AP Calculus AB and BC

Limit of arctan(x) at Negative Infinity

The limit of arctan x as x approaches negative infinity is negative pi over 2, roughly negative 1.5708. Arctangent is bounded below by that value and never reaches it, so y = negative pi over 2 is a horizontal asymptote on the left. At positive infinity the limit is pi over 2.

limxarctanx=π2\lim_{x \to -\infty} \arctan x = -\frac{\pi}{2}

Settled by the range of the arctangent.

Where the floor comes from

Tangent repeats, so its inverse has to be built on a single branch. The branch chosen runs between the two vertical asymptotes nearest the origin, and on it tangent takes every real value exactly once.

tan:(π2,π2)(,)\tan : \left(-\frac{\pi}{2}, \frac{\pi}{2}\right) \to (-\infty, \infty)

Inverting swaps domain and range, so arctan\arctan accepts any real number and returns an angle strictly inside that open interval. The interval answers the limit before any computation begins.

arctan:(,)(π2,π2)\arctan : (-\infty, \infty) \to \left(-\frac{\pi}{2}, \frac{\pi}{2}\right)

Now read the limit as a question about angles. A hugely negative input asks which angle in that interval has a hugely negative tangent, and tangent runs down to -\infty only as the angle drops toward π2-\frac{\pi}{2}. The output is pressed against the bottom edge.

xxarctanx\arctan x
1-10.7854-0.7854
10-101.4711-1.4711
100-1001.5608-1.5608
1000-10001.5698-1.5698

The values fall steadily and stay above π21.5708-\frac{\pi}{2} \approx -1.5708, closing the gap without ever shutting it.

Why substitution fails, and why the limit has to exist

As at any infinite endpoint there is nothing to substitute, and unlike a rational function no indeterminate form shows up once you look. There is no \frac{\infty}{\infty} and no \infty - \infty, because a single bounded function has no competitor.

Existence is settled in advance. Arctangent decreases as xx decreases and is bounded below, and a monotone function with a bound in the direction it is heading must converge. All that remains is identifying the value, which the range supplies.

An unbounded input does not force an unbounded output

The instinct that a runaway input drags the output with it is what makes this limit feel wrong. arctanx\arctan x, sinx\sin x, and xx2+1\frac{x}{\sqrt{x^{2}+1}} all accept every real input and return values confined to a fixed window.

Arctangent is odd, so the left end mirrors the right: arctan(x)=arctanx\arctan(-x) = -\arctan x converts the value π2\frac{\pi}{2} at \infty into π2-\frac{\pi}{2} here. The graph carries two horizontal asymptotes, which is what Topic 1.15 questions probe.

The mistakes students make

The first is answering -\infty, from picturing tangent instead of its inverse. Tangent does blow up near π2-\frac{\pi}{2}, but that is the same relationship read the other way. Arctangent is the bounded direction.

The second is answering 90-90. AP limits are in radians, so the exact answer is π2-\frac{\pi}{2}, and a decimal is acceptable only alongside it.

The third is naming one horizontal asymptote when the question asks for all of them.

LimitValue
limxarctanx\lim_{x \to -\infty} \arctan xπ2-\frac{\pi}{2}
limxarctanx\lim_{x \to \infty} \arctan xπ2\frac{\pi}{2}
limx0arctan(1x)\lim_{x \to 0^{-}} \arctan\left(\frac{1}{x}\right)π2-\frac{\pi}{2}
limxarctan(ex)\lim_{x \to -\infty} \arctan\left(e^{x}\right)00

The last row is a reminder to read the inside first. There the inner function heads to 0 rather than to an infinite end, so the outer arctangent contributes nothing dramatic.

Not sure which technique a limit wants?

The Limit Method Chooser walks the decision from direct substitution through factoring, the conjugate, and L'Hopital, and says why each one applies or fails.

Frequently asked questions

Does arctan x ever equal negative pi over 2?

No. The range is the open interval, so the value is approached and never attained. If some input returned exactly π2-\frac{\pi}{2}, then tan(π2)\tan\left(-\frac{\pi}{2}\right) would have to equal that input, and tangent is undefined there.

How can one function have two horizontal asymptotes?

The two ends are independent questions, and each one records its own limit. So a function can have an asymptote on the left, on the right, on both, or on neither. Arctangent has y=π2y = -\frac{\pi}{2} and y=π2y = \frac{\pi}{2}, and xx2+1\frac{x}{\sqrt{x^{2}+1}} has y=1y = -1 and y=1y = 1.

Where does this limit appear outside graphing?

In improper integrals. arctanx\arctan x antidifferentiates 11+x2\frac{1}{1+x^{2}}, so evaluating at a lower infinite endpoint means taking this limit. That gives 011+x2dx=π2\int_{-\infty}^{0} \frac{1}{1+x^{2}}\,dx = \frac{\pi}{2}, and π\pi across the whole real line.