y=gd⁡x=arcsin⁡(tanh⁡x)y = \operatorname{gd} x = \arcsin(\tanh x)

Graph of the Gudermannian y=gd⁡xy = \operatorname{gd} x

y=gd⁡x=arcsin⁡(tanh⁡x)y = \operatorname{gd} x = \arcsin(\tanh x) is called the Gudermannian function, and it ties the trigonometric and the hyperbolic functions together without the help of complex numbers1. It is named after the nineteenth-century German mathematician Gudermann, and it is also the latitude itself in the Mercator map projection.

Domain and range

  • The domain is all real numbers
  • The range is −π2<y<π2-\dfrac{\pi}{2} < y < \dfrac{\pi}{2}
  • It increases monotonically
  • It is an odd function

The range of tanh⁡x\tanh x is (−1,1)(-1, 1), so the argument of arcsin⁡\arcsin always falls within its domain. The lines y=±π2y = \pm\dfrac{\pi}{2} are horizontal asymptotes; the values close in on them without ever arriving.

Symmetry and monotonicity

Both tanh⁡\tanh and arcsin⁡\arcsin are odd, so their composition is odd and has point symmetry about the origin. The derivative simplifies remarkably.

ddxarcsin⁡(tanh⁡x)=11−tanh⁡2x⋅sech⁡2x=sech⁡2xsech⁡x=sech⁡x\begin{align*} \frac{d}{dx}\arcsin(\tanh x) &= \frac{1}{\sqrt{1 - \tanh^2 x}} \cdot \operatorname{sech}^2 x \\ &= \frac{\operatorname{sech}^2 x}{\operatorname{sech} x} = \operatorname{sech} x \end{align*}

So gd⁡′(x)=sech⁡x\operatorname{gd}'(x) = \operatorname{sech} x, and the function may equally be defined by an integral.

gd⁡x=∫0xsech⁡t dt\operatorname{gd} x = \int_0^{x} \operatorname{sech} t\,dt

Since sech⁡x\operatorname{sech} x is always positive the function increases monotonically. The slope at the origin is sech⁡0=1\operatorname{sech} 0 = 1, so the graph is tangent there to the line y=xy = x.

Concavity

The second derivative is y′′=−sech⁡xtanh⁡xy'' = -\operatorname{sech} x \tanh x, whose sign is the opposite of the sign of xx. The curve is concave up for x<0x < 0 and concave down for x>0x > 0, with the origin as its only inflection point.

A bridge between the two families

The point of this function is a set of identities that hold once one writes θ=gd⁡x\theta = \operatorname{gd} x.

On the trigonometric sideOn the hyperbolic side
sin⁡θ\sin\thetatanh⁡x\tanh x
cos⁡θ\cos\thetasech⁡x\operatorname{sech} x
tan⁡θ\tan\thetasinh⁡x\sinh x
sec⁡θ\sec\thetacosh⁡x\cosh x

All of them hold at once. The function is a dictionary that translates hyperbolic functions into trigonometric ones with no recourse to complex numbers.

The inverse and the integral of the secant

The inverse is gd⁡−1(y)=artanh⁡(sin⁡y)\operatorname{gd}^{-1}(y) = \operatorname{artanh}(\sin y), which can also be written as follows.

gd⁡−1(y)=ln⁡(sec⁡y+tan⁡y)\operatorname{gd}^{-1}(y) = \ln(\sec y + \tan y)

The right-hand side is nothing but the main part of the integral ∫sec⁡y dy=ln⁡∣sec⁡y+tan⁡y∣+C\int \sec y\,dy = \ln|\sec y + \tan y| + C. The formula for the integral of the secant, which looks so abrupt in a table of integrals, turns out to be this function.

The Mercator projection

In the Mercator projection a place at latitude φ\varphi is drawn at the vertical coordinate ∫0φsec⁡t dt\int_0^{\varphi} \sec t\,dt2. That is gd⁡−1(φ)\operatorname{gd}^{-1}(\varphi), so conversely the latitude corresponding to a vertical coordinate xx on the map is gd⁡x\operatorname{gd} x.

xx, the vertical coordinategd⁡x\operatorname{gd} x in radiansLatitude
00000∘0^\circ
11≈0.8657\approx 0.8657≈49.60∘\approx 49.60^\circ
22≈1.3018\approx 1.3018≈74.58∘\approx 74.58^\circ
π\pi≈1.4842\approx 1.4842≈85.05∘\approx 85.05^\circ

Being tangent to y=xy = x near the origin corresponds to the small distortion near the equator, and approaching ±π2\pm\dfrac{\pi}{2} at the ends corresponds to the poles being banished to infinity. Web Mercator, used widely by map services, cuts the latitude off at ±85.05∘\pm 85.05^\circ because it cuts the map at x=±πx = \pm\pi to make it square.

  1. Gudermannian function, Wikipedia
  2. Mercator projection, Wikipedia