expm1(3)

✨ expm1(3) — exponential minus 1 ✨

okay bestie, welcome to expm1(3) 💖 this is a page from the Library Functions Manual, and we're about to make the whole "subtract 1 from an exponential" situation make sense. no cap.

---

🫶 NAME 🫶

expm1, expm1f, expm1l — exponential minus 1

---

📚 LIBRARY 📚

Math library (libm, -lm) — yeah you gotta link it, that's just the vibe.

---

🌸 SYNOPSIS 🌸

#include <math.h>

double expm1(double x);
float expm1f(float x);
long double expm1l(long double x);

🧚 Feature Test Macro Requirements for glibc 🧚

(peep feature_test_macros(7) if you want the lore)

expm1():

_ISOC99_SOURCE || _POSIX_C_SOURCE >= 200112L
    || _XOPEN_SOURCE >= 500
    || /* Since glibc 2.19: */ _DEFAULT_SOURCE
    || /* glibc <= 2.19: */ _BSD_SOURCE || _SVID_SOURCE

expm1f(), expm1l():

_ISOC99_SOURCE || _POSIX_C_SOURCE >= 200112L
    || /* Since glibc 2.19: */ _DEFAULT_SOURCE
    || /* glibc <= 2.19: */ _BSD_SOURCE || _SVID_SOURCE

---

💅 DESCRIPTION 💅

So these functions return a value equivalent to

exp(x) - 1

but here's the iconic part, the reason she exists: the result gets computed in a way that stays accurate even when x is near zero. that's the case where plain old exp(x) - 1 would be a whole mess, because you'd be subtracting two numbers that are nearly equal and losing all your precision. it's giving catastrophic cancellation. not on her watch. 🔥

---

👑 RETURN VALUE 👑

On success, these functions return exp(x) - 1. periodt.

• If x is a NaN, a NaN is returned. (NaN in, NaN out, the circle of life 🦋)
• If x is +0 (-0), +0 (-0) is returned.
• If x is positive infinity, positive infinity is returned. 🌈
• If x is negative infinity, -1 is returned.
• If the result overflows, a range error occurs, and the functions return -HUGE_VAL, -HUGE_VALF, or -HUGE_VALL, respectively.

---

😭 ERRORS 😭

Go read math_error(7) if you wanna know how to figure out whether an error even happened when you called one of these. it's all in there.

The following errors can occur:

• Range error, overflow — errno is set to ERANGE (but see BUGS, bestie, there's always a BUGS section). An overflow floating-point exception (FE_OVERFLOW) is raised.

---

🎀 ATTRIBUTES 🎀

for what all these terms even mean, see attributes(7).

• Interface: expm1(), expm1f(), expm1l()
• Attribute: Thread safety
• Value: MT-Safe

thread safe, no drama, we love her for that 🫶

---

⭐ STANDARDS ⭐

C11, POSIX.1-2008.

---

🦋 HISTORY 🦋

C99, POSIX.1-2001. BSD.

ancient. legendary. still standing. 💜

---

💀 BUGS 💀

okay so the old versions were lowkey messy, we have receipts:

• Before glibc 2.17, on certain architectures (e.g., x86, but not x86_64) expm1() raised a bogus underflow floating-point exception for some large negative x values (where the function result approaches -1). 🙈
• Before approximately glibc 2.11, expm1() raised a bogus invalid floating-point exception in addition to the expected overflow exception, and returned a NaN instead of positive infinity, for some large positive x values. 💀
• Before glibc 2.11, the glibc implementation did not set errno to ERANGE when a range error occurred.

all fixed now, glow-up complete, but iykyk.

---

🌈 SEE ALSO 🌈

exp(3), log(3), log1p(3)

---

that's the whole page bestie, you're officially an expm1 girlie 💖 sending you good vibes and accurate floating-point results, mwah 🎀 this was Linux man-pages 6.20, dated 2026-02-08, and the page itself is expm1(3) — go forth and compute ✨