Element-Wise Arithmetic & Comparison Operations0%

The Rules of NumPy Broadcasting Explained

Element-Wise Arithmetic & Comparison Operations

Beginner12 min read•Updated: Jul 11, 2026

Element-wise Arithmetic & Logical Operators

In NumPy, mathematical and comparison operators applied to arrays operate element-by-element (element-wise) by default. This contrasts with linear algebra matrix conventions (where multiplication denotes dot products) or standard Python lists (where + concatenates lists and * duplicates them).

1. Basic Arithmetic Operators

All standard Python arithmetic operators have vectorized element-wise implementations in NumPy:

1import numpy as np
2 
3a = np.array([10, 20, 30, 40], dtype=float)
4b = np.array([2,  4,  5,  8], dtype=float)
5 
6print("Addition (a + b):        ", a + b)
7print("Subtraction (a - b):     ", a - b)
8print("Multiplication (a * b):  ", a * b)
9print("True Division (a / b):   ", a / b)
10print("Floor Division (a // b): ", a // b)
11print("Exponentiation (a ** 2): ", a ** 2)
12print("Modulus (a % b):         ", a % b)

Output:

1Addition (a + b):         [12. 24. 35. 48.]
2Subtraction (a - b):      [ 8. 16. 25. 32.]
3Multiplication (a * b):   [ 20.  80. 150. 320.]
4True Division (a / b):    [5. 5. 6. 5.]
5Floor Division (a // b):  [5. 5. 6. 5.]
6Exponentiation (a ** 2):  [ 100.  400.  900. 1600.]
7Modulus (a % b):          [0. 0. 0. 0.]

Pro Tip

Crucial Linear Algebra Distinction: In NumPy, a * b is element-wise Hadamard product. It is NOT matrix multiplication. For matrix multiplication, use the @ operator or np.matmul().

2. In-Place Arithmetic Operators

NumPy supports compound in-place operators (+=, -=, *=, /=), which modify the array's existing memory buffer directly without creating an intermediate array:

1counter = np.ones(5)
2print("Initial id:", id(counter))
3 
4# In-place addition modifies buffer directly
5counter += 5
6print("After += 5:", counter)
7print("Same object id?", id(counter))  # True!
8 
9# In contrast, counter = counter + 5 allocates a new object

Important Warning

Warning on In-Place Type Casting: In-place operators cannot automatically cast to a higher dtype. For instance, executing int_arr += 2.5 raises UFuncTypeError: Cannot cast ufunc 'add' output from dtype('float64') to dtype('int32') with casting rule 'same_kind'.

3. Comparison Operators

Applying comparison operators (==, !=, <, <=, >, >=) evaluates each element independently, producing a boolean array:

1readings = np.array([12.1, 15.4, 9.8, 22.0, 18.3])
2baseline = 15.0
3 
4print("Greater than baseline: ", readings > baseline)
5print("Equal to 22.0:         ", readings == 22.0)

Output:

1Greater than baseline:  [False  True False  True  True]
2Equal to 22.0:          [False False False  True False]

4. Array-Wide Logical Reductions: `all()` and `any()`

To evaluate whether all elements or at least one element satisfy a condition, use np.all() and np.any():

1quality_scores = np.array([98, 92, 95, 89, 94])
2 
3# Check if ALL parts pass quality threshold (> 85)
4all_passed = np.all(quality_scores > 85)
5print("Did all parts pass?", all_passed)  # True
6 
7# Check if ANY part has a near-perfect score (>= 98)
8has_perfect = np.any(quality_scores >= 98)
9print("Any near-perfect parts?", has_perfect)  # True

Evaluating Tolerant Floating-Point Equality:

Because of IEEE 754 floating-point rounding errors, direct equality a == b is dangerous for floats. Use np.isclose() or np.allclose():

1val1 = np.array([0.1 + 0.2])
2val2 = np.array([0.3])
3 
4print("Direct equality:", val1 == val2)                  # [False] !
5print("np.isclose:", np.isclose(val1, val2))             # [True]
6print("np.allclose:", np.allclose(val1, val2))           # True

Interactive Quiz• Knowledge Check

Multiple Choice Questions

Question 1

What is the output of np.array([1, 2, 3]) * np.array([2, 3, 4])?

Question 2

How does arr += 5 differ from arr = arr + 5?

Question 3

Which function reliably checks if two floating-point arrays are equal within a specified numerical tolerance?

Question 4

What does np.any(arr < 0) evaluate to?

Question 5

What happens if you try to perform int_arr += 1.5 on an integer array of dtype int32?

Next Lesson

The Rules of NumPy Broadcasting Explained

Continue learning with hands-on practice, examples, and exercises in the upcoming topic.

Related Lessons

Previous Lesson	Next Lesson
Vectorization: Eliminating Python Loops with SIMD	The Rules of NumPy Broadcasting Explained

Practice Quiz

Test your understanding of this lesson with 5 questions. Each question has one correct answer.

Previous Next

PreviousPrevNext

MSK Institute • Shikohabad

Loading Page Content...

Preparing courses, live labs, and learning resources.सामग्री लोड हो रही है, कृपया प्रतीक्षा करें...

Connecting serverPlease wait...

