variety of systems and it is hypothesized that CT state to polaron energy transfer recombination may be responsible for a share of nonradiative recombination in all organic photovoltaics and can explain numerous experimentally observed device trends regarding photocurrent generation and energy losses. Overall, this work enhances the understanding of photophysical processes in organic materials and allows the design of systems that can avoid this recombination pathway.

14 SOLAR ENERGY↗
DOE OSTI2020

Kinetic modeling of microbial growth, enzyme activity, and gene deletions: An integrated model of β-glucosidase function in Cellvibrio japonicus

Understanding the complex growth and metabolic dynamics in microorganisms requires advanced kinetic models containing both metabolic reactions and enzymatic regulation to predict phenotypic behaviors under different conditions and perturbations. Most current kinetic models lack gene expression dynamics and are separately calibrated to distinct media, which consequently makes them unable to account for genetic perturbations or multiple substrates. This challenge limits our ability to gain a comprehensive understanding of microbial processes towards advanced metabolic optimizations that are desired for many biotechnology applications. Here, we present an integrated computational and experimental approach for the development and optimization of mechanistic kinetic models for microbial growth and metabolic and enzymatic dynamics. Our approach integrates growth dynamics, gene expression, protein secretion, and gene-deletion phenotypes. We applied this methodology to build a dynamic model of the growth kinetics in batch culture of the bacterium Cellvibrio japonicus grown using either cellobiose or glucose media. The model parameters were inferred from an experimental data set using an evolutionary computation method. The resulting model was able to explain the growth dynamics of C. japonicus using either cellobiose or glucose media and was also able to accurately predict the metabolite concentrations in the wild-type strain as well as in β-glucosidase gene deletion mutant strains. We validated the model by correctly predicting the non-diauxic growth and metabolite consumptions of the wild-type strain in a mixed medium containing both cellobiose and glucose, made further predictions of mutant strains growth phenotypes when using cellobiose and glucose media, and demonstrated the utility of the model for designing industrially-useful strains. Importantly, the model is able to explain the role of the different β-glucosidases and their behavior under genetic perturbations. Here, this integrated approach can be extended to other metabolic pathways to produce mechanistic models for the comprehensive understanding of enzymatic functions in multiple substrates.

59 BASIC BIOLOGICAL SCIENCES↗
DOE OSTI2022

Mechanistic Differences between Electrochemical Hydrogenation and Hydrogenolysis of 5‐Hydroxymethylfurfural and Their pH Dependence

Abstract Hydrogenation and hydrogenolysis are two important reactions for electrochemical reductive valorization of biomass‐derived oxygenates such as 5‐hydroxymethylfurfural (HMF). In general, hydrogenolysis (which combines hydrogenation and deoxygenation) is more challenging than hydrogenation (which does not involve the cleavage of carbon–oxygen bonds). Thus, identifying factors and conditions that can promote hydrogenolysis is of great interest for reductive valorization of biomass‐derived oxygenates. For the electrochemical reduction of HMF and its derivatives, it is known that aldehyde hydrogenation is not a part of aldehyde hydrogenolysis but rather a competing reaction; however, no atomic‐level understanding is currently available to explain their electrochemical mechanistic differences. In this study, combined experimental and computational investigations were performed using Cu electrodes to elucidate the key mechanistic differences between electrochemical hydrogenation and hydrogenolysis of HMF. The results revealed that hydrogenation and hydrogenolysis of HMF involve the formation of different surface‐adsorbed intermediates via different reduction mechanisms and that lowering the pH promoted the formation of the intermediates required for aldehyde and alcohol hydrogenolysis. This study for the first time explains the origins of the experimentally observed pH‐dependent selectivities for hydrogenation and hydrogenolysis and offers a new mechanistic foundation upon which rational strategies to control electrochemical hydrogenation and hydrogenolysis can be developed.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
DOE OSTI2023

Temporal patterns of fine‐root dynamics have little influence on seasonal soil CO 2 efflux in a mixed, mesic forest

Among the contributors to soil CO 2 efflux, there remains uncertainty about the contribution of root activity to the overall soil efflux. Soil water and temperature frequently have been used to predict a large portion of the variation in soil CO 2 efflux. We hypothesized that fine-root dynamics explain most of the remaining variability in soil CO 2 efflux that cannot be explained by soil temperature and water content. We anticipated that seasonal increases in root production, mortality via decomposition, and standing crop would result in corresponding increases in soil CO 2 efflux. We tested our hypotheses by collecting and analyzing two years of minirhizotron and soil chamber CO 2 flux data from plots distributed throughout the Shale Hills Catchment of the Susquehanna-Shale Hills Critical Zone Observatory in Central Pennsylvania, USA. Here we showed that: (1) seasonal fluctuations in fine-root dynamics yielded only a very small increase in the predictability of soil CO 2 efflux; (2) fine-root mortality effects on soil CO 2 efflux were strongly tied to soil temperature; (3) fluctuations in fine-root presence or standing mass independent of temperature and moisture had little effect on soil CO 2 efflux; and (4) new fine-root length and root length mortality had limited impacts on soil CO 2 efflux rates. We conclude that, at least in temperate forests on rocky soils, characterizing fine-root dynamics may provide only limited improvement in the estimation of soil CO 2 efflux.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2023

Structural Determinants of the Binding and Activation of Estrogen Receptor α by Phenolic Thieno[2,3- d ]pyrimidines

Synthetic, structural, and computational approaches were used to solve the puzzle as to how a phenolic nonsteroidal estrogen 1 with only a single H-bond to its receptor was more potent than an isomer 2 which formed an intricate network of H-bonds. Synthesis of a series of substituted phenols revealed that pK a was not a determinant of estrogenic activity. First-principles calculation also failed to explain the difference in activity of 1 and 2. Molecular dynamics revealed that 1 formed a more stable receptor complex compared to 2, which may explain its increased activity despite forming fewer apparent H-bonds with the protein.

