Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Tuning”

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 1,099 records · Page 61
All sourcesNASA NTRSDOE OSTI
DOE OSTI2024

Understanding changes in cloud simulations from E3SM version 1 to version 2

Abstract. This study documents clouds simulated by the Energy Exascale Earth System Model (E3SM) version 2 (E3SMv2) and attempts to understand what causes the model behavior change in clouds relative to E3SMv1. This is done by analyzing the last 30-year (1985–2014) data from the 165-year historical simulations using E3SMv1 and v2 and four sensitivity tests to isolate the impact of changes in model parameter choices in its turbulence, shallow convection, and cloud macrophysics parameterization (Cloud Layers Unified By Binormals, CLUBB); microphysical parameterization (MG2); and deep-convection scheme (ZM), as well as model physics changes in convective triggering. It is shown that E3SMv2 significantly improves the simulation of subtropical coastal stratocumulus clouds and clouds with optical depth larger than 3.6 over the stratocumulus-to-cumulus transition regimes, where the shortwave cloud radiative effect (SWCRE) is also improved, and the Southern Ocean (SO) while seeing an overall slight degradation in low clouds over other tropical and subtropical oceans. The better performance in E3SMv1 over those regions is partially due to error compensation between its simulated optically thin and intermediate low clouds for which E3SMv2 actually improves simulation of optically intermediate low clouds. Sensitivity tests indicate that the changes in low clouds are primarily due to the tuning done in CLUBB. The impact of the ZM tuning is mainly on optically intermediate and thick high clouds, contributing to an improved SWCRE and longwave cloud radiative effect (LWCRE). The impact of the MG2 tuning and the new convective trigger is primarily on the high latitudes and the SO. They have a relatively smaller impact on clouds than CLUBB tuning and ZM tuning do. This study offers additional insights into clouds simulated in E3SMv2 by utilizing multiple data sets and the Cloud Feedback Model Intercomparison Project (CFMIP) Observation Simulator Package (COSP) diagnostic tool as well as sensitivity tests. The improved understanding will benefit future E3SM developments.

58 GEOSCIENCES↗
DOE OSTI2025

QuadTune version 1: a regional tuner for global atmospheric models

When a new, better-formulated physical parameterization is introduced into a global atmospheric model, aspects of the global model solutions are sometimes degraded. Then, in order to use the new global model to address science questions, there is an incentive to restore its accuracy. Oftentimes this restoration is achieved by tuning of model parameter values. Unfortunately, the retuning process is expensive because characterizing the parameter dependence requires numerous time-consuming global simulations. To reduce the cost of tuning, this manuscript introduces a “poor man's” model tuner, “QuadTune”. QuadTune carves the globe into regions and approximates the model parameter dependence through the use of an uncorrelated quadratic emulator (i.e., response surface). The simplicity of the emulator reduces the required number of global model simulations and aids explainability of tuner behavior. Tuning removes parametric error but leaves behind model structural error. Structural error manifests itself as regional residual biases, such as stubborn biases and tuning trade-offs. To visualize these residual biases, QuadTune's software includes a set of diagnostic plots. This paper illustrates the use of the plots for characterizing residual biases with an example tuning problem.

Environmental sciences↗
DOE OSTI2020

Classical Optimizers for Noisy Intermediate-Scale Quantum Devices

We present a collection of optimizers tuned for usage on Noisy Inter-mediate-Scale Quantum (NISQ) devices. Optimizers have a range of applications in quantum computing, including the Variational Quantum Eigensolver (VQE) and Quantum Approximate Optimization (QAOA) algorithms. They have further uses in calibration, hyperparameter tuning, machine learning, etc. We employ the VQE algorithm as a case study. VQE is a hybrid algorithm, with a classical minimizer step driving the next evaluation on the quantum processor. While most results to date concentrated on tuning the quantum VQE circuit, our study indicates that in the presence of quantum noise the classical minimizer step is a weak link and a careful choice combined with tuning is required for correct results. We explore state-of-the-art gradient-free optimizers capable of handling noisy, black-box, cost functions and stress-test them using a quantum circuit simulation environment with noise injection capabilities on individual gates. Our results indicate that specifically tuned optimizers are crucial to obtaining valid science results on NISQ hardware, as well as projecting forward on fault tolerant circuits.