💻 Practical Labs🚀 Free Demo📜 Verified Certificates

The Rules of NumPy Broadcasting Explained

Element-Wise Arithmetic & Comparison Operations

Beginner12 min read•Updated: Jul 11, 2026

Study Materials Cheatsheet

Element-wise Arithmetic & Logical Operators

1. Basic Arithmetic Operators

All standard Python arithmetic operators have vectorized element-wise implementations in NumPy:

1import numpy as np
2 
3a = np.array([10, 20, 30, 40], dtype=float)
4b = np.array([2,  4,  5,  8], dtype=float)
5 
6print("Addition (a + b):        ", a + b)
7print("Subtraction (a - b):     ", a - b)
8print("Multiplication (a * b):  ", a * b)
9print("True Division (a / b):   ", a / b)
10print("Floor Division (a // b): ", a // b)
11print("Exponentiation (a ** 2): ", a ** 2)
12print("Modulus (a % b):         ", a % b)

Output:

1Addition (a + b):         [12. 24. 35. 48.]
2Subtraction (a - b):      [ 8. 16. 25. 32.]
3Multiplication (a * b):   [ 20.  80. 150. 320.]
4True Division (a / b):    [5. 5. 6. 5.]
5Floor Division (a // b):  [5. 5. 6. 5.]
6Exponentiation (a ** 2):  [ 100.  400.  900. 1600.]
7Modulus (a % b):          [0. 0. 0. 0.]

Pro Tip

Crucial Linear Algebra Distinction: In NumPy, a * b is element-wise Hadamard product. It is NOT matrix multiplication. For matrix multiplication, use the @ operator or np.matmul().

2. In-Place Arithmetic Operators

NumPy supports compound in-place operators (+=, -=, *=, /=), which modify the array's existing memory buffer directly without creating an intermediate array:

1counter = np.ones(5)
2print("Initial id:", id(counter))
3 
4# In-place addition modifies buffer directly
5counter += 5
6print("After += 5:", counter)
7print("Same object id?", id(counter))  # True!
8 
9# In contrast, counter = counter + 5 allocates a new object

Important Warning

3. Comparison Operators

Applying comparison operators (==, !=, <, <=, >, >=) evaluates each element independently, producing a boolean array:

1readings = np.array([12.1, 15.4, 9.8, 22.0, 18.3])
2baseline = 15.0
3 
4print("Greater than baseline: ", readings > baseline)
5print("Equal to 22.0:         ", readings == 22.0)

Output:

1Greater than baseline:  [False  True False  True  True]
2Equal to 22.0:          [False False False  True False]

4. Array-Wide Logical Reductions: `all()` and `any()`

To evaluate whether all elements or at least one element satisfy a condition, use np.all() and np.any():

1quality_scores = np.array([98, 92, 95, 89, 94])
2 
3# Check if ALL parts pass quality threshold (> 85)
4all_passed = np.all(quality_scores > 85)
5print("Did all parts pass?", all_passed)  # True
6 
7# Check if ANY part has a near-perfect score (>= 98)
8has_perfect = np.any(quality_scores >= 98)
9print("Any near-perfect parts?", has_perfect)  # True

Evaluating Tolerant Floating-Point Equality:

Because of IEEE 754 floating-point rounding errors, direct equality a == b is dangerous for floats. Use np.isclose() or np.allclose():

1val1 = np.array([0.1 + 0.2])
2val2 = np.array([0.3])
3 
4print("Direct equality:", val1 == val2)                  # [False] !
5print("np.isclose:", np.isclose(val1, val2))             # [True]
6print("np.allclose:", np.allclose(val1, val2))           # True

Interactive Quiz• Knowledge Check

Multiple Choice Questions

Question 1

What is the output of np.array([1, 2, 3]) * np.array([2, 3, 4])?

Question 2

How does arr += 5 differ from arr = arr + 5?

Question 3

Which function reliably checks if two floating-point arrays are equal within a specified numerical tolerance?

Question 4

What does np.any(arr < 0) evaluate to?

Question 5

What happens if you try to perform int_arr += 1.5 on an integer array of dtype int32?

Next Lesson

The Rules of NumPy Broadcasting Explained

Continue learning with hands-on practice, examples, and exercises in the upcoming topic.

Related Lessons

Previous Lesson	Next Lesson
Vectorization: Eliminating Python Loops with SIMD	The Rules of NumPy Broadcasting Explained

Practice Quiz

Test your understanding of this lesson with 5 questions. Each question has one correct answer.

Previous Next

Element-wise Arithmetic & Logical Operators

1. Basic Arithmetic Operators

2. In-Place Arithmetic Operators

3. Comparison Operators

4. Array-Wide Logical Reductions: all() and any()

Evaluating Tolerant Floating-Point Equality:

Multiple Choice Questions

The Rules of NumPy Broadcasting Explained

Related Lessons

Practice Quiz

Loading Page Content...

Element-wise Arithmetic & Logical Operators

1. Basic Arithmetic Operators

2. In-Place Arithmetic Operators

3. Comparison Operators

4. Array-Wide Logical Reductions: all() and any()

Evaluating Tolerant Floating-Point Equality:

Multiple Choice Questions

The Rules of NumPy Broadcasting Explained

Related Lessons

Practice Quiz

4. Array-Wide Logical Reductions: `all()` and `any()`

4. Array-Wide Logical Reductions: `all()` and `any()`