Write with . On a fixed branch of the complex logarithm on which and , the equation becomes
Its real and imaginary parts give the same condition , or
This is a logarithmic spiral. For the principal complex logarithm, it is the portion parametrized by that avoids the branch cut.
Use the principal complex logarithm, for which , and set
Then the definition of complex exponentiation gives
and therefore
Other choices of branch of the complex logarithm produce further valid answers.
Since and ,
The powers of the imaginary unit have period four. Hence
while
Thus , so
Both and its complement are dense. Indeed, inside any sufficiently small ternary cylinder, extend its finite digit prefix by a tail containing infinitely many s to obtain a point of , or by a tail of all zeros to obtain a point outside . Therefore every nondegenerate interval has supremum one and infimum zero for the indicator function. Its upper integral is one and its lower integral is zero, so
Let be the union of the closed intervals remaining after steps of the Cantor set construction. Their total length is . Any point outside has a ternary expansion with a among its first digits, so . Endpoints of deleted intervals also belong to because they have an alternative ternary expansion containing a , as in .
A partition using all endpoints of has lower Darboux sum at least
Since is dense, every upper Darboux sum is one. Letting gives equal upper and lower integrals, and therefore
For a partition of an interval , put
The upper Darboux sum and lower Darboux sum are
The upper and lower integrals are
The bounded function is Riemann integrable when these values agree, and their common value is its Riemann integral.
For the indicator of the rational numbers, every nondegenerate interval contains both a rational number and an irrational number. Hence every upper sum is one and every lower sum is zero. The function is not Riemann integrable.
Suppose diverges. Positivity implies . Put for . Since ,
If converged, then eventually and
Summing would bound the telescoping quantity , contradicting . Therefore
The converse is true. If , then and
Taking the contrapositive shows that divergence of the ratio series forces divergence of .
The arithmetic-geometric mean inequality gives
Thus convergence of implies convergence of by the comparison test for series.
The converse is false. Define
The even terms make diverge, while
so the neighboring geometric-mean series converges.
For positive ,
The comparison test for series therefore gives
No. The property
is extensional: it depends only on the partial function computed, not on its program index. It is nontrivial, since some partial recursive functions have a nine-element domain and others do not. The Rice theorem therefore implies that its index set
is undecidable. Hence
Apply part (b) to
There is an index such that . Its domain of a partial function is exactly
Therefore
By the S-m-n theorem, there is a total recursive function such that
for all . Apply the recursion theorem to . It supplies an index for which
Consequently
Fix an effective enumeration of the -ary partial recursive functions.
The S-m-n theorem says that program inputs can be compiled into an index: for every there is a total recursive such that
Here means that either both sides are undefined or both are defined and equal.
The recursion theorem says that every total recursive transformation of program indices has a semantic fixed point: some satisfies
The Rice theorem says that every nontrivial extensional property of programs computed by partial recursive functions is undecidable.
If counts complex multiplications and a twiddle product is reused in the two butterfly outputs, then
For , solving the recurrence gives
The radix-two Fast Fourier transform recursively applies part (a) until it reaches one-point transforms, then combines adjacent even and odd transforms in butterfly operations. For , the decomposition tree is
followed by the reverse sequence of twiddle-factor butterfly combinations. This is the usual three-stage FFT network.
Let and . Since ,
Hence, for , the even--odd fast Fourier transform recursion is
Only one twiddle-factor multiplication is needed for each value of .

There are unlisted articles, also show them or only show them.