Skip to content

numpy/growth-in-loop

Rule growth-in-loop
Pack NumPy
Severity warning
Fix none, edit by hand
Library versions every version

Growing an array with np.append or np.concatenate in a loop copies every earlier value on each pass. Collect the pieces in a list and call np.concatenate once after the loop.

out = np.array([])
for part in parts:
out = np.append(out, part)
out = np.concatenate(parts)

The rule reports out = np.append(out, x), out = np.concatenate((out, x)) and the same with vstack, hstack, column_stack, dstack and stack inside a loop that doesn't reset out or read it elsewhere. Each call copies every earlier value.

xpd has no automatic fix, so make the change by hand.

xpd prints this message with the issue. A name that starts with $ stands for code from your file.

$OUT grows by one $NP.$F per pass, and each call copies every earlier value. Collect the pieces in a list and call np.concatenate once after the loop, which ran 130 to 170 times faster for 1,000 passes over 1,000,000 values.

Browse the other rules of the NumPy pack.