Hello author, your work is excellent. I am currently trying to utilize the forward operator in solving FWI, and have attempted to compute receiver signals based on velocity using both deepwave and handwritten numerical methods. Unfortunately, the solutions I obtained seem to have significant discrepancies compared to the dataset. May I ask if the solutions in the dataset have undergone any normalization processing? I believe I have implemented the methods based on the MATLAB code provided in your paper, but I am currently unable to resolve this issue.
Thank you very much for your attention.
`import deepwave
import matplotlib.pyplot as plt
import torch
import numpy as np
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
ny = 70
nx = 70
dx = 10.0
v = torch.from_numpy(np.load("data_download/curvevel-A/model1.npy")[0]).reshape(ny, nx).to(device)
n_shots = 5
n_sources_per_shot = 1
d_source = 14
first_source = 7
source_depth = 1
n_receivers_per_shot = 70
d_receiver = 1
first_receiver = 0
receiver_depth = 1
freq = 15
nt = 1000
dt = 0.001
peak_time = 1.5 / freq
source_locations = torch.zeros(
n_shots, n_sources_per_shot, 2, dtype=torch.long, device=device
)
source_locations[..., 1] = source_depth
source_locations[:, 0, 0] = torch.arange(n_shots) * d_source + first_source
receiver_locations = torch.zeros(
n_shots, n_receivers_per_shot, 2, dtype=torch.long, device=device
)
receiver_locations[..., 1] = receiver_depth
receiver_locations[:, :, 0] = (
torch.arange(n_receivers_per_shot) * d_receiver + first_receiver
).repeat(n_shots, 1)
source_amplitudes = (
deepwave.wavelets.ricker(freq, nt, dt, peak_time)
.repeat(n_shots, n_sources_per_shot, 1)
.to(device)
)
output = deepwave.scalar(
v.T,
dx,
dt,
source_amplitudes=source_amplitudes,
source_locations=source_locations,
receiver_locations=receiver_locations,
accuracy=8,
pml_width = 20,
pml_freq=freq,
)
receiver_amplitudes = output[-1].permute(0,2,1)
label = np.load("data_download/curvevel-A/data1.npy")[0]
vmin, vmax = torch.quantile(
receiver_amplitudes[0], torch.tensor([0.05, 0.95]).to(device)
)
_, ax = plt.subplots(1, 2, figsize=(10.5, 7), sharey=True)
ax[0].imshow(
receiver_amplitudes[1].cpu(),
aspect="auto",
cmap="gray",
vmin = vmin, vmax = vmax
)
ax[1].imshow(
label[1],
aspect="auto",
cmap="gray",
vmin = vmin, vmax = vmax
)
ax[0].set_xlabel("Channel")
ax[0].set_ylabel("Time Sample")
ax[1].set_xlabel("Shot")
plt.tight_layout()
plt.savefig("example_forward_model.jpg")`
Hello author, your work is excellent. I am currently trying to utilize the forward operator in solving FWI, and have attempted to compute receiver signals based on velocity using both deepwave and handwritten numerical methods. Unfortunately, the solutions I obtained seem to have significant discrepancies compared to the dataset. May I ask if the solutions in the dataset have undergone any normalization processing? I believe I have implemented the methods based on the MATLAB code provided in your paper, but I am currently unable to resolve this issue.
Thank you very much for your attention.
`import deepwave
import matplotlib.pyplot as plt
import torch
import numpy as np
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
ny = 70
nx = 70
dx = 10.0
v = torch.from_numpy(np.load("data_download/curvevel-A/model1.npy")[0]).reshape(ny, nx).to(device)
n_shots = 5
n_sources_per_shot = 1
d_source = 14
first_source = 7
source_depth = 1
n_receivers_per_shot = 70
d_receiver = 1
first_receiver = 0
receiver_depth = 1
freq = 15
nt = 1000
dt = 0.001
peak_time = 1.5 / freq
source_locations = torch.zeros(
n_shots, n_sources_per_shot, 2, dtype=torch.long, device=device
)
source_locations[..., 1] = source_depth
source_locations[:, 0, 0] = torch.arange(n_shots) * d_source + first_source
receiver_locations = torch.zeros(
n_shots, n_receivers_per_shot, 2, dtype=torch.long, device=device
)
receiver_locations[..., 1] = receiver_depth
receiver_locations[:, :, 0] = (
torch.arange(n_receivers_per_shot) * d_receiver + first_receiver
).repeat(n_shots, 1)
source_amplitudes = (
deepwave.wavelets.ricker(freq, nt, dt, peak_time)
.repeat(n_shots, n_sources_per_shot, 1)
.to(device)
)
output = deepwave.scalar(
v.T,
dx,
dt,
source_amplitudes=source_amplitudes,
source_locations=source_locations,
receiver_locations=receiver_locations,
accuracy=8,
pml_width = 20,
pml_freq=freq,
)
receiver_amplitudes = output[-1].permute(0,2,1)
label = np.load("data_download/curvevel-A/data1.npy")[0]
vmin, vmax = torch.quantile(
receiver_amplitudes[0], torch.tensor([0.05, 0.95]).to(device)
)
_, ax = plt.subplots(1, 2, figsize=(10.5, 7), sharey=True)
ax[0].imshow(
receiver_amplitudes[1].cpu(),
aspect="auto",
cmap="gray",
vmin = vmin, vmax = vmax
)
ax[1].imshow(
label[1],
aspect="auto",
cmap="gray",
vmin = vmin, vmax = vmax
)
ax[0].set_xlabel("Channel")
ax[0].set_ylabel("Time Sample")
ax[1].set_xlabel("Shot")
plt.tight_layout()
plt.savefig("example_forward_model.jpg")`