Lavrijsen, Wim↗
DOE OSTI2022

Wavelength Scaling of Widely-Tunable Terahertz Quantum-Cascade Metasurface Lasers

Terahertz (THz) external-cavity lasers based on quantum-cascade (QC) metasurfaces are emerging as widely-tunable, single-mode sources with the potential to cover the 1--6 THz range in discrete bands with milliwatt-level output power. By operating on an ultra-short cavity with a length on the order of the wavelength, the QC vertical-external-cavity surface-emitting-laser (VECSEL) architecture enables continuous, broadband tuning while producing high quality beam patterns and scalable power output. The methods and challenges for designing the metasurface at different frequencies are discussed. As the QC-VECSEL is scaled below 2 THz, the primary challenges are reduced gain from the QC active region, increased metasurface quality factor and its effect on tunable bandwidth, and larger power consumption due to a correspondingly scaled metasurface area. At frequencies above 4.5 THz, challenges arise from a reduced metasurface quality factor and the excess absorption that occurs from proximity to the Reststrahlen band. The results of four different devices — with center frequencies 1.8 THz, 2.8 THz, 3.5 THz, and 4.5 THz — are reported. Each device demonstrated at least 200 GHz of continuous single-mode tuning, with the largest being 650 GHz around 3.5 THz. The limitations of the tuning range are well modeled by a Fabry-Pérot cavity which accounts for the reflection phase of the metasurface and the effect of the metasurface quality factor on laser threshold. Lastly, the effect of different output couplers on device performance is studied, demonstrating a significant trade-off between the slope efficiency and tuning bandwidth.

47 OTHER INSTRUMENTATION↗
DOE OSTI2024

Experimental Safe Extremum Seeking for Accelerators

Here, we demonstrate the recent designs of safe extremum seeking (Safe ES) on the 1-km-long charged particle accelerator at the Los Alamos Neutron Science Center (LANSCE). Safe ES is a modification of extremum seeking (ES) which, in addition to minimizing an analytically unknown cost, also employs a safety filter based on an analytically unknown control barrier function (CBF) safety metric. Tuning is necessitated by accelerators being large complex systems, with many drifting parameters due to thermal effects and degradation. At the same time, safe operation (the maintenance of state constraints) is crucial, as damage brings astronomical costs, both financially and in operation downtime. Our measured (but analytically unknown) safety metric is the beam current. We perform multivariable Safe ES on three accelerator applications, in which we adapt 4, 6, and 3 magnet strength parameters, respectively. Two of the three applications are for validated simulation models of beamlines at LANSCE: the first for the proton radiography (pRad) beamline of 800-MeV protons for spot size tuning; the second on a high-performance code, HPSim, for tuning the low-energy beam transport (LEBT) region that contains a beam of 750-keV protons. The third is an experimental tuning of the steering magnets in the LEBT at LANSCE.

43 PARTICLE ACCELERATORS↗
DOE OSTI2022

Not All Aggregates Are Made the Same: Distinct Structures of Solution Aggregates Drastically Modulate Assembly Pathways, Morphology, and Electronic Properties of Conjugated Polymers

Abstract Tuning structures of solution‐state aggregation and aggregation‐mediated assembly pathways of conjugated polymers is crucial for optimizing their solid‐state morphology and charge‐transport property. However, it remains challenging to unravel and control the exact structures of solution aggregates, let alone to modulate assembly pathways in a controlled fashion. Herein, aggregate structures of an isoindigo–bithiophene‐based polymer (PII‐2T) are modulated by tuning selectivity of the solvent toward the side chain versus the backbone, which leads to three distinct assembly pathways: direct crystallization from side‐chain‐associated amorphous aggregates, chiral liquid crystal (LC)‐mediated assembly from semicrystalline aggregates with side‐chain and backbone stacking, and random agglomeration from backbone‐stacked semicrystalline aggregates. Importantly, it is demonstrated that the amorphous solution aggregates, compared with semicrystalline ones, lead to significantly improved alignment and reduced paracrystalline disorder in the solid state due to direct crystallization during the meniscus‐guided coating process. Alignment quantified by the dichroic ratio is enhanced by up to 14‐fold, and the charge‐carrier mobility increases by a maximum of 20‐fold in films printed from amorphous aggregates compared to those from semicrystalline aggregates. This work shows that by tuning the precise structure of solution aggregates, the assembly pathways and the resulting thin‐film morphology and device properties can be drastically tuned.

