Hi,Sir
I try run the sample.py by using marmousi velocity model.

But ,when set honor_grid=True.

Runtime Error will appears.

Description of the problem

I can't set the poisition to (10.0,10.0) but to (10.0+1e-3,10.0+1e-3) or (10.5,10.5)

Full code that generated the error

My test code is in the attachment.

sample.zip

Full error message

Traceback (most recent call last):
  File "sample.py", line 37, in <module>
    trays = [tt(ray) for ray in rays]
  File "sample.py", line 37, in <listcomp>
    trays = [tt(ray) for ray in rays]
  File "C:\Users\xcq\AppData\Roaming\Python\Python38\site-packages\fteikpy\_grid.py", line 91, in __call__
    return vinterp2d(
ZeroDivisionError: division by zero

System information

• Operating system:Winodws 10
• Python install

ation (Anaconda, system, ETS):Anaconda

• Version of Python:3.8
• Version of this package:2.0

Description of the problem

When I run the example, the kernel will die at the last line (see the figure below);

Full code that generated the error

import numpy
from fteikpy import Eikonal3D

# Velocity model
velocity_model = numpy.ones((8, 8, 8))
dz, dx, dy = 1.0, 1.0, 1.0

# Solve Eikonal at source
eik = Eikonal3D(velocity_model, gridsize=(dz, dx, dy))
tt = eik.solve((0.0, 0.0, 0.0))

# Get traveltime at specific grid point
t1 = tt[0, 1, 2]

# Or get traveltime at any point in the grid
t2 = tt(numpy.random.rand(3) * 7.0)

Full error message

Kernel died, restarting

System information

• Operating system: Windows 10
• Python installation (Anaconda, system, ETS): Anaconda
• Version of Python: 3.8.8
• Version of this package: latest

Description of the desired feature

Are you willing to help implement and maintain this feature? Yes/No

Description of the problem
When I run the code to get traveltime at some points,Some error like ZeroDivisionError: division by zero.
I want to give a random array of positions within a circle to get traveltime.The source is in the center of the circular region.I tried different speed models and grid spacing sizes, which can pass when the array is linear, but may fail when the array becomes random.I wonder if this could have been avoided by code optimization or certain design principles?

Full code that generated the error

PASTE YOUR CODE HERE

import numpy as np
import pyvista
import random

from fteikpy import Eikonal3D, grid_to_meshio

pyvista.set_plot_theme("document")

def getRandomPointInCircle(num, radius, centerx, centery):
    
    # Polar coordinates generate a random array within a ring
    samplePoint = []
    for i in range(num):
        theta = random.random() * 2 * np.pi
        r = random.uniform(0, radius ** 2) 
        x = np.cos(theta) * (r ** 0.5) + centerx
        y = np.sin(theta) * (r ** 0.5) + centery
        samplePoint.append([x, y])

    return samplePoint
def main():
 
    # Observation system ：Polar coordinates generate a random array within a ring
    num = 50
    radius = 12
    centerx,centery = 5, 5
    receiver = np.array(getRandomPointInCircle(num, radius, centerx, centery))

    # Velocity model
    velocity_model = np.ones((60, 200, 200))
    dz, dx, dy = 0.5, 0.5, 0.5
    velocity_model[:10]=velocity_model[:10]*500
    velocity_model[11:30]=velocity_model[11:30]*800
    velocity_model[31:60]=velocity_model[31:60]*1200

    # Solve Eikonal at source
    eik = Eikonal3D(velocity_model, gridsize=(dz, dx, dy),origin=(0,-50,-50))
    #eik.smooth(2.5)
    tt = eik.solve((0, 0, 0), nsweep=3, return_gradient=True)
    ttgrid = tt.grid.ravel()

    # Trace rays for 50 locations
    end_points = np.zeros((num, 3)) #z、x、y
    end_points[:,1] = receiver[:,0] 
    end_points[:,2] = receiver[:,1]
    rays = tt.raytrace(end_points)

    trays = [tt(ray) for ray in rays]

if __name__ == '__main__':
    main()

Full error message

PASTE ERROR MESSAGE HERE

System information

• Operating system: Windows 10
• Python installation (Anaconda, system, ETS):Anaconda
• Version of Python:3.6.1
• Version of this package:latest

Description of the desired feature

It would be very helpful to have the code which generates the figure on the doc webpage, especially how the raytracing is done.

From the paper of Noble et al.2014.

The hybrid scheme has very good performance that combines a spherical approximation when close to the source and a plane
wave approximation when far away.

From _fteik2d.py I know there is some condition to choose plane wave or spherical wave.

I am not sure is the hybrid scheme applied in _fteik3d.py.

Do you know this ?

Description of the desired feature
Hi, @keurfonluu
In the laboratory, the sample is a usually cylinder column. So, would you please add a feature to calculte the traveltime and trace the ray inside a cylinder area ? Thank you

Are you willing to help implement and maintain this feature? Yes/No
Yes

Hi Keurfon,

Thanks for your efforts of releasing/maintaining your code projects. fteikpy has been very useful to me so far.

From the documentation and the tests it is not entirely clear to me what the tt.gradient vector represents. Are these the gradients of the travel times with respect to the source location, for each receiver? If yes, that would be convenient for gradient-based hypocentre inversion :)

How can I get the Jacobian matrix from the existing code?