The Challenge of Materials with Memory
Classical fluid dynamics relies heavily on the Newtonian assumption: the stress at any point in a fluid depends strictly on its instantaneous velocity gradient. If you stop shearing water, its viscous stress drops to zero immediately.
However, many real-world materials do not behave this way. Complex fluids—such as polymer melts, biological tissues, and discontinuous shear-thickening (DST) suspensions—possess "memory." As Deng et al. note, polymeric liquids contain long-chain molecules that stretch, entangle, and recoil over time. This creates a complex coupling between macroscopic flow variables and microscopic molecular distributions.
Because these internal structures take time to relax, the material's current stress state depends on its entire history of past deformations. This history-dependent behavior isn't limited to liquids; even solid-like geological structures like tight sandstone exhibit unexpected viscoelastic relaxation over time, as demonstrated by Ding et al..
To model these materials accurately, researchers replace classical Newtonian laws with viscoelastic constitutive equations, such as the Maxwell or fractional Maxwell models discussed by Hu. These models mathematically express stress as a time-integral of past strain history.
The Eulerian Bottleneck
To simulate a fluid with memory, a computer must track the deformation history of every single fluid element over time. In a traditional Eulerian simulation framework, this is incredibly difficult.
Eulerian methods look at the world through a fixed spatial grid, observing fluid properties as they pass through specific coordinates. If you need to compute the history-dependent stress at a specific grid point right now, you have to figure out where that specific parcel of fluid came from. This requires "back-tracking" the fluid's trajectory through time across the grid.
As a fluid twists, turns, and shears, back-tracking becomes a computational nightmare. It demands complex interpolation algorithms, massive data storage, and intense bookkeeping to map past states onto the current grid configuration.
Particles as Moving Ledgers
The core insight behind Lagrangian history tracking is to change the frame of reference entirely. Instead of watching fluid pass through a fixed grid, the fluid is discretized into moving particles using mesh-free methods like Smoothed Particle Hydrodynamics (SPH).
In a Lagrangian framework, particles move with the flow. This physical tracking solves the history problem elegantly: each particle naturally acts as its own local ledger.
Because a particle represents a distinct, moving element of the material, any microstructural state or deformation history can be stored directly as an attribute of that particle. As it flows through complex geometries, it carries its history along with it.
As Santelli et al. (2023) explain, this inherent bookkeeping completely eliminates the need for complex back-tracking algorithms. To evaluate an integral constitutive model, the simulation simply reads the historical data already sitting inside the particle.
Microstructural and Multiscale Tracking
This particle-centric approach scales seamlessly from macroscopic fluid dynamics down to microstructural interactions. For instance, Moreno et al. developed a Generalized Lagrangian Heterogeneous Multiscale Method (LHMM). By attaching microscopic relaxation states directly to moving macroscopic particles, the authors ensured that past deformation history correctly influences the current rheological response, even when microscopic relaxation times are much longer than macroscopic timescales.
Similarly, for complex phenomena like Discontinuous Shear-Thickening (DST) fluids—materials like cornstarch and water that solidify upon sudden impact—Angerman et al. used a Lagrangian framework to track how local microstructures evolve along previous flow paths. This allows the simulation to capture the sudden, history-dependent transitions from liquid to solid behavior.
From Theory to Physical Reality
By transforming a global tracking problem into local particle bookkeeping, researchers can now simulate highly non-linear, real-world processes that were previously intractable.
For example, Santelli et al. (2025) combined SPH with a fractional integral K-BKZ constitutive model to simulate polymer melts under Large Amplitude Oscillatory Shear (LAOS). Because the particles maintained their own deformation history over time, the simulation successfully matched experimental data for polypropylene under high-deformation industrial manufacturing conditions.
Understanding this history tracking also sheds light on emergent physical behaviors in complex environments. Puljiz et al. demonstrated that rigid particles embedded in a viscoelastic medium experience long-ranged, history-dependent interactions. Because the surrounding matrix remembers past deformations and relaxes slowly, the medium continues to transmit mediated forces to neighboring particles long after the initial motion has stopped.
Ultimately, Lagrangian history tracking works because it aligns the computational data structure with the physical reality of the fluid: memory belongs to the material element, so the data belongs on the moving particle.
Key Foundational Work
Memory-based mediated interactions between rigid particulate inclusions in viscoelastic environments (2019) — contributed 2 concept elements to this direction.
Unexpected viscoelastic deformation of tight sandstone: Insights and predictions from the fractional Maxwell model (2017) — contributed 1 concept elements to this direction.