MeshAC
A 3D Mesh Generation and Adaptation Package for Multiscale Coupling Simulation for Materials Defects
Install / Use
/learn @kjfu/MeshACREADME
MeshAC
A 3D Mesh Generation and Adaptation Package for Multiscale (atomistic-to-continuum) Coupling Simulation for Materials Defects. Written by Kejie Fu, Mingjie Liao and Yangshuai Wang.
Getting Started
Installation
You can install MeshAC with the following steps:
- Clone the repository into your local machine:
git clone https://github.com/kjfu/MeshAC.git
- Compile the code using cmake:
cd MeshAC
mkdir build
cd build
cmake ..
make
cp ./build/MeshAC ./MeshAC
Command Line Switches
MeshAC supports usage via command line. Here is an overview of all command line switches:
Usage: ./MeshAC [OPTIONS] input [output]
Postionals:
input TEXT REQUIRED Input a file of initial mesh(.mesh) for mesh generation or middle files
(.mesh, .remesh and .value) for mesh adaptation.(string, required)
output TEXT Output a file of resulting mesh (.mesh) or a file of interpolation solutions
(.value).
Options:
-h Print this help message and exit.
-i TEXT REQUIRED Input a file of initial mesh(.mesh) for mesh generation or middle files
(.mesh, .remesh and .value) for mesh adaptation.(string, required)
-o TEXT Output a file of resulting mesh (.mesh) or a file of interpolation solutions
(.value).
-s FLOAT Input the max sizing value for mesh generation.
-hd Generate a mesh with edge dislocation.
-r Refine an existing mesh adaptively.
-rr Refine an existing mesh with edge dislocation adaptively.
Examples
To generate 3d mesh from points（with 8 points as bounding points with label 1, and several atomistic points with label 0）
>> ./MeshAC -s 5 -i sample.mesh -o outmesh.mesh
To generate a 3d mesh from points with edge dislocation
>>./MeshAC -hd -i test3d.mesh -s 15 -o out3d.mesh
To remesh a 3d mesh adaptively
You must keep 3 files (*.mesh, *.remesh, *.value) in same path.
>>./MeshAC -r -i test3d -o out3d
To remesh a 3d mesh with edge dislocation adaptively
You must keep 3 files(*.mesh, *.remesh, and *.value) in same path
>>./MeshAC -rr -i test3d -o out3d
Label Meanings in .mesh file
Labels for nodes
| Label | Significance |Tip| |:------|:-------|:-----| |0|Nodes inside the atomic area|| |1|Nodes on the border of the continuous area|| |2|Nodes on the border of the atomic area|| |3|Nodes between the border of the continuous area and the border of the atomic area||
Labels for tets
| Label | Significance |Tip| |:------|:-------|:-----| |0|Tets of the atomic area|| |1|Tets of the continuous area||
<!-- ## Overview We now summarize the main components of the library. 1. Mesh generation 2. Mesh adaptation 3. ... The following functionals are currently supported: - [`...`](...) .... From Mesher3DForSJTU package ... -->A/C coupling method in Julia
Atomistic-to-continuum coupling method in Julia. The current implementation is based on the BGFC (Blended Ghost Force Correction) method.
The atomistic computations involved are heavily depends on the pure Julia package JuLIP (Julia Library for Interatomic Potentials).
- Install Julia 1.10.1 from here;
- Install necessary registry and then open a new activate environment:
] registry add https://github.com/JuliaRegistries/General";
] registry add https://github.com/ACEsuit/ACEregistry";
] activate .
- Run the following Julia command to install the required packages:
] add JuLIP, DelimitedFiles, Printf, NeighbourLists, QHull, Optim, LineSearches, SparseArrays, Isaac, PyCall, Plots, ASE, ScatteredInterpolation
Tips:
- Use
Pkg.activate(".")to use a local project and set environment variableJULIA_PROJECTaccordingly. - Please do remember to modify the path of mesher3d in AtC constructor (AtC.jl).
FIO
The module named ACFIO is used to cope with the geometrical operations.
