def to_img(input: torch.Tensor, mode: str, colormap=True, normalize=True):
if (mode == "2D"):
img = input[0].clone().detach()
if (normalize):
img -= img.min()
img *= (1 / img.max() + 1e-6)
if (colormap and img.shape[0] == 1):
img = cm.coolwarm(img[0].cpu().numpy())
#img = np.transpose(img, (2, 0, 1))
img = (255 * img).astype(np.uint8)
else:
img *= 255
img = img.permute(1, 2, 0).cpu().numpy().astype(np.uint8)
elif (mode == "3D"):
img = input[0, :, :, :, int(input.shape[4] / 2)].clone()
if (normalize):
img -= img.min()
img *= (1 / img.max() + 1e-6)
if (colormap and img.shape[0] == 1):
img = cm.coolwarm(img[0].cpu().numpy())
#img = np.transpose(img, (2, 0, 1))
img = (255 * img).astype(np.uint8)
else:
img *= 255
img = img.permute(1, 2, 0).cpu().numpy().astype(np.uint8)
#print(img.shape)
return img
def toImg(vectorField, renorm_channels = False):
vf = vectorField.copy()
if(len(vf.shape) == 3):
if(vf.shape[0] == 1):
return cm.coolwarm(vf[0]).swapaxes(0,2).swapaxes(1,2)
elif(vf.shape[0] == 2):
vf += 1
vf *= 0.5
vf = vf.clip(0, 1)
z = np.zeros([1, vf.shape[1], vf.shape[2]])
vf = np.concatenate([vf, z])
return vf
elif(vf.shape[0] == 3):
if(renorm_channels):
for j in range(vf.shape[0]):
vf[j] -= vf[j].min()
vf[j] *= (1 / vf[j].max())
return vf
elif(len(vf.shape) == 4):
return toImg(vf[:,:,:,0], renorm_channels)
from matplotlib.pyplot import cm
#Plotting
figure=plt.figure()
axes = figure.add_axes([0.1,0.1,1.2,1.2])
plt.xticks(fontsize=14)
plt.yticks(fontsize=14)
# Data
G=1
J=1
tau=np.array([0.1, 1, 2, 5, 10, 50, 100, 500]);
N = 500 #number of points
L = np.linspace(0,100, N)
color=iter(cm.coolwarm(np.linspace(0,1,np.size(tau))))
for i in range(0,np.size(tau)):
c=next(color)
n= J / G * np.exp(-L/G/tau[i])
axes.plot(L,n, marker=' ' , c=c)
plt.title('n(L) vs. residence time', fontsize=18);
axes.set_xlabel('$L$', fontsize=18);
axes.set_ylabel('$n$',fontsize=18);
#########################################################
# Plotting
figure=plt.figure()
axes = figure.add_axes([0.1,0.1,1.2,1.2])
plt.xticks(fontsize=14)