def resolution_study():
run_shell_command("make")
N = 256
num_stencil = [ int(i*N) for i in [
1/2,
1,
2,
]]
dt = 0.0005
re = 80
steps = 10_000
initial_steps = 25_000
resolution_folder = "resolution_study_3"
for n in num_stencil:
for skip_diffusion in [0,1]:
diff_str = skip_diffusion_to_str(skip_diffusion)
folder_name = f"{n}"
case = RunCase(
skip_diffusion=skip_diffusion,
size = n,
dt = dt,
steps = steps,
restarts = 0,
reynolds_number=re,
path = f"{BASE_SAVE}/{resolution_folder}/{diff_str}/{folder_name}",
load_initial_data=2,
export_vtk=True,
skip_steps=initial_steps
)
wrap_error_case(case, f"{BASE_SAVE}/{resolution_folder}/errors.txt")
def temporal_study():
run_shell_command("make")
DT = 0.0005
STEPS = 10_000
INITIAL_STEPS = 25_000
timesteps = [ [i*DT, int(STEPS / i), int(INITIAL_STEPS / i)] for i in [
1/2,
1,
2,
]]
N = 256
re = 80
temporal_folder = "temporal_study_3"
print("timesteps are", timesteps)
for dt, steps, initial_steps in timesteps:
for skip_diffusion in [0,1]:
diff_str = skip_diffusion_to_str(skip_diffusion)
folder_name = f"{dt}"
case = RunCase(
skip_diffusion=skip_diffusion,
size = N,
dt = dt,
steps = steps,
restarts = 0,
reynolds_number=re,
path = f"{BASE_SAVE}/{temporal_folder}/{diff_str}/{folder_name}",
load_initial_data=2,
export_vtk=True,
skip_steps=initial_steps
)
wrap_error_case(case, f"{BASE_SAVE}/{temporal_folder}/errors.txt")
def one_case():
TIME_END = 0.5
batch_name = "viscous_baseline_2"
job_name = "single-case"
save_json_folder = f"{BASE_SAVE}/{batch_name}"
size = 128
dt = 0.0005
steps = int(TIME_END / dt)
nprocs = 16
extra_caps = []
io_steps = None
load_initial_data = 0
export_divergence = 0
ep1 = 0.001
ep2 = 0.001
skip_diffusion = 0
output_folder = define_output_folder()
if not (load_initial_data == 2):
copy_init_files(size)
# if the directory exists remove any older files from the dir
if os.path.exists(save_json_folder):
for f in os.listdir(save_json_folder):
os.remove(os.path.join(save_json_folder, f))
os.makedirs(save_json_folder, exist_ok=True)
run_shell_command("make")
case = RunCase(
skip_diffusion=skip_diffusion,
size=size,
dt=dt,
steps=steps,
restarts=0,
reynolds_number=40,
path=output_folder,
load_initial_data=load_initial_data,
epsilon1=ep1,
# delta2 is negative, this will decrease helicity
epsilon2=ep2,
export_vtk=True,
scalar_type=14,
require_forcing=1,
viscous_compensation=0,
validate_viscous_compensation=0,
io_steps=io_steps,
nprocs=nprocs,
export_divergence=export_divergence,
)
if IS_DISTRIBUTED:
print("creating files to run on distributed compute")
case.write_to_json(job_name, save_json_folder)
copy_distribute_files(save_json_folder, batch_name, extra_caps)
build = Build("master", "master")
build.to_json(save_json_folder)
else:
print("running the case locally")
# init files have already been copied above
case.run(1)
def figure2():
# n=1..7 had delta_2 as .1 : delta_1 = 0.1
# n= 8..9 had delta_2 as 10.0 : delta_1 = 0.1
# n= 10 had delta_2=1 : delta_1 = 0.1
# n= 11..17 had delta_2=0.7 : delta_1 = 0.05
delta_1 = 0.05
delta_2 = 0.7
run_shell_command("make")
n = 20
batch_name = f"figure2_{n}"
save_json_folder = f"{BASE_SAVE}/{batch_name}"
if not os.path.exists(save_json_folder):
os.mkdir(save_json_folder)
for f in os.listdir(save_json_folder):
os.remove(os.path.join(save_json_folder, f))
END_TIME = 5.0
dt = 0.0005
size = 128
re = 40
steps = int(END_TIME / dt)
save_vtk = True
extra_caps = []
io_steps = 38
copy_init_files(size)
epsilon_generator = EpsilonControl.load_json()
cases = [
# baseline
# n=7 has baseline
[0.0, 0.0, "baseline_viscous", 0],
# viscous stuff
#[-1 * delta_1, 0.0, "energy_modification_viscous", 0],
#[0.0, -1 * delta_2, "helicity_modification_viscous", 0],
#[-1 * delta_1, -1 * delta_2, "both_modification_viscous", 0],
# inviscid stuff
# [-1*delta_1, 0., "energy_modification_inviscid", 1],
# [ 0., -1*delta_2, "helicity_modification_inviscid", 1],
# [-1*delta_1, -1*delta_2, "both_modification_inviscid", 1],
]
output_folder = define_output_folder()
for delta_1, delta_2, folder, skip_diffusion in cases:
epsilon1 = epsilon_generator.epsilon_1(delta_1)
epsilon2 = epsilon_generator.epsilon_2(delta_2)
case = RunCase(
skip_diffusion=skip_diffusion,
size=size,
dt=dt,
steps=steps,
restarts=0,
restart_time=1.0,
reynolds_number=re,
path=output_folder,
load_initial_data=0,
epsilon1=epsilon1,
epsilon2=epsilon2,
export_vtk=save_vtk,
# scalar_type=14,
scalar_type=0,
io_steps=io_steps,
)
case.write_to_json(folder, save_json_folder)
copy_distribute_files(save_json_folder, batch_name, extra_caps)
build = Build("master", "master")
build.to_json(save_json_folder)
def full_system_test():
delta_1 = 0.1
delta_2 = 0.1
n = 20
run_shell_command("make")
batch_name = f"system_test_{n}"
save_json_folder = f"{BASE_SAVE}/{batch_name}"
if not os.path.exists(save_json_folder):
os.mkdir(save_json_folder)
for _ in os.listdir(save_json_folder):
raise ValueError("folder should be empty")
dt = 0.0005
size = 128
re = 40
steps = 500
save_vtk = True
extra_caps = []
io_steps = 10
copy_init_files(size)
epsilon_generator = EpsilonControl.load_json()
cases = [
# [0., 0., "baseline"],
[-1 * delta_1, 0.0, "ep1-neg"],
[0.0, -1 * delta_2, "ep2-neg"],
]
output_folder = define_output_folder()
for delta_1, delta_2, folder in cases:
diffusion_str = skip_diffusion_to_str(0)
epsilon1 = epsilon_generator.epsilon_1(delta_1)
epsilon2 = epsilon_generator.epsilon_2(delta_2)
case = RunCase(
skip_diffusion=0,
size=size,
dt=dt,
steps=steps,
restarts=0,
reynolds_number=re,
path=output_folder,
load_initial_data=0,
epsilon1=epsilon1,
epsilon2=epsilon2,
export_vtk=save_vtk,
scalar_type=14,
# scalar_type=0,
io_steps=io_steps,
)
case.write_to_json(f"{folder}_{diffusion_str}", save_json_folder)
copy_distribute_files(save_json_folder, batch_name, extra_caps)
build = Build("master", "master")
build.to_json(save_json_folder)
def forcing_cases():
delta_1 = 0.1
delta_2 = 0.1
run_shell_command("make")
batch_name = f"forcing_viscous_7"
save_json_folder = f"{BASE_SAVE}/{batch_name}"
if not os.path.exists(save_json_folder):
os.mkdir(save_json_folder)
for f in os.listdir(save_json_folder):
os.remove(os.path.join(save_json_folder, f))
END_TIME = 5
dt = 0.0005
size = 128
re = 40
steps = int(END_TIME / dt)
save_vtk = True
extra_caps = []
io_steps = 38
copy_init_files(size)
epsilon_generator = EpsilonControl.load_json()
cases = [
[0.0, 0.0, "baseline"],
[-1 * delta_1, 0.0, "ep1-neg"],
[0.0, -1 * delta_2, "ep2-neg"],
]
output_folder = define_output_folder()
for skip_diffusion in [0, 1]:
# TODO: remove this after generating
if skip_diffusion == 1:
continue
for delta_1, delta_2, folder in cases:
diffusion_str = skip_diffusion_to_str(skip_diffusion)
epsilon1 = epsilon_generator.epsilon_1(delta_1)
epsilon2 = epsilon_generator.epsilon_2(delta_2)
case = RunCase(
skip_diffusion=skip_diffusion,
size=size,
dt=dt,
steps=steps,
restarts=0,
restart_time=1.0,
reynolds_number=re,
path=output_folder,
load_initial_data=0,
epsilon1=epsilon1,
epsilon2=epsilon2,
export_vtk=save_vtk,
# scalar_type=14,
scalar_type=0,
io_steps=io_steps,
)
case.write_to_json(f"{folder}_{diffusion_str}", save_json_folder)
copy_distribute_files(save_json_folder, batch_name, extra_caps)
build = Build("master", "master")
build.to_json(save_json_folder)
def forcing_sweep():
# Conclusion:
# for viscous cases, the max I values are
# ep1 : 1 -> 10^(i-2) -> 0.1
# ep2 : 3 -> 10^(i-2) -> 10
run_shell_command("make")
n = 5
forcing_folder = f"epsilon_parameter_sweep_{n}"
save_json_folder = f"{BASE_SAVE}/{forcing_folder}"
if not os.path.exists(save_json_folder):
os.mkdir(save_json_folder)
for _ in os.listdir(save_json_folder):
raise ValueError("directory is not empty")
# os.remove(os.path.join(save_json_folder, f))
END_TIME = 10
dt = 0.0005
size = 128
re = 40
steps = int(END_TIME / dt)
save_vtk = False
batch_name = forcing_folder
extra_caps = []
copy_init_files(size)
epsilon_generator = EpsilonControl.load_json()
cases = []
NUM_CASES = 10
delta_1_max = 0.5
delta_1_min = 0.01
delta_1_inc = (delta_1_max - delta_1_min) / (NUM_CASES - 1)
delta_2_max = 100
delta_2_min = 0.1
delta_2_inc = (delta_2_max - delta_2_min) / (NUM_CASES - 1)
for epsilon_value in [1, 2]:
for i in range(0, NUM_CASES):
if epsilon_value == 1:
delta = delta_1_min + (delta_1_inc * i)
newcase = [delta, 0, f"ep1-{i}"]
else:
delta = delta_2_min + (delta_2_inc * i)
newcase = [0, delta, f"ep2-{i}"]
cases.append(newcase)
delta_1_cases = [i for i, _, _ in cases[0:NUM_CASES]]
epsilon_1_cases = [epsilon_generator.epsilon_1(i) for i in delta_1_cases]
z1 = list(zip(delta_1_cases, epsilon_1_cases))
delta_2_cases = [i for _, i, _ in cases[NUM_CASES:]]
epsilon_2_cases = [epsilon_generator.epsilon_2(i) for i in delta_2_cases]
z2 = list(zip(delta_2_cases, epsilon_2_cases))
print("asdf")
print(f"forced energy cases:")
pprint(z1)
print(f"forced helicity cases:")
pprint(z2)
if IS_SINGULARITY and IS_DISTRIBUTED:
output_folder = f"/distribute_save/"
else:
output_folder = f"../../distribute_save/"
for skip_diffusion in [0, 1]:
# TODO: remove this after generating
if skip_diffusion == 1:
continue
for delta_1, delta_2, folder in cases:
diffusion_str = skip_diffusion_to_str(skip_diffusion)
epsilon1 = epsilon_generator.epsilon_1(delta_1)
epsilon2 = epsilon_generator.epsilon_2(delta_2)
if epsilon1 == -0.0:
epsilon1 = 0.0
if epsilon2 == -0.0:
epsilon2 = 0.0
print("ep1: {:.5E} ep2: {:.5E}".format(epsilon1, epsilon2))
case = RunCase(
skip_diffusion=skip_diffusion,
size=size,
dt=dt,
steps=steps,
restarts=0,
restart_time=1.0,
reynolds_number=re,
path=output_folder,
load_initial_data=0,
epsilon1=epsilon1,
epsilon2=epsilon2,
export_vtk=save_vtk,
scalar_type=14,
)
case.write_to_json(f"{folder}_{diffusion_str}", save_json_folder)
copy_distribute_files(save_json_folder, batch_name, extra_caps)
build = Build("master", "master")
build.to_json(save_json_folder)