Scalable in situ compression of transient simulation data using time-dependent bases

Large-scale simulations of time-dependent problems generate a massive amount data and with the explosive increase in computational resources size generated by these has increased significantly. This imposed severe limitations on that can be stored elevated issue input/output (I/O) into one major bottlenecks high-performance computing. In this work, we present an situ compression technique to reduce storage orders magnitude. methodology is based subspaces it extracts low-rank structures from multidimensional streaming decomposing set bases core tensor. We derive closed-form evolution equations for tensor as well bases. The presented does not require history cost its extractions scales linearly — making suitable large-scale datasets. To control error, adaptive strategy add/remove modes maintain reconstruction error below given threshold. four demonstration cases: (i) analytical example, (ii) incompressible unsteady reactive flow, (iii) stochastic turbulent (iv) three-dimensional channel flow.