torch.js has not been released yet.

torch.true_divide

function true_divide(input: unknown, other: unknown, options: unknown): Tensor

Divides input by other, always producing floating-point output.

Unlike integer division which truncates, true_divide always performs floating-point division even when both operands are integers. This ensures mathematically correct division results. Essential for:

Numerical accuracy: Avoiding integer truncation in division
PyTorch compatibility: Matching Python 3's / operator behavior
Ratio computations: Computing fractions, percentages, proportions
Normalization: Dividing by counts or totals

\text{true\_divide}(a, b) = \frac{a}{b} \in \mathbb{R}

Parameters

inputunknown: The dividend tensor (numerator)
otherunknown: The divisor tensor or scalar (denominator)
optionsunknown: Optional settings including out parameter

Returns

Tensor– A tensor with floating-point division results

Examples

// Integer division vs true division
const a = torch.tensor([5, 7, 9]);
const b = torch.tensor([2, 3, 4]);
torch.true_divide(a, b);  // [2.5, 2.333..., 2.25]

// Scalar division
const x = torch.tensor([1, 2, 3, 4]);
torch.true_divide(x, 2);  // [0.5, 1.0, 1.5, 2.0]

// Computing percentages
const counts = torch.tensor([10, 25, 15]);
const total = torch.sum(counts);
const percentages = torch.true_divide(counts, total).mul(100);

See Also

PyTorch torch.true_divide()
div - Standard division (same behavior for floats)
floor_divide - Integer division with floor rounding
fmod - Floating-point remainder