laboetie is a computational fluid dynamics code for chemical applications.
It uses the Lattice-Boltzmann algorithm.
[more to come]
It is written by Maximilien Levesque¹ and Benjamin Rotenberg²
¹ École Normale Supérieure and CNRS, Paris, France
² Sorbonne Universités, UPMC Univ. Paris 06, CNRS, UMR 8234, PHENIX, F-75005 Paris, France
We are researchers: our work is evaluated on the basis of citations to our publications. If you use Laboetie, please cite us:
doi: 10.1103/PhysRevE.88.013308
- Accounting for adsorption and desorption in lattice Boltzmann simulations
Maximilien Levesque, Magali Duvail, Ignacio Pagonabarraga, Daan Frenkel and Benjamin Rotenberg
Phys. Rev. B 88, 013308 (2013)
http://dx.doi.org/10.1103/PhysRevE.88.013308 - Unexpected coupling between flow and adsorption in porous media
Jean-Mathieu Vanson, François-Xavier Coudert, Benjamin Rotenberg, Maximilien Levesque, Caroline Tardivat, Michaela Klotz and Anne Boutin
Soft Matter 11, 6125-6133 (2015)
http://dx.doi.org/10.1039/C5SM01348H - Computational Chemistry Methods for Nanoporous Materials
Jack D. Evans, Guillaume Fraux, Romain Gaillac, Daniela Kohen, Fabien Trousselet, Jean-Mathieu Vanson and François-Xavier Coudert
Chemistry of Materials 29, 199 (2017)
http://dx.doi.org/10.1021/acs.chemmater.6b02994 - Transport and adsorption under liquid flow: the role of pore geometry
Jean-Mathieu Vanson, Anne Boutin, Michaela Klotz and François-Xavier Coudert
Soft Matter 13, 875-885 (2017)
http://dx.doi.org/10.1039/C6SM02414A - Kinetic Accessibility of Porous Material Adsorption Sites Studied through the Lattice Boltzmann Method
Jean-Mathieu Vanson, François-Xavier Coudert, Michaela Klotz and Anne Boutin
Langmuir 33, 1405-1411 (2017)
http://dx.doi.org/10.1021/acs.langmuir.6b04472 - Transient hydrodynamic finite size effects in simulations under periodic boundary conditions
Adelchi J. Asta, Maximilien Levesque, Rodolphe Vuilleumier and Benjamin Rotenberg
Phys. Rev. E 95, 061301 (2017)
https://dx.doi.org/10.1103/PhysRevE.95.061301 - Moment propagation method for the dynamics of charged adsorbing/desorbing species at solid-liquid interfaces
Adelchi Asta, Maximilien Levesque and Benjamin Rotenberg
Molecular Physics, Frenkel Special Issue (2018)
https://doi.org/10.1080/00268976.2018.1461944
git clone https://github.com/maxlevesque/laboetie
cd laboetie
makeThe moment propagation is parallelized. It uses the OPENMP API. It is enabled by default (see -fopenmp in the Makefile).
To disable openmp parallelism, remove -fopenmp from line 13 of Makefile.
By default, laboetie will use all the threads of your computer.
How do you control the number of threads used by OPENMP?
You should export OMP_NUM_THREADS=3 in your terminal before executing laboetie if you want laboetie to use 3 threads:
Thus, to compile and execute laboetie limited to 8 threads on my computer, I type in my terminal:
$ export OMP_NUM_THREADS=8
$ ./laboetiePlease note that the run time, user time, system time and cpu time printed to you by laboetie are not usable anymore when OMP_NUM_THREADS > 1. This is an issue that needs to be corrected.
Parallelism is implemented by dividing the system along almost independent slices along the z direction: if your system has {100,100,1} nodes along the x, y and z directions, respectively, then it is absolutely useless to multithread your job: use export OMP_NUM_THREADS=1.
For now, OPENMP in laboetie is memory bound. Preliminary speed-ups for systems of few hundred of nodes are as follow:
# number of threads, speed-up
1, 1
2, 1.9
3, 2.6
4, 3.1
6, 2.7
8, 2.7
I would recommand to use 2 or 4 threads only.
lxNumber of nodes in x directionlyNumber of nodes in y directionlzNumber of nodes in z directionlbmodelLattice Boltzmann geometric model for velocities, e.g., D3Q19timestepmax_for_PoissonNernstPlanckMaximum number of timesteps in trying to find the equilibrium distribution of charged solutes. Should be 1 salt-free systems.D_equilnumber of steps for equilibrating charges (finding PB solution). 1 if charge (salt) free fluid.t_equilnumber of steps for equilibrating the flux without constraintstmomnumber of steps for equilibrating the flux without constraintstmaxbetween tmom and tmax, moment propagation is donegeometryLabel# 0 for a custom cell (written in geom.in) # 1 for a slit, i.e. two walls at z=zmin and z=zmax # 2 for a cylinder along Z. lx have to be equal to ly. # 3 for a body centered cubic cell with solid spheres in contact # 4 for disc benichou with exists at 0, 3, 6 and 9 oclock # 5 for corrugated wall (sinusoidal and mirored wrt to y=ly/2. # 6 for a slit with various diameter (2D) # 7 for a tube with various diameter (3D) # 8 for a spherical cavity (3D) # 9 for sphere with 6 exits as asked by benichou # 10 for the supercell given by Xudong, Vincent, Marie and Benjamin from COMSOLstripes?initialSolventDensityfluid density in LB unitsf_extexternal forcecharge_distribsol) charge distributed in the whole solid. int) charge distributed on interfacial nodes onlysigma= 0.0 # charge distributed in solidbjl= 0.4 # bjerum lengthlambda_D= -2.0 # old debye_lD_plus= 0.05D_minus= 0.05D_iter= 1tracer_Db= 0.01 # bulk diffusion coefficient of the tracertracer_Ds= 0.0 # surface diffusion coefficient of the tracertracer_z= 0.0 # charge of the tracertracer_ka= 0.1 # adsorption coefficient of the tracertracer_kd= 0.01 # desorption coefficient of the tracerc_tracer_0= 0.1left wall vel= 0.0right wall vel= 0.0Tprint_eq= 1000Tprint_run= 10000elec_slope= 0.0 0.0 0.0 # external electric fieldlncb_slope= 0.0 0.0 0.0 # external gradient of salt concentrationf_gen= 0.0 0.0 0.0
All outputs files are found in output/.
A 3-dimensional representation of the supercell in xsf format.
supercelf.xsf can be opened with VMD: vmd -xsf output/supercell.xsf.
The color code is:
pinkSolid nodesgreenInterfacial fluid nodes, i.e., fluid nodes close to a solid nodewhiteNon-interfacial fluid nodes