The root test is a criterion in mathematical analysis for deciding whether an infinite series converges absolutely or diverges. It examines the exponential rate of growth or decay of the terms by taking their th roots. The test applies to series with real or complex terms and is particularly useful when the terms contain expressions raised to the th power. It is named after Augustin-Louis Cauchy.
Statement
Let
be a series of real or complex numbers, and define
Here denotes the absolute value or complex modulus, and is the limit superior, which may equal .
The root test states:
- If , the series converges absolutely.
- If , including , the series diverges.
- If , the test is inconclusive: the series may converge absolutely, converge conditionally, or diverge.
When the ordinary limit
exists, it equals , so the same conclusions apply. The limit-superior formulation is more general because it remains applicable when the sequence of roots oscillates.
Changing, adding, or deleting finitely many initial terms does not affect convergence or the limiting root value.
Principle and proof
The root test compares the terms with those of a geometric series.
Absolute convergence when
Choose a real number such that
By the definition of limit superior, there is an integer such that
Consequently,
Since converges, the comparison test shows that converges. Thus converges absolutely.
This argument also gives a bound on the tail. Whenever for all ,
Divergence when
Choose with
There are infinitely many indices for which
and hence
The terms therefore do not tend to zero, violating a necessary condition for convergence.
In the limit-superior formulation, this does not mean that every sufficiently late term is large. Some subsequences may still tend to zero; infinitely many large terms are enough to establish divergence.
Examples
Absolute convergence
Consider
Taking th roots gives
Because , the series converges absolutely.
For the series
the corresponding calculation is
The factor does not alter the limiting exponential decay rate.
Divergence
For
the root limit is
so the series diverges.
The inconclusive case
Each of the following series has root limit :
Nevertheless, the first converges absolutely, the second diverges, and the third converges conditionally.
Thus is genuinely undecided by the root test. It is also possible for the terms not to approach zero in this case: the series has root limit .
Relationship to the ratio test
The ratio test examines consecutive terms through
If the terms are eventually nonzero and this ratio has a limit , finite or infinite, then
Accordingly, whenever the ordinary ratio test gives a decisive result, the root test gives the same result.
The root test can also succeed when the ratios have no limit. For example, let
The roots alternate between and , so
The series converges, although the consecutive-term ratios tend to zero along one parity and grow without bound along the other.
In practice, the ratio test often simplifies expressions involving factorials, while the root test often simplifies expressions containing th powers.
Application to power series
For a power series
define
The root-test quantity at a fixed point is
It follows that the radius of convergence is
with the conventions and . This is the Cauchy–Hadamard formula.
The series converges absolutely for and diverges for . When , points on the boundary give root-test value , so convergence there requires separate analysis.
Scope and limitations
The root test detects exponential-scale behavior. For eventually nonzero terms, the identity
expresses this in terms of the natural logarithm: the test measures the average logarithmic growth or decay per index.
Polynomial decay, such as for fixed , produces root limit , regardless of whether the resulting series converges. More generally, subexponential decay can fall outside the test’s decisive range.
Because it uses absolute values, the root test does not detect convergence caused by cancellation between positive and negative terms or between complex phases. Every conditionally convergent series has root limit superior : a smaller value would imply absolute convergence, while a larger value would imply divergence.