Python Arrays

Python Arrays: Built-in array Module & Memory Optimization

While standard Python list objects store pointers to arbitrary objects, the array module provides compact, homogeneous numerical arrays stored in contiguous C-style memory buffers.

Memory Layout: List vs Array

flowchart LR
    subgraph Python List (Pointer Array)
        p0["Ptr [0]"] --> PyObject1["PyObject (int 10)"]
        p1["Ptr [1]"] --> PyObject2["PyObject (int 20)"]
    end
    subgraph Python Array (Contiguous C Buffer)
        buf["Contiguous Memory: [ 10 | 20 | 30 | 40 ]"]
    end

Array Type Codes

• 'i': Signed C integer (2 or 4 bytes).
• 'f': Floating-point number (4 bytes).
• 'd': Double-precision float (8 bytes).

Practical Code Example

import array
import sys

def demonstrate_array_efficiency() -> None:
    count = 100_000
    
    # Standard List of Ints
    py_list = list(range(count))
    
    # Contiguous C Array of Ints ('i' typecode)
    c_array = array.array('i', range(count))

    # Memory Size Comparison
    list_size = sys.getsizeof(py_list)
    array_size = sys.getsizeof(c_array)

    print(f"Memory overhead for {count:,} integers:")
    print(f" Standard List: {list_size / (1024*1024):.2f} MB")
    print(f" Contiguous C Array: {array_size / (1024*1024):.2f} MB")
    print(f" Memory Saved: {((list_size - array_size) / list_size) * 100:.1f}%")

if __name__ == "__main__":
    demonstrate_array_efficiency()

Best Practices & Gotchas

• Homogeneous Requirement: array.array can only store items of the specified type code. Appending a string to an integer array raises TypeError.
• Use NumPy for Heavy Math: For matrix math, linear algebra, or multidimensional arrays, use NumPy (numpy.ndarray) instead of the standard array module.
• Fast Binary I/O: array.tofile() and array.fromfile() allow extremely fast raw binary disk reads and writes.

Self-Check Challenge

Create a floating-point array using type code 'd' with values [1.1, 2.2, 3.3], append 4.4, and print its memory representation.