Creating Arrays

Zeros, Ones, and Full

Three fill loops build constant-value float buffers of length 4: all zeros, all ones, and all sevens. The replay shows the same append pattern repeated three times, with each list growing one element per iteration.

By hand

With NumPy

np.zeros(n), np.ones(n), and np.full(n, c) each allocate a typed buffer and fill it in one call without an explicit loop. The snapshot shows each array's shape, element type, and values in a labeled block.

naive.py 
n = 4
c = 7.0
zeros = []
for i in range(n):
    zeros.append(0.0)
ones = []
for i in range(n):
    ones.append(1.0)
full = []
for i in range(n):
    full.append(c)
print('RESULT:', (zeros, ones, full))
library.py 
import numpy as np

n = 4
c = 7.0
zeros = np.zeros(n)
ones = np.ones(n)
full = np.full(n, c)
for name, arr in [('zeros', zeros), ('ones', ones), ('full', full)]:
    print(f'{name}: shape: {arr.shape} dtype: {arr.dtype} values: {arr.tolist()}')
print('RESULT:', (zeros.tolist(), ones.tolist(), full.tolist()))

zeros: shape: (4,) dtype: float64 values: [0.0, 0.0, 0.0, 0.0]
ones: shape: (4,) dtype: float64 values: [1.0, 1.0, 1.0, 1.0]
full: shape: (4,) dtype: float64 values: [7.0, 7.0, 7.0, 7.0]
RESULT: ([0.0, 0.0, 0.0, 0.0], [1.0, 1.0, 1.0, 1.0], [7.0, 7.0, 7.0, 7.0])

Implementation notes

  • All three functions default to dtype=float64. Pass dtype=int (or another dtype) to change the element type.
  • np.zeros is the conventional way to pre-allocate an output buffer before filling it with computed values — a common pattern in the reduction and broadcasting lessons that follow.
  • np.empty(n) is faster than np.zeros(n) because it skips the initialisation step, but the contents are undefined. Use it only when every element will be overwritten before it is read.
  • np.zeros_like(arr) and np.ones_like(arr) create a zero/one array with the same shape and dtype as an existing array — useful when the shape is not known until runtime.
  • Shape, dtype, and values are shown explicitly here because ndarray.__repr__ output varies with NumPy version and print options.