This package originates from the non-trivial to use ep2-3De v1.0, and it is fully witten in Julia. It aims at fast prototyping with decent production capabilities.
It addresses the following key aspects throughout a modern and easy-to-use MPM framework that considers:
- solves elastoplastic problems under the following:
- an Updated Lagrangian explicit formulation
- a finite or infinitesimal deformation framework; adopting logarithmic strains and Kirchoff stresses for finite deformation and Jaumann rate formulation for infinitesimal deformation
- uses the following shape function basis:
- standard linear shape function
$N_n(\mathbf{x}_p)$ - GIMP shape function
$S_n(\mathbf{x}_p)$ - boundary modified cubic B-spline shape function
$\phi_n(\mathbf{x}_p)$
- standard linear shape function
- uses the following mapping between nodes (denoted
$n$ or$v$ ) and material points (denoted$p$ )- FLIP with augmented mUSL procedure
- TPIC with standard USL procedure
The solver relies on random gaussian fields to generate initial fields
In the context of High-Performance Computing (HPC), it’s essential to understand the different tiers of performance based on the computational resources available:
- standard: This tier is characterized by single-core CPU usage, suitable for basic tasks like light simulations and data processing.
- moderate: This level utilizes multi-core CPUs and a single GPU, allowing for medium-scale simulations and machine learning tasks.
- high Performance: This tier represents the use of multi-node systems with multiple CPUs and GPUs, ideal for large-scale simulations and deep learning applications.
The stylized term
-
(opt.) Get Julia here and follow instructions for installation
-
Clone
elastoPlasm.jlandcdto your local repo -
Launch Julia (on macOS, drag & drop
start_macOS.shin the terminal) and enter pkg mode], thenactivate .the projectelastoPlasmandinstantiateits environment and related packages. -
Once
elastoPlasmhas been correctly instantiated, you canusing elastoPlasm