59 BASIC BIOLOGICAL SCIENCES↗
DOE OSTI2022

Tidal frequencies and quasiperiodic subsurface water level variations dominate redox dynamics in a salt marsh system

Salt marshes are hotspots of nutrient processing en route to sensitive coastal environments. Whilst our understanding of these systems has improved over the years, we still have limited knowledge of the spatiotemporal variability of critical biogeochemical drivers within salt marshes. Sea-level rise will continue to force change on salt marsh functioning, highlighting the urgency of filling this knowledge gap. Our study was conducted in a central California estuary experiencing extensive marsh drowning and relative sea-level rise, making it a model system for such an investigation. Here we instrumented three marsh positions subjected to different degrees of tidal inundation (6.7%, 8.9%, and 11.2% of the time for the upper, middle, and lower marsh positions, respectively), providing locations with varied biogeochemical characteristics and hydrological interactions at the site. We continuously monitored redox potential (Eh) at depths of 0.1, 0.3, and 0.5 m, subsurface water levels (WL), and temperature at 0.7 m depth at each marsh position. To understand how drivers of subsurface biogeochemical processes fluctuate across tidal cycles, we used wavelet analyses to explain the interactions between Eh and WL. We found that tidal forcing significantly affects key drivers of biogeochemical processes by imparting controls on Eh variability, likely driving subsurface hydro-biogeochemistry of the salt marsh. Wavelet coherence showed that the Eh-WL relationship is nonlinear, and their lead–lag relationship is variable. We found that precipitation events perturb Eh at depth over timescales of hours, even though WL shows relatively minimal change during events. Furthermore, this work highlights the importance of high frequency in situ measurements, such as Eh, to help explain factors that govern subsurface biogeochemistry and hydrological processes in salt marshes.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2021

Oxidation and pyrolysis of methyl propyl ether

The ignition, oxidation, and pyrolysis chemistry of methyl propyl ether (MPE) was probed experimentally at several different conditions, and a comprehensive chemical kinetic model was constructed to help understand the observations, with many of the key parameters computed using quantum chemistry and transition state theory. Experiments were carried out in a shock tube measuring time variation of CO concentrations, in a flow tube measuring product concentrations, and in a rapid compression machine (RCM) measuring ignition delay times. The detailed reaction mechanism was constructed using the Reaction Mechanism Generator software. Sensitivity and flux analyses were used to identify key rate and thermochemical parameters, which were then computed using quantum chemistry to improve the mechanism. Validation of the final model against the 1-20 bar 600-1500 K experimental data is presented with a discussion of the kinetics. The model is in excellent agreement with most of the shock tube and RCM data. Strong non-monotonic variation in conversion and product distribution is observed in the flow-tube experiments as the temperature is increased, and unusually strong pressure dependence and significant heat release during the compression stroke is observed in the RCM experiments. These observations are largely explained by a close competition between radical decomposition and addition to O2 at different sites in MPE; this causes small shifts in conditions to lead to big shifts in the dominant reaction pathways. The validated mechanism was used to study the chemistry occurring during ignition in a diesel engine, simulated using Ignition Quality Test (IQT) conditions. At the IQT conditions, where the MPE concentration is higher, bimolecular reactions of peroxy radicals are much more important than in the RCM.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
DOE OSTI2022

Nonnegative canonical tensor decomposition with linear constraints: nnCANDELINC

Abstract There is an emerging interest for tensor factorization applications in big‐data analytics and machine learning. To speed up the factorization of extra‐large datasets, organized in multidimensional arrays (also known as tensors), easy to compute compression‐based tensor representations, such as, Tucker and tensor train formats, are used to approximate the initial large‐tensor. Further, tensor factorization is used to extract latent features that can facilitate discoveries of new mechanisms and signatures hidden in the data, where the explainability of the latent features is of principal importance. Nonnegative tensor factorization extracts latent features that are naturally sparse and parts of the data, which makes them easily interpretable. However, to take into account available domain knowledge and subject matter expertise, often additional constraints need to be imposed, which lead us to canonical decomposition with linear constraints (CANDELINC), a canonical polyadic decomposition with rank deficient factors. In CANDELINC, Tucker compression is used as a preprocessing step, which lead to a larger residual error but to more explainable latent features. Here, we propose a nonnegative CANDELINC (nnCANDELINC) accomplished via a specific nonnegative Tucker decomposition; we refer to as minimal or canonical nonnegative Tucker. We derive several results required to understand the specificity of nnCANDELINC, focusing on the difficulties of preserving the nonnegative rank of a tensor to its Tucker core and comparing the real valued to nonnegative case. Finally, we demonstrate nnCANDELINC performance on synthetic and real‐world examples.

97 MATHEMATICS AND COMPUTING↗