Informatics for combinatorial materials science

Broderick, Scott; Suh, Changwon; Nowers, Joseph R.; Vogel, Brandon M.; Mallapragada, Surya K.; Narasimhan, Balaji; Rajan, Krishna

doi:10.1007/s11837-008-0035-x

Cited by 28 publications

(17 citation statements)

References 16 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…There has been extensive previous work focusing on the application of Data Science and Analytics to material data sets. The field of Material Informatics (cf., e. g., [30,26,11,28,25,27,29,10,14,15]) uses data searching and sorting techniques to survey large material data sets. It also uses machine-learning regression [9,34] and other techniques to identify patterns and correlations in the data for purposes of combinatorial materials design and selection.…”

Section: Materials Informaticsmentioning

confidence: 99%

Data Driven Computing with noisy material data sets

Kirchdoerfer

Ortiz

2017

Computer Methods in Applied Mechanics and Engineering

180

181

View full text Add to dashboard Cite

We formulate a Data Driven Computing paradigm, termed maxent Data Driven Computing, that generalizes distance-minimizing Data Driven Computing and is robust with respect to outliers. Robustness is achieved by means of clustering analysis. Specifically, we assign data points a variable relevance depending on distance to the solution and on maximum-entropy estimation. The resulting scheme consists of the minimization of a suitably-defined free energy over phase space subject to compatibility and equilibrium constraints. Distanceminimizing Data Driven schemes are recovered in the limit of zero temperature. We present selected numerical tests that establish the convergence properties of the max-ent Data Driven solvers and solutions.

show abstract

Section: Materials Informaticsmentioning

confidence: 99%

Data Driven Computing with noisy material data sets

Kirchdoerfer

Ortiz

2017

Computer Methods in Applied Mechanics and Engineering

180

181

View full text Add to dashboard Cite

show abstract

“…While the results for a larger number of grain boundaries and more complex grain boundaries i.e., mixed tilt-twist) could also be added or another Fe interatomic potential could be used, the trends calculated within the present study provide both qualitative and quantitative understanding of grain boundaries acting as sinks for point defects. There are still a number of avenues for future work, e.g., the influence of temperature/entropy [39][40][41] , the influence of strain, the multiplicity of grain boundary structures [52][53][54][55] , uncertainty in results due to the interatomic potential development process 111,112 , or even data mining/informatics approaches for creating knowledge from the present simulations [113][114][115][116][117] , etc. We leave these avenues for future studies.…”

Section: A Examining the Structures And Energies Of Symmetric Tilt Gmentioning

confidence: 99%

Probing grain boundary sink strength at the nanoscale: Energetics and length scales of vacancy and interstitial absorption by grain boundaries inα-Fe

et al. 2012

View full text Add to dashboard Cite

The energetics and length scales associated with the interaction between point defects (vacancies and self-interstitial atoms) and grain boundaries in BCC Fe was explored. Molecular statics simulations were used to generate a grain boundary structure database that contained ≈ 170 grain boundaries with varying tilt and twist character. Then, vacancy and self-interstitial atom formation energies were calculated at all potential grain boundary sites within 15Å of the boundary. The present results provide detailed information about the interaction energies of vacancies and selfinterstitial atoms with symmetric tilt grain boundaries in iron and the length scales involved with absorption of these point defects by grain boundaries. Both low and high angle grain boundaries were effective sinks for point defects, with a few low-Σ grain boundaries (e.g., the Σ3{112} twin boundary) that have properties different from the rest. The formation energies depend on both the local atomic structure and the distance from the boundary center. Additionally, the effect of grain boundary energy, disorientation angle, and Σ-designation on the boundary sink strength was explored; the strongest correlation occurred between the grain boundary energy and the mean point defect formation energies. Based on point defect binding energies, interstitials have ≈ 80% more grain boundary sites per area and ≈ 300% greater site strength than vacancies. Last, the absorption length scale of point defects by grain boundaries is over a full lattice unit larger for interstitials than for vacancies (mean of 6-7Å vs. 10-11Å for vacancies and interstitials, respectively).

show abstract

“…Existing encoders for material representations can be categorized as physical and statistical, some of which have led to accelerated design of various material systems [8,9,10,11,12]. Among all, physical encoders characterize microstructures using composition (e.g., the percentage of each material constituent) [13,14], dispersion (e.g., inclusions' spatial relation, pair correlation,the ranked neighbor distance [15,16,17,18,19,20]), and geometry features (e.g., the radius/size distribution, roundness, eccentricity, and aspect ratio of elements of the microstructure [17,15,8,21,22,23,24,25]). Among statistical encoders are the N-point correlation functions [26,18,8,21,22].…”

Section: Data Science Challenges In Computational Materials Sciencementioning

confidence: 99%

Improving direct physical properties prediction of heterogeneous materials from imaging data via convolutional neural network and a morphology-aware generative model

Cang

Yao

et al. 2018

Computational Materials Science

166

View full text Add to dashboard Cite

Direct prediction of material properties from microstructures through statistical models has shown to be a potential approach to accelerating computational material design with large design spaces. However, statistical modeling of highly nonlinear mappings defined on highdimensional microstructure spaces is known to be data-demanding. Thus, the added value of such predictive models diminishes in common cases where material samples (in forms of 2D or 3D microstructures) become costly to acquire either experimentally or computationally. To this end, we propose a generative machine learning model that creates an arbitrary amount of artificial material samples with negligible computation cost, when trained on only a limited amount of authentic samples. The key contribution of this work is the introduction of a morphology constraint to the training of the generative model, that enforces the resultant artificial material samples to have the same morphology distribution as the authentic ones. We show empirically that the proposed model creates artificial samples that better match with the authentic ones in material property distributions than those generated from a state-of-the-art Markov Random Field model, and thus is more effective at improving the prediction performance of a predictive structure-property model.

show abstract

Informatics for combinatorial materials science

Cited by 28 publications

References 16 publications

Data Driven Computing with noisy material data sets

Data Driven Computing with noisy material data sets

Probing grain boundary sink strength at the nanoscale: Energetics and length scales of vacancy and interstitial absorption by grain boundaries inα-Fe

Improving direct physical properties prediction of heterogeneous materials from imaging data via convolutional neural network and a morphology-aware generative model

Contact Info

Product

Resources

About