FiniteVolumeGPU/GPUSimulators/helpers/initial_conditions.py
2025-07-24 13:42:03 +02:00

341 lines
9.6 KiB
Python

# -*- coding: utf-8 -*-
"""
This python module implements Cuda context handling
Copyright (C) 2018 SINTEF ICT
This program is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation, either version 3 of the License, or
(at your option) any later version.
This program is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program. If not, see <http://www.gnu.org/licenses/>.
"""
import gc
import numpy as np
from GPUSimulators.simulator.boundary import BoundaryCondition
def get_extent(width, height, nx, ny, grid, index=None):
if grid is not None:
gx = grid.x
gy = grid.y
if index is not None:
i, j = grid.get_coordinate(index)
else:
i, j = grid.get_coordinate()
dx = (width / gx) / nx
dy = (height / gy) / ny
x0 = width * i / gx + 0.5 * dx
y0 = height * j / gy + 0.5 * dy
x1 = width * (i + 1) / gx - 0.5 * dx
y1 = height * (j + 1) / gy - 0.5 * dx
else:
dx = width / nx
dy = height / ny
x0 = 0.5 * dx
y0 = 0.5 * dy
x1 = width - 0.5 * dx
y1 = height - 0.5 * dy
return [x0, x1, y0, y1, dx, dy]
def downsample(highres_solution, x_factor, y_factor=None):
if y_factor is None:
y_factor = x_factor
assert (highres_solution.shape[1] % x_factor == 0)
assert (highres_solution.shape[0] % y_factor == 0)
if x_factor * y_factor == 1:
return highres_solution
if len(highres_solution.shape) == 1:
highres_solution = highres_solution.reshape((1, highres_solution.size))
nx = highres_solution.shape[1] / x_factor
ny = highres_solution.shape[0] / y_factor
return highres_solution.reshape([int(ny), int(y_factor), int(nx), int(x_factor)]).mean(3).mean(1)
def bump(nx: int, ny: int, width: int, height: int,
bump_size=None,
ref_nx=None, ref_ny=None,
x_center=0.5, y_center=0.5,
h_ref=0.5, h_amp=0.1, u_ref=0.0, u_amp=0.1, v_ref=0.0, v_amp=0.1):
if ref_nx is None:
ref_nx = nx
assert (ref_nx >= nx)
if ref_ny is None:
ref_ny = ny
assert (ref_ny >= ny)
if bump_size is None:
bump_size = width / 5.0
ref_dx = width / float(ref_nx)
ref_dy = height / float(ref_ny)
x_center = ref_dx * ref_nx * x_center
y_center = ref_dy * ref_ny * y_center
x = ref_dx * (np.arange(0, ref_nx, dtype=np.float32) + 0.5) - x_center
y = ref_dy * (np.arange(0, ref_ny, dtype=np.float32) + 0.5) - y_center
xv, yv = np.meshgrid(x, y, sparse=False, indexing='xy')
r = np.sqrt(xv ** 2 + yv ** 2)
xv = None
yv = None
gc.collect()
# Generate highres then downsample
# h_highres = 0.5 + 0.1*np.exp(-(xv**2/size + yv**2/size))
h_highres = h_ref + h_amp * 0.5 * (1.0 + np.cos(np.pi * r / bump_size)) * (r < bump_size)
h = downsample(h_highres, ref_nx / nx, ref_ny / ny)
h_highres = None
gc.collect()
# hu_highres = 0.1*np.exp(-(xv**2/size + yv**2/size))
u_highres = u_ref + u_amp * 0.5 * (1.0 + np.cos(np.pi * r / bump_size)) * (r < bump_size)
hu = downsample(u_highres, ref_nx / nx, ref_ny / ny) * h
u_highres = None
gc.collect()
# hu_highres = 0.1*np.exp(-(xv**2/size + yv**2/size))
v_highres = v_ref + v_amp * 0.5 * (1.0 + np.cos(np.pi * r / bump_size)) * (r < bump_size)
hv = downsample(v_highres, ref_nx / nx, ref_ny / ny) * h
v_highres = None
gc.collect()
dx = width / nx
dy = height / ny
return h, hu, hv, dx, dy
def gen_shock_bubble(nx, ny, gamma, grid=None):
"""
Generate a Shock-bubble interaction case for the Euler equations
"""
width = 4.0
height = 1.0
g = 0.0
rho = np.ones((ny, nx), dtype=np.float32)
u = np.zeros((ny, nx), dtype=np.float32)
v = np.zeros((ny, nx), dtype=np.float32)
E = np.zeros((ny, nx), dtype=np.float32)
p = np.ones((ny, nx), dtype=np.float32)
x0, x1, y0, y1, dx, dy = get_extent(width, height, nx, ny, grid)
x = np.linspace(x0, x1, nx, dtype=np.float32)
y = np.linspace(y0, y1, ny, dtype=np.float32)
xv, yv = np.meshgrid(x, y, sparse=False, indexing='xy')
# Bubble
radius = 0.25
x_center = 0.5
y_center = 0.5
bubble = np.sqrt((xv - x_center) ** 2 + (yv - y_center) ** 2) <= radius
rho = np.where(bubble, 0.1, rho)
# Left boundary
left = (xv < 0.1)
rho = np.where(left, 3.81250, rho)
