Data Types

Integers

Data Type	Bit Size	Value Range
int8	8-bit	-128 to 127
int16	16-bit	-32,768 to 32,767
int32	32-bit	-2,147,483,648 to 2,147,483,647
int64	64-bit	-9,223,372,036,854,775,808 to 9,223,372,036,854,775,807

Unsigned Integers (Only Positive Values)

Data Type	Bit Size	Value Range
uint8	8-bit	0 to 255
uint16	16-bit	0 to 65,535
uint32	32-bit	0 to 4,294,967,295
uint64	64-bit	0 to 18,446,744,073,709,551,615

Floating-Points

Data Type	Bit Size	Value Range (Approximate)
float16	16-bit	±6.1 × 10⁻⁵ to ±65,504
float32	32-bit	±1.2 × 10⁻³⁸ to ±3.4 × 10³⁸
float64	64-bit	±2.2 × 10⁻³⁰⁸ to ±1.8 × 10³⁰⁸

These are approximate; some numbers might be represented more precisely, while others less so, depending on their exact binary representation.

• float32 has about 7 decimal digits of precision.

• float64 has about 15–17 decimal digits of precision.

Terms

Significand is the part of a floating-point number that contains its significant digits.

Machine epsilon is the smallest positive number that, when added to 1.0, results in a number distinguishable from 1.0:

• float32 (single precision): The machine epsilon is approximately 1.1920929e-07.

• float64 (double precision): The machine epsilon is approximately 2.220446049250313e-16.

float32 example:

• 1.0 + 1.1920929e-07, you’ll get a value that is slightly greater than 1.0.

• 1.0 + (a number smaller than 1.1920929e-07) might still result in 1.0.

import numpy as np

print(np.array(1, dtype='float32') + np.array(1.1920929e-07, dtype='float32'))
print(np.array(1, dtype='float32') + np.array(1e-08, dtype='float32'))

1.0000001
1.0