![Welcome to [Daby-Seesaram, Skardova & Genet, Submitted]'s demos! - Home](../_static/logo.png){kind=logo}
2D case - Interactive plot#
Two-parameter 2D problem - Interactive widget with sliders
Libraries import#
import sys
import torch
import torch.nn as nn
from neurom.HiDeNN_PDE import MeshNN, NeuROM, MeshNN_2D, MeshNN_1D
import neurom.src.Pre_processing as pre
from neurom.src.PDE_Library import Strain, Stress,VonMises_plain_strain
from neurom.src.Training import Training_NeuROM_multi_level
import neurom.Post.Plots as Pplot
import time
import os
import torch._dynamo as dynamo
from importlib import reload
import tomllib
import numpy as np
import argparse
torch.manual_seed(0)
<torch._C.Generator at 0x1238e1610>
Load the config file#
Configuration_file = 'Configurations/config_2D_ROM_interactive.toml'
with open(Configuration_file, mode="rb") as file:
config = tomllib.load(file)
Definition of the space domain and mechanical proprieties of the structure#
The initial Material parameters, the geometry, mesh and the boundary conditions are set.
# Material parameters definition
Mat = pre.Material( flag_lame = False, # If True should input lmbda and mu instead of E and nu
coef1 = config["material"]["E"], # Young Modulus
coef2 = config["material"]["nu"] # Poisson's ratio
)
# Create mesh object
MaxElemSize = pre.ElementSize(
dimension = config["interpolation"]["dimension"],
L = config["geometry"]["L"],
order = config["interpolation"]["order"],
np = config["interpolation"]["np"],
MaxElemSize2D = config["interpolation"]["MaxElemSize2D"]
)
Excluded = []
Mesh_object = pre.Mesh(
config["geometry"]["Name"], # Create the mesh object
MaxElemSize,
config["interpolation"]["order"],
config["interpolation"]["dimension"]
)
Mesh_object.AddBorders(config["Borders"]["Borders"])
Mesh_object.AddBCs( # Include Boundary physical domains infos (BCs+volume)
config["geometry"]["Volume_element"],
Excluded,
config["DirichletDictionryList"]
)
Mesh_object.MeshGeo() # Mesh the .geo file if .msh does not exist
Mesh_object.ReadMesh()
Mesh_object.ExportMeshVtk()
_ _ ____ ___ __ __
| \ | | ___ _ _| _ \ / _ \| \/ |
| \| |/ _ \ | | | |_) | | | | |\/| |
| |\ | __/ |_| | _ <| |_| | | | |
|_| \_|\___|\__,_|_| \_\ ___/|_| |_|
unknown version
************ MESH READING COMPLETE ************
* Dimension of the problem: 2D
* Elements type: t3: 3-node triangle
* Number of Dofs: 224
fatal: No names found, cannot describe anything.
Parametric study definition#
The hypercube describing the parametric domain used for the tensor decomposition is set-up here
ParameterHypercube = torch.tensor([ [ config["parameters"]["para_1_min"],
config["parameters"]["para_1_max"],
config["parameters"]["N_para_1"]],
[ config["parameters"]["para_2_min"],
config["parameters"]["para_2_max"],
config["parameters"]["N_para_2"]]])
Initialisation of the surrogate model#
ROM_model = NeuROM( # Build the surrogate (reduced-order) model
Mesh_object,
ParameterHypercube,
config,
config["solver"]["n_modes_ini"],
config["solver"]["n_modes_max"]
)
Plot initial parametric nodal values#
ROM_model.Para_modes[0][0].InterpoLayer.weight
Parameter containing:
tensor([[ 0.0359, 0.1644, -0.1118, -0.2690, 0.0633, -0.0255, 0.1551, 0.2669,
0.2174, 0.1172]], requires_grad=True)
Training the model#
ROM_model.Freeze_Mesh() # Set space mesh coordinates as untrainable
ROM_model.Freeze_MeshPara() # Set parameters mesh coordinates as untrainable
ROM_model.TrainingParameters(
loss_decrease_c = config["training"]["loss_decrease_c"],
Max_epochs = config["training"]["n_epochs"],
learning_rate = config["training"]["learning_rate"]
)
ROM_model.train() # Put the model in training mode
ROM_model, Mesh_object = Training_NeuROM_multi_level(ROM_model,config, Mat)
* Refinement level: 0
**************** START TRAINING ***************
epoch 100 loss = -3.27861e-04 modes = 1
epoch 200 loss = -1.25096e-03 modes = 1
epoch 300 loss = -2.29776e-03 modes = 1
epoch 400 loss = -3.44243e-03 modes = 1
epoch 500 loss = -4.01850e-03 modes = 1
