Awesome
LaMEM.jl
This is the Julia interface to LaMEM) (Lithosphere and Mantle Evolution Model), which is the easiest way to install LaMEM on any system. It allows you to start a (parallel) LaMEM simulation, and read back the output files to julia for further processing. Below we give some brief steps in how to use it. More examples can be found in the user guide.
1. Installation
Go to the package manager & install it with:
julia>]
pkg>add LaMEM
It will automatically download a binary version of LaMEM which runs in parallel (along with the correct PETSc version). This will work on linux, mac and windows. If you want to check that it works on your machine type:
pkg>test LaMEM
which will run the build-in testsuite.
2. Create a model setup & run LaMEM
You can directly create a LaMEM setup in julia with:
julia> using LaMEM, GeophysicalModelGenerator
julia> model = Model(Grid(nel=(16,16,16), x=[-1,1], y=[-1,1], z=[-1,1]))
LaMEM Model setup
|
|-- Scaling : GeoParams.Units.GeoUnits{GEO}
|-- Grid : nel=(16, 16, 16); xϵ(-1.0, 1.0), yϵ(-1.0, 1.0), zϵ(-1.0, 1.0)
|-- Time : nstep_max=50; nstep_out=1; time_end=1.0; dt=0.05
|-- Boundary conditions : noslip=[0, 0, 0, 0, 0, 0]
|-- Solution parameters : eta_min=1.0e18; eta_max=1.0e25; eta_ref=1.0e20; act_temp_diff=0
|-- Solver options : direct solver; mumps; penalty term=10000.0
|-- Model setup options : Type=files;
|-- Output options : filename=output; pvd=1; avd=0; surf=0
|-- Materials : 0 phases;
Add materials to the setup:
julia> matrix = Phase(ID=0,Name="matrix",eta=1e20,rho=3000);
julia> sphere = Phase(ID=1,Name="sphere",eta=1e23,rho=3200)
Phase 1 (sphere):
rho = 3200.0
eta = 1.0e23
julia> add_phase!(model, sphere, matrix)
Create an initial geometry using the GeophysicalModelGenerator interface:
julia> add_sphere!(model,cen=(0.0,0.0,0.0), radius=0.5)
and run the simulation with:
julia> run_lamem(model,1)
Saved file: Model3D.vts
Writing LaMEM marker file -> ./markers/mdb.00000000.dat
--------------------------------------------------------------------------
Lithosphere and Mantle Evolution Model
Compiled: Date: Apr 7 2023 - Time: 22:11:23
Version : 1.2.4
--------------------------------------------------------------------------
STAGGERED-GRID FINITE DIFFERENCE CANONICAL IMPLEMENTATION
--------------------------------------------------------------------------
Parsing input file : output.dat
Finished parsing input file : output.dat
--------------------------------------------------------------------------
...
Note that if you have a linux/mac machine you can run it in parallel (change 1 to 2 or 4, for example). On windows you would have to install Linux for Windows first, using WSL. Once the simulation is done, you can open it with Paraview, or directly plot it within julia (see the documentation).
3. Starting a simulation
The julia way of running LaMEM simulations, described above, is the preferred way (see also the documentation). In the background it will create a LaMEM (*.dat) input file, along with an initial marker setup. If you have such as in input file already, you can run that in parallel (here on 4 cores) with:
julia> using LaMEM
julia> ParamFile="input_files/FallingBlock_Multigrid.dat";
julia> run_lamem(ParamFile, 4,"-time_end 1")
--------------------------------------------------------------------------
Lithosphere and Mantle Evolution Model
Compiled: Date: Sep 10 2022 - Time: 06:21:30
--------------------------------------------------------------------------
STAGGERED-GRID FINITE DIFFERENCE CANONICAL IMPLEMENTATION
--------------------------------------------------------------------------
Parsing input file : input_files/FallingBlock_Multigrid.dat
Adding PETSc option: -snes_type ksponly
Adding PETSc option: -js_ksp_monitor
Adding PETSc option: -crs_pc_type bjacobi
Finished parsing input file : input_files/FallingBlock_Multigrid.dat
--------------------------------------------------------------------------
Time stepping parameters:
Simulation end time : 1. [ ]
Maximum number of steps : 10
Time step : 10. [ ]
Minimum time step : 1e-05 [ ]
Maximum time step : 100. [ ]
Time step increase factor : 0.1
CFL criterion : 0.5
CFLMAX (fixed time steps) : 0.5
Output time step : 0.2 [ ]
Output every [n] steps : 1
Output [n] initial steps : 1
--------------------------------------------------------------------------
The last parameter are optional PETSc command-line options. By default it runs on one processor.
