Hyperbolic Tangent Calculator

The Hyperbolic Tangent Calculator computes $$\tanh(x)$$ for any real number $$x$$. The hyperbolic tangent is defined as the ratio of hyperbolic sine to hyperbolic cosine and is one of the most important activation functions in modern machine learning and neural networks.

The hyperbolic tangent is defined as:

$$\tanh(x) = \frac{\sinh(x)}{\cosh(x)} = \frac{e^x - e^{-x}}{e^x + e^{-x}}$$

This function maps any real number to a value strictly between -1 and 1, producing a smooth, S-shaped (sigmoid-like) curve that passes through the origin. It smoothly saturates to $$\pm 1$$ as $$x \to \pm\infty$$.

Key Properties

• Domain: All real numbers $$(-\infty, +\infty)$$
• Range: $$(-1, 1)$$ — strictly between -1 and 1
• Symmetry: $$\tanh(-x) = -\tanh(x)$$ (odd function, zero-centered)
• At zero: $$\tanh(0) = 0$$
• Limits: $$\lim_{x \to +\infty} \tanh(x) = 1$$, $$\lim_{x \to -\infty} \tanh(x) = -1$$
• Derivative: $$\frac{d}{dx}\tanh(x) = 1 - \tanh^2(x) = \text{sech}^2(x)$$
• Taylor series: $$\tanh(x) = x - \frac{x^3}{3} + \frac{2x^5}{15} - \frac{17x^7}{315} + \cdots$$

Relationship to the Logistic Sigmoid

The hyperbolic tangent is closely related to the logistic sigmoid function $$\sigma(x) = \frac{1}{1 + e^{-x}}$$:

$$\tanh(x) = 2\sigma(2x) - 1$$

Both functions are sigmoid-shaped, but $$\tanh$$ is zero-centered (output range $$(-1,1)$$) while the logistic sigmoid maps to $$(0,1)$$. This zero-centered property often makes $$\tanh$$ preferable as an activation function in neural networks, since it helps keep the mean of activations near zero.

The tanh in Neural Networks

In deep learning, $$\tanh$$ was historically one of the most popular activation functions before ReLU became dominant. Its advantages include:

• Zero-centered output, which can speed up convergence
• Smooth and differentiable everywhere
• Bounded output prevents exploding activations

However, tanh suffers from the vanishing gradient problem: for large $$|x|$$, the derivative $$1 - \tanh^2(x)$$ approaches zero, making gradient-based learning extremely slow in deep networks.

Applications

Beyond machine learning, $$\tanh$$ appears in many scientific contexts. In special relativity, the velocity addition formula for collinear velocities uses $$\tanh$$: if rapidities add as $$\phi = \phi_1 + \phi_2$$, then $$v/c = \tanh(\phi)$$. In fluid dynamics, $$\tanh$$ profiles describe shear layers and mixing zones. In magnetism, the Brillouin function (which describes paramagnetic magnetization) reduces to $$\tanh$$ for spin-1/2 particles. In electrical engineering, $$\tanh$$ appears in impedance matching and transmission line calculations. The function also models dose-response curves in pharmacology and population saturation in ecology.