Convert this implementation to a more efficient usage of numpy
Essentially for every instance D[i] <= num, I want to use that index to add numpy x[i] to nx and numpy y[i] to ny it to the new numpy array nx, ny.
row, = D.shape
for i in range(row):
if D[i] <= num:
nx.append(x[i])
ny.append(y[i])
You can use np.where
w = np.where(D <= num)[0]
nx, ny = x[w], y[w]
demo
D = np.arange(10)
x = np.arange(10)
y = np.arange(10)[::-1]
num = 4
w = np.where(D <= num)[0]
nx, ny = x[w], y[w]
print(nx, ny)
[0 1 2 3 4] [9 8 7 6 5]
You can also use np.flatnonzero to the same effect.
D = np.arange(10)
x = np.arange(10)
y = np.arange(10)[::-1]
num = 4
w = np.flatnonzero(D <= num)
nx, ny = x[w], y[w]
print(nx, ny)
[0 1 2 3 4] [9 8 7 6 5]
naive time testing
with large-ish data
k = 100000
D = np.arange(k)
x = np.arange(k)
y = np.arange(k)[::-1]
num = k // 2
where, flatnonzero, or some other tool.