Please note that you will have to be in the correct directory or indicate where that directory is. If you are in a different directory, the easiest way to change to the correct one is by using the changefolder function (on Windows and Mac):
julia> changefolder()
Alternatively, you can use the build-in terminal/shell in julia, which you can access with:
julia>;
shell>cd ~/LaMEM/input_models/BuildInSetups/
use the Backspace key to return to the julia REPL.
Once you have performed a simulation, you can look at the results by opening the *.pvd files with Paraview. In this example, that would be FB_multigrid.pvd and FB_multigrid_phase.pvd.
4. Reading LaMEM output back into julia
If you want to quantitatively do something with the results, there is an easy way to read the output of a LaMEM timestep back into julia. All routines related to that are part of the LaMEM.IO module.
julia> using LaMEM
You can first read the *.pvd file in the directory to see which timesteps are available. If you used julia to run the simulation (as under 2 above ), this is done with:
julia> Timestep, Filenames, t = read_LaMEM_simulation(model)
([0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13], ["Timestep_00000000_0.00000000e+00/output.pvtr", "Timestep_00000001_4.40000000e-02/output.pvtr", "Timestep_00000002_9.24000000e-02/output.pvtr", "Timestep_00000003_1.45640000e-01/output.pvtr", "Timestep_00000004_2.04204000e-01/output.pvtr", "Timestep_00000005_2.68624400e-01/output.pvtr", "Timestep_00000006_3.39486840e-01/output.pvtr", "Timestep_00000007_4.17435524e-01/output.pvtr", "Timestep_00000008_5.03179076e-01/output.pvtr", "Timestep_00000009_5.97496984e-01/output.pvtr", "Timestep_00000010_7.01246682e-01/output.pvtr", "Timestep_00000011_8.15371351e-01/output.pvtr", "Timestep_00000012_9.40908486e-01/output.pvtr", "Timestep_00000013_1.07899933e+00/output.pvtr"], [0.0, 0.044, 0.0924, 0.14564, 0.204204, 0.2686244, 0.3394868, 0.4174355, 0.5031791, 0.597497, 0.7012467, 0.8153714, 0.9409085, 1.078999])
If you instead have an existing LaMEM simulation, you can specify the *.pvd file:
julia> pvdname="output"
julia> Timestep, Filenames, t = read_LaMEM_simulation(pvdname)
We can read a particular timestep (say 1) with:
julia> data, time = read_LaMEM_timestep(model, 1)
(CartData
size : (17, 17, 17)
x ϵ [ -1.0 : 1.0]
y ϵ [ -1.0 : 1.0]
z ϵ [ -1.0 : 1.0]
fields : (:phase, :density, :visc_total, :visc_creep, :velocity, :pressure, :temperature, :j2_dev_stress, :j2_strain_rate)
attributes: ["note"]
, [0.044])
The output is in a CartData structure (as defined in GeophysicalModelGenerator).
More details are given in the documentation.
5. Dependencies
We rely on the following packages:
- GeophysicalModelGenerator - Data structure in which we store the info of a LaMEM timestep. The package can also be used to generate setups for LaMEM.
- LaMEM_jll - this contains the LaMEM binaries, precompiled for most systems. It also contains a precompiled version of PETSc, along with MPI. Note that on windows, MPI does not work, so you can only use one processor. We therefore recommend that you install linux on windows (using WSL) and run LaMEM through that.
- ReadVTK - This reads the LaMEM
*.vtkfiles (or the rectilinear and structured grid versions of it) baxck into julia.
6. Funding
Funding for this julia interface has been provided by the European Research Council (ERC CoG MAGMA # 771143), and by the EuroHPC-JU Center of Excellence CHEESE-2P.