36 MATERIALS SCIENCE↗
DOE OSTI2025

Impact of Halogen Groups on the Properties of PEA–Based 2D Pb–Sn Halide Perovskites

Tuning broad emission in 2D Pb–Sn halide perovskites (HPs) is essential for advancing optoelectronic applications, particularly for color-tunable and white-light-emitting devices. This broad emission is linked to structural factors, such as defects and phase segregation of the Pb component within the Pb–Sn system, which are strongly influenced by the molecular structure and chemical properties of spacer cations. Atomic tuning of the spacers via halogenation opens up a new way to fine-tune the molecular properties, enabling further augmentations of HP functionalities. Nevertheless, the distinct broad emission's sensitivity to spacer chemistry remains underexplored. Here, halogenation's influence is systematically investigated on 2D HP emission characteristics using a high-throughput workflow. These findings reveal that the F-containing phenethylammonium (4F-PEA) spacer narrows the broadband PL, whereas Cl broadens it. Through a correlative study, it is found that 4F-PEA reduces not only the local phase segregation but also the defect levels and microstrains in 2D HPs. This is likely attributed to the manifestation of less lattice distortion via stronger surface coordination of the dipole-augmented 4F-PEA. Furthermore, these results highlight halogenation as a key factor in modulating phase segregation and defect density in 2D Pb–Sn HPs, offering a promising pathway to tune the emission for enhanced optoelectronic performance.

2D Pb-Sn halide perovskites↗
DOE OSTI2025

What can solve the strong CP problem?

Three possible strategies have been advocated to solve the strong CP problem. The first is the axion, a dynamical mechanism that relaxes any initial value of the CP violating angle $\overline{θ}$ to zero. The second is the imposition of new symmetries that are believed to set $\overline{θ}$ to zero in the UV. The third is the acceptance of the fine tuning of parameters. We argue that the latter two solutions do not solve the strong CP problem. The θ term of QCD is not a parameter — it does not exist in the Hamiltonian. Rather, it is a property of the quantum state that our universe finds itself in, arising from the fact that there are CP violating states of a CP preserving Hamiltonian. It is not eliminated by imposing parity as a symmetry since the underlying theory is already parity symmetric and that does not preclude the existence of CP violating states. Moreover, since the value of θ realized in our universe is a consequence of measurement, it is inherently random and cannot be fine tuned by choice of parameters. Rather any fine tuning would require a tuning between parameters in the theory and the random outcome of measurement. Our results considerably strengthen the case for the existence of the axion and axion dark matter. The confusion around θ arises from the fact that unlike classical mechanics, the Hamiltonian and Lagrangian are not equivalent in quantum mechanics. The Hamiltonian defines the differential time evolution, whereas the Lagrangian is a solution to this evolution. Consequently, initial conditions could in principle appear in the Lagrangian but not in the Hamiltonian. This results in aspects of the initial condition such as θ misleadingly appearing in the Lagrangian as parameters. We comment on the similarity between the θ vacua and the violations of the constraint equations of classical gauge theories in quantum mechanics.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗
DOE OSTI2020

Production and Fuel properties of Iso-Olefins with Controlled Molecular Structure and Obtained from Butene Oligomerization

The ability to control the molecular structure (e.g., degree of branching) of iso-olefins produced from oligomerizing light olefins is a valuable tool for tuning the final compositions of hydrocarbon fuels and targeting specific fuel properties. In this study, we demonstrated that the degree of branching of iso-olefins obtained from butene(s) oligomerization can be controlled by tuning process conditions (i.e., temperature, weight hourly space velocity [WHSV], and nature of butene feedstock) and by choosing the proper catalyst (i.e., Amberlyst-36 vs. Y/ZSM-22). In this study, we produced three types of iso-olefin mixtures: 1) a methyl-heptenes rich (74 wt.%) mixture, 2) dimethyl-hexenes rich (80–96 wt.%) mixtures, and 3) highly branched (i.e., more than three methyl substitutions) iso-olefins rich (>50 wt.%) mixtures. While dimethyl-hexenes are preferentially formed at lower temperatures (60–100°C) and WHSV (i.e., 2 hr-1), methyl-heptenes are favorably produced at higher temperatures (>100°C) and WHSVs (i.e., 7 hr-1) over Amberlyst-36. The use of either 1-butene or 2-butene as feedstock resulted in liquid products with similar branching because facile intramolecular isomerization occurs prior to oligomerization. However, the use of isobutylene feedstock forms a significantly more branched olefin product. The formation of highly branched iso-olefins also is favored at higher temperature (140°C) over Amberlyst-36. Using Y/ZSM-22 instead of Amberlyst-36 allowed preferential formation of less-branched methyl-heptenes. For each type of iso-olefins mixture, we determined fuel properties including research octane number (RON), motor octane number (MON), and octane sensitivity (S), which is the difference between RON and MON. We found that not only RON and MON but also S values increased with the degree of branching of these complex mixtures of iso-olefins. The highest RON of 99.7 and S value of 9.3 were obtained for a mixture of highly branched iso-olefins.

Dagle, Vanessa↗
DOE OSTI2024

Enhancement of Conjugated Polymer Microstructure and Mixed-Conducting Properties via Chalcogenophene Heteroatom Substitution

Heteroatom substitution is a powerful tool to tune the intra- and intermolecular structure of conjugated polymers as well as their resulting optoelectronic and electrochemical properties. Here, a series of oligoethylene glycol bithiophene chalcogenophene polymers (p(g3T2-X)) with systematically varied furan, thiophene, selenophene, and tellurophene comonomers have been synthesized for mixed ionic-electronic conducting applications. Their microstructures have been thoroughly characterized ex situ and in situ with X-ray scattering, and their mixed conducting properties have been probed in electrochemical transistor testbeds. Chalcogenophene heteroatom choice was found to clearly dictate the polymer microstructure (crystallite dimensionality and orientation) and tune mixed conducting properties. Proceeding down Group 16, from O to Se systematically directed the molecular ordering of 2D polymer crystallites from face-on (O) to mixed (S) to edge-on (Se) orientations, with Te driving the polymer to form well-oriented edge-on 3D crystallites. Heteroatom dictated crystallite quality, and orientation tuned relative ionic transport by 2 orders of magnitude. Hole mobility (μ hole ) and mixed conducting figure of merit (μC*) were each tuned over an order of magnitude depending on heteroatom choice, with the Te-containing polymer reaching μ hole = 3.60 cm 2 V –1 s –1 and μC* = 483 F cm –1 V –1 s –1 , due to improved molecular ordering. Insights from this polymer series highlight target microstructures for enhanced mixed conduction in future conjugated polymers.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
DOE OSTI2025

Autogenerating a Domain-Specific Question-Answering Data Set from a Thermoelectric Materials Database to Enable High-Performing BERT Models

We present a method for autogenerating a large domain-specific question-answering (QA) dataset from a thermoelectric materials database. We show that a small language model, BERT, once fine-tuned on this automatically generated dataset of 99,757 QA pairs about thermoelectric materials, affords better performance in the field of thermoelectric materials compared to a BERT model fine-tuned on the generic English-language QA data set, SQuAD-v2. We further show that mixing the two data sets (ours and SQuAD-v2), which have significantly different syntactic and semantic scopes, allows the BERT model to achieve even better performance. The best-performing BERT model fine-tuned on the mixed data set outperforms the models fine-tuned on the other two data sets by scoring an exact match of 67.93% and an F1 score of 72.29% when evaluated on our test data set. This has important implications as it demonstrates the ability to realize high-performing small language models, with modest computational resources, empowered by domain-specific materials data sets which can be generated according to our method.

biological databases↗
DOE OSTI2023

Cyclobutanedicarboxylate Metal–Organic Frameworks as a Platform for Dramatic Amplification of Pore Partition Effect

Ultrafine tuning of MOF structures at sub-angstrom or picometer levels can help improve separation selectivity for gases with subtle differences. However, for MOFs with large-enough pore size, the effect from ultrafine tuning on sorption can be muted. Here we show an integrative strategy that couples together extreme pore compression with ultrafine pore tuning. Furthermore, this strategy is made possible by unique combination of two features of the pacs platform (pacs = partitioned acs): multi-modular framework and exceptional tolerance towards isoreticular replacement. Specifically, we use one module (Ligand 1, L1) to shrink pore size to an extreme minimum on pacs. A compression ratio of about 30% was achieved (based on unit-cell c/a ratio) from prototypical 1,4-benzenedicarboxylate-pacs to trans-1,3-cyclobutanedicarboxylate-pacs. This is followed by using another module (Ligand 2, L2) for ultrafine pore tuning (< 3% compression). This L1-L2 strategy increases C 2 H 2 /CO 2 selectivity from 2.6 to 20.8 and gives rise to excellent experi-mental breakthrough performance. Being the shortest cyclic dicarboxylate that mimics para-benzene-based moieties us-ing bioisosteric (BIS) strategy on pacs, trans-1,3-cyclobutanedicarboxylate-based MOF chemistry offers new opportuni-ties in MOF chemistry.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
DOE OSTI2025