u = np.where(left, 2.57669, u)
# Energy
p = np.where(left, 10.0, p)
E = 0.5 * rho * (u ** 2 + v ** 2) + p / (gamma - 1.0)
bc = BoundaryCondition({
'north': BoundaryCondition.Type.Reflective,
'south': BoundaryCondition.Type.Reflective,
'east': BoundaryCondition.Type.Periodic,
'west': BoundaryCondition.Type.Periodic
})
# Construct simulator
arguments = {
'rho': rho, 'rho_u': rho * u, 'rho_v': rho * v, 'E': E,
'nx': nx, 'ny': ny,
'dx': dx, 'dy': dy,
'g': g,
'gamma': gamma,
'boundary_conditions': bc
}
return arguments
def gen_kelvin_helmholtz(nx, ny, gamma, roughness=0.125, grid=None, index=None):
"""
Roughness parameter in (0, 1.0] determines how "squiggly"
the interface between the zones is
"""
def gen_zones(nx, ny, n):
"""
Generates the zones of the two fluids of K-H
"""
zone = np.zeros((ny, nx), dtype=np.int32)
def gen_smooth_random(nx, n):
n = max(1, min(n, nx))
if n == nx:
return np.random.random(nx) - 0.5
else:
from scipy.interpolate import interp1d
# Control points and interpolator
xp = np.linspace(0.0, 1.0, n)
yp = np.random.random(n) - 0.5
if n == 1:
kind = 'nearest'
elif n == 2:
kind = 'linear'
elif n == 3:
kind = 'quadratic'
else:
kind = 'cubic'
f = interp1d(xp, yp, kind=kind)
# Interpolation points
x = np.linspace(0.0, 1.0, nx)
return f(x)
x0, x1, y0, y1, _, dy = get_extent(1.0, 1.0, nx, ny, grid, index)
x = np.linspace(x0, x1, nx)
y = np.linspace(y0, y1, ny)
_, y = np.meshgrid(x, y)
# print(y+a[0])
a = gen_smooth_random(nx, n) * dy
zone = np.where(y > 0.25 + a, zone, 1)
a = gen_smooth_random(nx, n) * dy
zone = np.where(y < 0.75 + a, zone, 1)
return zone
width = 2.0
height = 1.0
g = 0.0
gamma = 1.4
rho = np.empty((ny, nx), dtype=np.float32)
u = np.empty((ny, nx), dtype=np.float32)
v = np.zeros((ny, nx), dtype=np.float32)
p = 2.5 * np.ones((ny, nx), dtype=np.float32)
# Generate the different zones
zones = gen_zones(nx, ny, max(1, min(nx, int(nx * roughness))))
# Zone 0
zone0 = zones == 0
rho = np.where(zone0, 1.0, rho)
u = np.where(zone0, 0.5, u)
# Zone 1
zone1 = zones == 1
rho = np.where(zone1, 2.0, rho)
u = np.where(zone1, -0.5, u)
E = 0.5 * rho * (u ** 2 + v ** 2) + p / (gamma - 1.0)
_, _, _, _, dx, dy = get_extent(width, height, nx, ny, grid, index)
bc = BoundaryCondition({
'north': BoundaryCondition.Type.Periodic,
'south': BoundaryCondition.Type.Periodic,
'east': BoundaryCondition.Type.Periodic,
'west': BoundaryCondition.Type.Periodic
})
# Construct simulator
arguments = {
'rho': rho, 'rho_u': rho * u, 'rho_v': rho * v, 'E': E,
'nx': nx, 'ny': ny,
'dx': dx, 'dy': dy,
'g': g,
'gamma': gamma,
'boundary_conditions': bc
}
return arguments
def gen_rayleigh_taylor(nx, ny, gamma, version=0, grid=None):
"""
Generates a Rayleigh-Taylor instability case
"""
width = 0.5
height = 1.5
g = 0.1
rho = np.zeros((ny, nx), dtype=np.float32)
u = np.zeros((ny, nx), dtype=np.float32)
v = np.zeros((ny, nx), dtype=np.float32)
p = np.zeros((ny, nx), dtype=np.float32)
x0, x1, y0, y1, dx, dy = get_extent(width, height, nx, ny, grid)
x = np.linspace(x0, x1, nx, dtype=np.float32) - width * 0.5
y = np.linspace(y0, y1, ny, dtype=np.float32) - height * 0.5
xv, yv = np.meshgrid(x, y, sparse=False, indexing='xy')
# This gives a squiggly interfaction
if version == 0:
y_threshold = 0.01 * np.cos(2 * np.pi * np.abs(x) / 0.5)
rho = np.where(yv <= y_threshold, 1.0, rho)
rho = np.where(yv > y_threshold, 2.0, rho)
elif version == 1:
rho = np.where(yv <= 0.0, 1.0, rho)
rho = np.where(yv > 0.0, 2.0, rho)
v = 0.01 * (1.0 + np.cos(2 * np.pi * xv / 0.5)) / 4
else:
assert False, "Invalid version"
p = 2.5 - rho * g * yv
E = 0.5 * rho * (u ** 2 + v ** 2) + p / (gamma - 1.0)
bc = BoundaryCondition({
'north': BoundaryCondition.Type.Reflective,
'south': BoundaryCondition.Type.Reflective,
'east': BoundaryCondition.Type.Reflective,
'west': BoundaryCondition.Type.Reflective
})
# Construct simulator
arguments = {
'rho': rho, 'rho_u': rho * u, 'rho_v': rho * v, 'E': E,
'nx': nx, 'ny': ny,
'dx': dx, 'dy': dy,
'g': g,
'gamma': gamma,
'boundary_conditions': bc
}
return arguments