Complex Number Paradoxes Explained: Why Exponent Rules Fail

Complex exponentiation breaks the familiar high-school rules for powers because those rules — (x^a)^b = x^ab, (xy)^a = x^a y^a, and x^a = x^b ⇒ a = b — only hold when the base is a nonnegative real number and the exponents are real. Applying them to negative or complex bases produces contradictions such as -1 = 1. A rigorous complex exponential and a multivalued complex logarithm resolve the paradoxes, at the cost of losing single-valuedness.

Key Takeaways

• Three high-school exponent rules — (x^a)^b = x^ab, (xy)^a = x^a y^a, and x^a = x^b ⇒ a = b — are valid only for nonnegative real bases and real exponents, which is why naive manipulation of square roots of negative numbers produces false results like 1 = -1.
• The complex exponential is defined as e^(a+bi) = e^a(cos(b) + i sin(b)), a definition chosen so it agrees with the real exponential, stays continuous, and preserves the rule e^(z+w) = e^z e^w.
• The complex logarithm, written Log(z), is inherently multivalued: Log(1) equals every value in the set …, -4πi, -2πi, 0, 2πi, 4πi, …, not just 0.
• Restricting the complex logarithm to a single “principal value” (choosing k so that -π < θ + 2kπ ≤ π) makes it single-valued but destroys the identity log(xy) = log(x) + log(y).
• Complex exponentiation is defined as z^w = e^(w·Log(z)), which means expressions like (-1)^(1/2) are also multivalued — in this case yielding exactly the two values i and -i.

Why Do Complex Number Paradoxes Happen?

Several classic “proofs” appear to show that complex numbers behave inconsistently. Each relies on extending a real-number exponent rule into complex territory where that rule no longer applies.

Example A: √(-1)·√(-1) Seems to Equal Both -1 and 1

The chain of equalities -1 = i^2 = √-1 · √-1 = √((-1)(-1)) = √1 = 1 misapplies the rule (xy)^a = x^a y^a to x = y = -1 and a = 1/2. That rule only holds when x and y are nonnegative, so the step from √-1·√-1 to √((-1)(-1)) is invalid.

Example B: Rewriting (-1)^(1/2) as (-1)^(2/4) Seems to Force i = 1

The manipulation i = √-1 = (-1)^(1/2) = (-1)^(2/4) = [(-1)^2]^(1/4) = 1 misapplies the rule (x^a)^b = x^ab with x = -1, a = 2, and b = 1/4. This rule is valid only for positive x, so the substitution fails for a negative base.

Example C: Euler’s Identity Seems to Prove 2πi = 0

Euler’s identity gives e^(2πi) = 1, so it might seem that Log(1) = 2πi and Log(1) = 0 simultaneously, implying 2πi = 0. This conclusion relies on the rule x^a = x^b ⇒ a = b, which holds only for real exponents a and b, not complex ones.

Example D: A Chain of Substitutions Seems to Prove e^(-4π²) = 1

Starting from e^(2πi) = 1 and e^(1+2πi) = e, squaring the exponent produces e^((1+2πi)^2) = e^(1+4πi-4π²), which appears to force e^(-4π²) = 1 — an absurd result. This error comes from applying (x^a)^b = x^ab with complex exponents, where the rule does not hold.

How Is the Complex Exponential Defined?

The complex exponential is defined for any complex number a + bi as e^(a+bi) = e^a(cos(b) + i sin(b)), where e is Euler’s constant. This definition is not arbitrary. It is the unique function satisfying three properties simultaneously: it agrees with the real exponential when the imaginary part is zero, it is continuous for all complex inputs, and it preserves the addition rule e^(z+w) = e^z e^w. A full derivation of why these three properties uniquely determine the formula requires calculus beyond the scope of this explanation.

How Is the Complex Logarithm Defined?

Because e^0 = 1 and e^(2πi) = 1 simultaneously, any consistent definition of the complex logarithm Log(1) must allow both 0 and 2πi as valid values. The complex logarithm is therefore defined as a multivalued function: Log(z) = w whenever e^w = z.

Deriving the Formula for Log(z)

Writing z in polar form as z = r(cos(θ) + i sin(θ)), where r = √(x²+y²), leads to the general formula:

Log(z) = log(r) + i(θ + 2kπ), for any integer k.

Log(0) is left undefined, just as it is in the real-number case, because 0 cannot be written in polar form with a well-defined angle θ.

Worked Example: Log(1)

Writing 1 in polar form as 1 = 1(cos(0) + i sin(0)) gives Log(1) = log(1) + i(0 + 2kπ). Log(1) therefore takes on every value in the set …, -6πi, -4πi, -2πi, 0, 2πi, 4πi, 6πi, …, one value for each integer k.

The Principal Value Trade-off

To avoid working with a multivalued function, many authors restrict Log(z) to a single “principal value” by choosing the integer k such that -π < θ + 2kπ ≤ π. Under this convention, the principal value of Log(1) is 0. The trade-off is that the identity log(xy) = log(x) + log(y), which holds for real logarithms, no longer holds for the principal complex logarithm.

How Is Complex Exponentiation Defined?

Given complex numbers z and w with z nonzero, complex exponentiation is defined as z^w = e^(w·Log(z)). Because this definition depends on the multivalued complex logarithm, complex exponentiation is itself multivalued unless restricted to a principal value.

Worked Example: (-1)^(1/2)

Writing -1 in polar form as -1 = 1(cos(π) + i sin(π)) gives Log(-1) = (2k+1)πi, taking values such as …, -3πi, -πi, πi, 3πi, …. Substituting into the exponentiation formula gives (-1)^(1/2) = e^(kπi) · e^((π/2)i). Since e^((π/2)i) = i and e^(kπi) equals 1 for even k and -1 for odd k, (-1)^(1/2) takes on exactly two values: i and -i.

Frequently Asked Questions

Why doesn’t √(-1)·√(-1) equal √((-1)(-1))?

The rule that lets you combine two roots like this, (xy)^a = x^a y^a, is only valid when x and y are both nonnegative real numbers. Since -1 is negative, splitting or combining the square roots this way is invalid and produces the false result 1 = -1.

Is the complex logarithm a single function?

No. The complex logarithm Log(z) is defined as a multivalued function, meaning it produces infinitely many outputs for a single input, one for each integer k in the formula Log(z) = log(r) + i(θ + 2kπ). Log(1) alone equals every multiple of 2πi.

What is the “principal value” of a complex logarithm?

The principal value is a single chosen output from the infinitely many values of Log(z), selected using the integer k for which -π < θ + 2kπ ≤ π. It makes the logarithm single-valued but sacrifices the identity log(xy) = log(x) + log(y).

Does e^(z+w) = e^z · e^w still hold for complex numbers?

Yes. This is one of the properties preserved when extending the exponential function to complex numbers, along with agreement with the real exponential and continuity for all complex inputs.

What are the values of (-1)^(1/2)?

Using the definition z^w = e^(w·Log(z)), (-1)^(1/2) takes on exactly two values: i and -i. This follows from Log(-1) equaling (2k+1)πi for any integer k.

Why do the familiar exponent rules fail for complex numbers?

Rules such as (x^a)^b = x^ab, (xy)^a = x^a y^a, and x^a = x^b ⇒ a = b were only ever proven for nonnegative real bases and real exponents. Extending them to negative or complex bases without justification is what produces apparent paradoxes like 2πi = 0.

Related reading: Fixing Things Which Can Go Wrong With Complex Numbers