Reinforcement Learning‐Based Adaptation of Grid Following Inverter's Internal Controller to Networked Microgrids' Strengths

The varying topological configurations, generator commitments and dispatches, and dynamic load demand lead to changing system's strengths during the operations of networked microgrids. When the system's strengths significantly change, the fixed control gains at large devices may result in unsatisfactory system performance; this necessitates the tuning of the control gains at large devices to adapt to the changing system's strengths. In this paper, observer-based reinforcement learning (RL) is utilised to automatically tune the proportional-integral (PI) gains of phase lock loop (PLL) controller of grid-following (GFL) inverters to adapt to the changing strengths of microgrids and networked microgrids. The RL agent in this framework augments an observer predicting system's strengths, from which the RL control policy will adjust accordingly to tune the PLL controller's gains towards the system's strengths. Also, to enhance the control performance, the recently introduced Barrier function-based RL framework is leveraged for the design of reward function to prevent the high frequency nadir. An operational 26 kV electric distribution system, which is modelled as networked microgrids, is used to illustrate the need and effectiveness of the proposed RL-tuned control.

frequency response↗
DOE OSTI2026

Automated ICRF heating surrogate modeling via machine learning

This work introduces automated machine learning workflows that address critical bottlenecks in surrogate model development for Ion Cyclotron Range of Frequencies (ICRF) heating applications. The automated framework includes data analysis tools that transform raw datasets into actionable insights in seconds, replacing weeks of manual exploratory effort and ensuring consistent, reproducible dataset characterization. By integrating advanced hyperparameter optimization (HPO) methods including Bayesian optimization via BoTorch and Tree-structured Parzen Estimators (TPE), the framework significantly reduces model development time from weeks to hours, decreasing computational cost and required expertise, while enabling high-accuracy surrogate models. Compared to traditional hyperparameter scanning (HPS) techniques such as methodical, randomized, and grid searches, HPO methods achieve superior convergence and predictive performance, even when compared to already well-tuned reference models. On NSTX High Harmonic Fast Wave (HHFW) heating datasets, both Random Forest Regressor (RFR) and neural network surrogates demonstrate improved accuracy, achieving R 2 values beyond 0.97 and 0.98, respectively. The results show that while HPO gains are modest for robust architectures like RFR, they become essential for more sensitive models such as neural networks, highlighting the trade-offs across optimization strategies. Through automated workflows that eliminate manual hyperparameter tuning and require minimal ML expertise, this work enables widespread adoption of high-fidelity surrogate models across the fusion community for real-time plasma control, uncertainty quantification, rapid experimental scenario development, and integrated system optimization.

Sanchez-Villar, Alvaro [Princeton Plasma Physics L↗
DOE OSTI2023

Sterically attenuated electronic communication in cobalt complexes of meridional isoquinoline-derived ligands for applications in electrocatalysis

The ability to synthetically tune the ligand frameworks of redox-active molecules is of critical importance to the economy of solar fuels because manipulating their redox properties can afford control over the operating potentials of sustained electrocatalytic or photoelectrocatalytic processes. The electronic and steric properties of 2,2′:6′,2″-terpyridine (Terpy) ligand frameworks can be tuned by functional group substitution on ligand backbones, and these correlate strongly to their Hammett parameters. The synthesis of a new series of tridentate meridional ligands of 2,4,6-trisubstituted pyridines that engineers the ability to finely tune the redox potentials of cobalt complexes to more positive potentials than that of their Terpy analogs is achieved by aryl-functionalizing at the four-position and by including isoquinoline at the two- and six-positions of pyridine (Aryl-DiQ). Their cobalt complex syntheses, their electronic properties, and their catalytic activity for carbon dioxide (CO2) reduction are reported and compared to their Terpy analogs. The cobalt derivatives generally experience a positive shift in their redox features relative to the Terpy-based analogs, covering a complementary potential range. Although those evaluated fail to produce any quantifiable products for the reduction of CO2 and suffer from long-term instability, these results suggest possible alternate strategies for stabilizing these compounds during catalysis. We speculate that lower equilibrium association constants to the cobalt center are intrinsic to these ligands, which originate from a steric interaction between protons on the pyridine and isoquinoline moieties. Nevertheless, the new Aryl-DiQ ligand framework has been engineered to selectively tune homoleptic cobalt complexes’ redox potentials.