epoch 600 loss = -4.17413e-03 modes = 1
epoch 700 loss = -4.20585e-03 modes = 1
epoch 800 loss = -4.23706e-03 modes = 2
epoch 900 loss = -4.78608e-03 modes = 2
epoch 1000 loss = -5.07869e-03 modes = 2
epoch 1100 loss = -5.22825e-03 modes = 2
epoch 1200 loss = -5.33641e-03 modes = 2
epoch 1300 loss = -5.37339e-03 modes = 2
*************** END FIRST PHASE ***************
* Training time: 5.019593000411987s
* Saving time: 7.152557373046875e-07s
* Evaluation time: 2.730597734451294s
* Backward time: 2.04931902885437s
* Update time: 0.1777338981628418s
* Average epoch time: 0.003751564275345282s
************** START SECOND PAHSE *************
epoch 5 loss = -5.377e-03
*************** END OF TRAINING ***************
* Training time: 5.286859035491943s
************ MESH READING COMPLETE ************
* Dimension of the problem: 2D
* Elements type: t3: 3-node triangle
* Number of Dofs: 434
* Refinement level: 1
**************** START TRAINING ***************
epoch 100 loss = -5.71027e-03 modes = 2
*************** END FIRST PHASE ***************
* Training time: 6.105348110198975s
* Saving time: 0s
* Evaluation time: 0.4201951026916504s
* Backward time: 0.3636465072631836s
* Update time: 0.025295495986938477s
* Average epoch time: 0.00597437280808052s
************** START SECOND PAHSE *************
epoch 5 loss = -5.7137e-03
*************** END OF TRAINING ***************
* Training time: 6.436589241027832s
************ MESH READING COMPLETE ************
* Dimension of the problem: 2D
* Elements type: t3: 3-node triangle
* Number of Dofs: 1194
* Refinement level: 2
**************** START TRAINING ***************
*************** END FIRST PHASE ***************
* Training time: 7.43602728843689s
* Saving time: 0s
* Evaluation time: 0.4173312187194824s
* Backward time: 0.5454256534576416s
* Update time: 0.026410341262817383s
* Average epoch time: 0.010746645671065135s
************** START SECOND PAHSE *************
epoch 5 loss = -5.9394e-03
*************** END OF TRAINING ***************
* Training time: 7.599175453186035s
************ MESH READING COMPLETE ************
* Dimension of the problem: 2D
* Elements type: t3: 3-node triangle
* Number of Dofs: 4038
* Refinement level: 3
**************** START TRAINING ***************
*************** END FIRST PHASE ***************
* Training time: 9.333802461624146s
* Saving time: 0s
* Evaluation time: 0.6638827323913574s
* Backward time: 1.0338337421417236s
* Update time: 0.026207923889160156s
* Average epoch time: 0.02168283760547638s
************** START SECOND PAHSE *************
epoch 5 loss = -6.0577e-03
*************** END OF TRAINING ***************
* Training time: 9.775207281112671s
************ MESH READING COMPLETE ************
* Dimension of the problem: 2D
* Elements type: t3: 3-node triangle
* Number of Dofs: 14230
* Refinement level: 4
**************** START TRAINING ***************
*************** END FIRST PHASE ***************
* Training time: 15.05105209350586s
* Saving time: 0s
* Evaluation time: 1.7289412021636963s
* Backward time: 3.502650260925293s
* Update time: 0.033141136169433594s
* Average epoch time: 0.06356439533003841s
************** START SECOND PAHSE *************
epoch 5 loss = -6.1054e-03
epoch 10 loss = -6.1054e-03
*************** END OF TRAINING ***************
* Training time: 27.139592170715332s
Plotting area#
Reproducing figure 15
tikz = False
import matplotlib.pyplot as plt
from matplotlib.ticker import MaxNLocator
import pyvista as pv
pv.start_xvfb()
Pplot.Plot_2D_PyVista(ROM_model,
Mesh_object,
config,
E = config["postprocess"]["Default_E"],
theta = config["postprocess"]["Default_theta"],
scalar_field_name = config["postprocess"]["scalar_field_name"],
scaling_factor = config["postprocess"]["scaling_factor"],
Interactive_parameter = config["postprocess"]["Interactive_parameter"],
Plot_mesh = config["postprocess"]["Plot_mesh"],
color_map = config["postprocess"]["colormap"])
************** Available commands *************
* Take a screenshot : s
* Reset the camera : r