The functionals of FIO are to read/write:
- .mesh
- .remesh
- .value
and write .dump files for visualization.
AtC
Major struct contains geometry and computational information.
To construct an AtC objective invokes:
function AtC(Ra::Int64, bw::Int64, Lmsh, h; Rbuf=2, sp=:W, r0=rnn(:W), defects=:SingVac, meshpath="YOUR PATH")
Note that meshpath is the path of mesher toolkit that may differ from each devices.
Example
Step 1: Construct the atomistic model in atomistic region:
] activate .
include("./JuliaAC/dihole.jl")
# construct atomistic region with di-hole defects
# L: repeated size (keep it sufficiently large); Lb: the inner ball size; Lw: the width of torus.
L = 20; Lb = 13.0; Lw = 16.0;
atdef = get_atdef(L, Lb, Lw);
Step 2: Write atomistic mesh:
Xat = positions(atdef)
cb = [-1 1 1 -1 -1 1 1 -1; -1 -1 1 1 -1 -1 1 1; -1 -1 -1 -1 1 1 1 1]
Xcb = (L + 20) .* cb
X = hcat(mat(Xat), Xcb)
Xtype = zeros(Int64, length(atdef))
append!(Xtype, ones(Int64, 8))
fn = "./data/dihole_A.mesh"
# X: atom positions and outer continuum points; Xtype: interface information
ACFIO.write_mesh(fn, X, Xtype)
Step 3: Build AtC mesh and read the coupled mesh in Julia:
# call `mesher3d` to build coupled mesh
ofn = "./data/dihole_AC.mesh"
# the density of nodes in continuum region
h = 10
run(`./MeshAC -s $h -i $fn -o $ofn`)
Step 4: Construct a/c coupling structure:
# X: nodes; T: mesh topology
X, T = ACFIO.read_mesh(ofn)
iBdry = findall(x->x==1.0, X[4,:])
XType = X[4,:]
nat = [findall(x->x==0.0, XType); findall(x->x==2.0, XType);]
Xat = deepcopy(X[1:3, nat])
set_positions!(atdef, Xat)
# alternative data
data = Dict{String, Real}()
U = zeros(3, size(X, 2))
∇U, volT, J = gradient(T, X, U)
wat = bulk(:W, cubic = true)
V0 = det(cell(wat))/2
# final a/c stucture
atc = AtC{eltype(X)}(atdef, V0, X[1:3, :], X[4, :], U, ∇U, T[1:4, :], T[5,:], 3, 2, J, wat, iBdry, data);
Step 5: Alternatively, one can call the following command directly:
] activate .
include("./JuliaAC/dihole.jl")
# collect model parameters
L = 20; Lb = 13.0; Lw = 16.0;
Lc = 20; h = 10;
input_name = "./data/double-voids.mesh"
output_name = "./data/double-voids-coupled.mesh"
# construct atomistic region with defects
atdef = get_atdef(L, Lb, Lw)
# construct a/c coupling structure
atc = AtC_di(atdef, h, L, Lc; fn = input_name, ofn = output_name);
Development
MeshAC is under active development. Please don't hesitate to open feature requests to help us guide development. We more than welcome contributions!
Publications
MeshAC has been used in the following publications.
- Adaptive Multigrid Strategy for Geometry Optimization of Large-Scale Three Dimensional Molecular Mechanics (J. Comp. Phys. 2023)<br> K. Fu, M. Liao, Y. Wang, J. Chen and L. Zhang
Citation
If you use the codes in a publication, please cite the repo using the .bib,
@misc{fu2024meshac,
title={MeshAC: A 3D Mesh Generation and Adaptation Package for Multiscale Coupling Methods},
author={Kejie Fu and Mingjie Liao and Yangshuai Wang and Jianjun Chen and Lei Zhang},
eprint={2402.09446},
archivePrefix={arXiv},
year={2024}
}
Acknowledgements
We used several useful libraries in our implement and testing listed as follows. We would like to especially thank their authors for their great work and publishing the code.