Chemistry↗
DOE OSTI2024

In situ electric-field control of ferromagnetic resonance in the low-loss organic-based ferrimagnet V[TCNE] x ∼2

We demonstrate indirect electric-field control of ferromagnetic resonance (FMR) in devices that integrate the low-loss, molecule-based, room-temperature ferrimagnet vanadium tetracyanoethylene (V[TCNE]x∼2) mechanically coupled to PMN-PT piezoelectric transducers. Upon straining the V[TCNE]x films, the FMR frequency is tuned by more than 6 times the resonant linewidth with no change in Gilbert damping for samples with α = 6.5 × 10−5. We show this tuning effect is due to a strain-dependent magnetic anisotropy in the films and find the magnetoelastic coefficient |λs| ∼ (1–4.4) ppm, backed by theoretical predictions from density-functional theory calculations and magnetoelastic theory. Noting the rapidly expanding application space for strain-tuned FMR, we define a new metric for magnetostrictive materials, magnetostrictive agility, given by the ratio of the magnetoelastic coefficient to the FMR linewidth. This agility allows for a direct comparison between magnetostrictive materials in terms of their comparative efficacy for magnetoelectric applications requiring ultra-low loss magnetic resonance modulated by strain. With this metric, we show V[TCNE]x is competitive with other magnetostrictive materials, including YIG and Terfenol-D. This combination of ultra-narrow linewidth and magnetostriction, in a system that can be directly integrated into functional devices without requiring heterogeneous integration in a thin film geometry, promises unprecedented functionality for electric-field tuned microwave devices ranging from low-power, compact filters and circulators to emerging applications in quantum information science and technology.

Materials Science↗
DOE OSTI2025

Inertial Confinement Fusion Design Search Using Bayesian Optimization

Inertial confinement fusion (ICF) experiments rely on complex multi-physics simulation codes such as the Lawrence Livermore National Laboratory-developed HYDRA to guide design work. However, these simulations have several dozen tunable parameters and can be computationally expensive. This makes searching the parameter space challenging and time-consuming. Recently developed automated tools utilize Bayesian optimization to search these high-dimensional parameter spaces for optimal designs. The optimization tools run 2D integrated simulations in HYDRA to converge on a design that produces specified scalar or vector outputs. In this paper, we apply the Bayesian optimization tools to two common tuning scenarios. First, we tune simulation inputs to match measurements of a well-characterized experiment at the National Ignition Facility. This type of tuning is commonly performed to compensate for the use of simplified simulation settings (e.g. reduced resolution) or to account for missing physics in the simulations. Second, we search for an ICF simulation design that has a particular radiation drive profile. These optimizations replicate the kinds of tuning researchers routinely perform, but do so with significantly reduced manual effort. This approach demonstrates a powerful and efficient pathway toward autonomous, high-fidelity design optimization for future ICF experiments.

Bayesian optimization↗
DOE OSTI2022

Practical galaxy morphology tools from deep supervised representation learning

Astronomers have typically set out to solve supervised machine learning problems by creating their own representations from scratch. We show that deep learning models trained to answer every Galaxy Zoo DECaLS question learn meaningful semantic representations of galaxies that are useful for new tasks on which the models were never trained. We exploit these representations to outperform several recent approaches at practical tasks crucial for investigating large galaxy samples. The first task is identifying galaxies of similar morphology to a query galaxy. Given a single galaxy assigned a free text tag by humans (e.g. ‘#diffuse’), we can find galaxies matching that tag for most tags. The second task is identifying the most interesting anomalies to a particular researcher. Our approach is 100 per cent accurate at identifying the most interesting 100 anomalies (as judged by Galaxy Zoo 2 volunteers). The third task is adapting a model to solve a new task using only a small number of newly labelled galaxies. Models fine-tuned from our representation are better able to identify ring galaxies than models fine-tuned from terrestrial images (ImageNet) or trained from scratch. We solve each task with very few new labels; either one (for the similarity search) or several hundred (for anomaly detection or fine-tuning). This challenges the longstanding view that deep supervised methods require new large labelled data sets for practical use in astronomy. To help the community benefit from our pretrained models, we release our fine-tuning code zoobot. Zoobot is accessible to researchers with no prior experience in deep learning.

79 ASTRONOMY AND ASTROPHYSICS↗