Vector operations in Python Calculator

ParaView Support
1

Dear support,

I am struggling to perform some basic operations to get the moments around an axis. My objective is to find the normal force around an axis following this sketch:

image
image1801×975 63 KB

Where AB is the axis that I’ve defined using a line filter and P is the point where a vector force is stored in my dataset.

The first step is to get the AP vector which works fine using this python calculator:

# Get reference to the inputs
surf = inputs[0]
axis = inputs[1]

# Get axis vector
A = axis.Points[ 0,:]
B = axis.Points[-1,:]
AB = B-A # VTKArray - Shape(3,)
AB = AB.reshape((3, 1))

# AP vector
AP = surf.Points-A # VTKCompositeDataArray - Shape(35248, 3)

return AP

And there you have the magnitude of the AP vector which increases as you get away from A

image
image1299×804 71.9 KB

But then I need to perform dot and cross operations between the AB and AP vectors I the following code crashes:

# Get reference to the inputs
surf = inputs[0]
axis = inputs[1]

# Get axis vector
A = axis.Points[ 0,:]
B = axis.Points[-1,:]
AB = B-A # VTKArray - Shape(3,)
AB = AB.reshape((3, 1))

# AP vector
AP = surf.Points-A # VTKCompositeDataArray - Shape(35248, 3)

return cross(AB,AP)

Is it possible to get the cross and dot product of 2 vectors which belong to different filters in a simple way? I know how to do this using the regular calculators but I am trying to do this in a python calculator to simplify a bit my pipeline :wink:

Thank you in advance for your time!

2

I’ve found a way to bypass this issue (after MANY tries) but I am a bit clueless of why this is the only solution that I could find so far:

import numpy as np

# Get reference to the inputs
surf = inputs[0]
axis = inputs[1]

# Get axis vector
A = axis.Points[ 0,:]
B = axis.Points[-1,:]
AB = B-A # VTKArray - Shape(3,)

# AP vector
AP = surf.Points-A # VTKCompositeDataArray - Shape(N Pts, 3)

# Bypass: Using an existing VTKCompositeDataArray, 
#         set it to 0 and sum your desired vector
AB_arr = AP * 0 + [AB[0], AB[1], AB[2]]

return cross(AP,AB_arr)

Also I’ve found another issue when using the “norm“ command to normalize the array that I had to bypass in a similar way. Instead of doing this:

n = norm(cross(AB_arr,r))

I have to do this:

# Bypass: Using an existing VTKCompositeDataArray, 
#         set it to 0 and sum your desired vector
Ones4dot_arr = AP * 0 + [1, 1, 1]

# Moment vector direction (n).
n = cross(AB_arr,r) / dot(Ones4dot_arr,Ones4dot_arr)
n = norm(n)

Is there any better/elegant way to perform these operations?

3

It would be very convenient to know how to declare new VTKArrays or VTKCompositeDataArrays inside the python calculator in an efficient manner to understand the structure of the data but I couldn’t find answers online :smiley: