In situ data-driven adaptive sampling for large-scale simulation data summarization

Multivariate Pointwise Information-Driven Data Sampling and Visualization

“…In this section, we briefly discuss existing generic sampling techniques and then introduce the proposed multivariate association-driven sampling algorithm. Statistical data sampling has been shown to be effective in summarizing and analyzing data sets by researchers in the past [ 22 , 63 ]. One key advantage of sampling-based data summarization over sophisticated data modeling-based approaches is that the sampling techniques keep a true subset of representative points selected from the original raw data.…”

“…This technique is called regular sampling. Regular sampling does not consider any data properties while selecting samples and due to the regular nature of sample selection, it produces artifacts and discontinuities during sample-based visual analysis [ 17 , 22 ].…”

“…Use of information theory has also been an important direction for researchers [ 56 , 57 , 58 ] while finding the optimal subset of the original data. Biswas et al [ 22 ] used similar ideas of entropy maximization for inventing an in situ data-driven sampling scheme that can preserve important properties of the data. They used probability distribution of the original datasets to find important samples in a generic sampling scheme.…”

“…In this work, we propose a multivariate data summarization technique that selects a small fraction of the original data set (i.e., subsamples the data) considering multiple variables simultaneously and picks data points with higher fidelity from the regions where the selected variables show a strong association. The benefits of sampling-based data representations for a single variable has been shown by the researchers in the past [ 22 , 23 , 24 , 25 ]. In this work, we pursue a novel sampling-based data summarization scheme and extend the sampling capability to the multi-variable domain.…”

Priority research directions for in situ data management: Enabling scientific discovery from diverse data sources

“…In situ analysis algorithms may transform data into reduced representations or surrogate models in order to mitigate large data size, high dimensionality, or long computation times. Low-rank approximation (Austin et al, 2016), statistical summarization (Biswas et al, 2018;Dutta et al, 2017;Hazarika et al, 2018;Lawrence et al, 2017;Lohrmann et al, 2017;Thompson et al, 2011), topological segmentation (Gyulassy et al, 2012(Gyulassy et al, , 2019Landge et al, 2014;Weber, 2013, 2014), wavelet transformation (Li et al, 2017;Salloum et al, 2018), lossy compression (Brislawn et al, 2012;Di and Cappello, 2016;Lindstrom, 2014), geometric modeling (Nashed et al, 2019; Peterka et al, 2018), and feature detection (Guo et al, 2017) may be used to generate reduced or surrogate models.…”