Fourier analysis

Most often, the unqualified term Fourier transform refers to the transform of functions of a continuous real argument, and it produces a continuous function of frequency, known as a frequency distribution. One function is transformed into another, and the operation is reversible. When the domain of the input (initial) function is time (t), and the domain of the output (final) function is ordinary frequency, the transform of function s(t) at frequency f is given by the complex number:

${\displaystyle S(f)=\int _{-\infty }^{\infty }s(t)\cdot e^{-i2\pi ft}\,dt.}$

Evaluating this quantity for all values of f produces the frequency-domain function. Then s(t) can be represented as a recombination of complex exponentials of all possible frequencies:

${\displaystyle s(t)=\int _{-\infty }^{\infty }S(f)\cdot e^{i2\pi ft}\,df,}$

which is the inverse transform formula. The complex number, S(f), conveys both amplitude and phase of frequency f.

See Fourier transform for much more information, including:

• conventions for amplitude normalization and frequency scaling/units
• transform properties
• tabulated transforms of specific functions
• an extension/generalization for functions of multiple dimensions, such as images.

The Fourier transform of a periodic function, s_P(t), with period P, becomes a Dirac comb function, modulated by a sequence of complex coefficients:

${\displaystyle S[k]={\frac {1}{P}}\int _{P}s_{P}(t)\cdot e^{-i2\pi {\frac {k}{P}}t}\,dt,\quad k\in \mathbb {Z} ,}$     (where ∫_P is the integral over any interval of length P).

The inverse transform, known as Fourier series, is a representation of s_P(t) in terms of a summation of a potentially infinite number of harmonically related sinusoids or complex exponential functions, each with an amplitude and phase specified by one of the coefficients:

${\displaystyle s_{P}(t)\ \ =\ \ {\mathcal {F}}^{-1}\left\{\sum _{k=-\infty }^{+\infty }S[k]\,\delta \left(f-{\frac {k}{P}}\right)\right\}\ \ =\ \ \sum _{k=-\infty }^{\infty }S[k]\cdot e^{i2\pi {\frac {k}{P}}t}.}$

Any s_P(t) can be expressed as a periodic summation of another function, s(t):

${\displaystyle s_{P}(t)\,\triangleq \,\sum _{m=-\infty }^{\infty }s(t-mP),}$

and the coefficients are proportional to samples of S(f) at discrete intervals of 1/P:

${\displaystyle S[k]={\frac {1}{P}}\cdot S\left({\frac {k}{P}}\right).}$[upper-alpha 1]

Note that any s(t) whose transform has the same discrete sample values can be used in the periodic summation. A sufficient condition for recovering s(t) (and therefore S(f)) from just these samples (i.e. from the Fourier series) is that the non-zero portion of s(t) be confined to a known interval of duration P, which is the frequency domain dual of the Nyquist–Shannon sampling theorem.

See Fourier series for more information, including the historical development.

The DTFT is the mathematical dual of the time-domain Fourier series. Thus, a convergent periodic summation in the frequency domain can be represented by a Fourier series, whose coefficients are samples of a related continuous time function:

${\displaystyle S_{\frac {1}{T}}(f)\ \triangleq \ \underbrace {\sum _{k=-\infty }^{\infty }S\left(f-{\frac {k}{T}}\right)\equiv \overbrace {\sum _{n=-\infty }^{\infty }s[n]\cdot e^{-i2\pi fnT}} ^{\text{Fourier series (DTFT)}}} _{\text{Poisson summation formula}}={\mathcal {F}}\left\{\sum _{n=-\infty }^{\infty }s[n]\ \delta (t-nT)\right\},\,}$

which is known as the DTFT. Thus the DTFT of the s[n] sequence is also the Fourier transform of the modulated Dirac comb function.[upper-alpha 2]

The Fourier series coefficients (and inverse transform), are defined by:

${\displaystyle s[n]\ \triangleq \ T\int _{\frac {1}{T}}S_{\frac {1}{T}}(f)\cdot e^{i2\pi fnT}\,df=T\underbrace {\int _{-\infty }^{\infty }S(f)\cdot e^{i2\pi fnT}\,df} _{\triangleq \,s(nT)}.}$

Parameter T corresponds to the sampling interval, and this Fourier series can now be recognized as a form of the Poisson summation formula.  Thus we have the important result that when a discrete data sequence, s[n], is proportional to samples of an underlying continuous function, s(t), one can observe a periodic summation of the continuous Fourier transform, S(f). Note that any s(t) with the same discrete sample values produces the same DTFT  But under certain idealized conditions one can theoretically recover S(f) and s(t) exactly. A sufficient condition for perfect recovery is that the non-zero portion of S(f) be confined to a known frequency interval of width 1/T.  When that interval is [−1/2T, 1/2T], the applicable reconstruction formula is the Whittaker–Shannon interpolation formula. This is a cornerstone in the foundation of digital signal processing.

Another reason to be interested in S_1/T(f) is that it often provides insight into the amount of aliasing caused by the sampling process.

Applications of the DTFT are not limited to sampled functions. See Discrete-time Fourier transform for more information on this and other topics, including:

• normalized frequency units
• windowing (finite-length sequences)
• transform properties
• tabulated transforms of specific functions

Similar to a Fourier series, the DTFT of a periodic sequence, ${\displaystyle s_{N}[n]}$, with period ${\displaystyle N}$, becomes a Dirac comb function, modulated by a sequence of complex coefficients (see DTFT § Periodic data):

${\displaystyle S[k]=\sum _{n}s_{N}[n]\cdot e^{-i2\pi {\frac {k}{N}}n},\quad k\in \mathbb {Z} ,}$     (where Σ_n is the sum over any sequence of length N).

The S[k] sequence is what is customarily known as the DFT of one cycle of s_N. It is also N-periodic, so it is never necessary to compute more than N coefficients. The inverse transform, also known as a discrete Fourier series, is given by:

${\displaystyle s_{N}[n]={\frac {1}{N}}\sum _{k}S[k]\cdot e^{i2\pi {\frac {n}{N}}k},}$   where Σ_k is the sum over any sequence of length N.

When s_N[n] is expressed as a periodic summation of another function:

${\displaystyle s_{N}[n]\,\triangleq \,\sum _{m=-\infty }^{\infty }s[n-mN],}$   and   ${\displaystyle s[n]\,\triangleq \,s(nT),}$[upper-alpha 3]

the coefficients are proportional to samples of S_1/T(f) at disrete intervals of 1/P = 1/NT:

${\displaystyle S[k]={\frac {1}{T}}\cdot S_{\frac {1}{T}}\left({\frac {k}{P}}\right).}$[upper-alpha 4]

Conversely, when one wants to compute an arbitrary number (N) of discrete samples of one cycle of a continuous DTFT, S_1/T(f), it can be done by computing the relatively simple DFT of s_N[n], as defined above. In most cases, N is chosen equal to the length of non-zero portion of s[n]. Increasing N, known as zero-padding or interpolation, results in more closely spaced samples of one cycle of S_1/T(f). Decreasing N, causes overlap (adding) in the time-domain (analogous to aliasing), which corresponds to decimation in the frequency domain. (see Discrete-time Fourier transform § L=N×I) In most cases of practical interest, the s[n] sequence represents a longer sequence that was truncated by the application of a finite-length window function or FIR filter array.

The DFT can be computed using a fast Fourier transform (FFT) algorithm, which makes it a practical and important transformation on computers.

See Discrete Fourier transform for much more information, including:

• transform properties
• applications
• tabulated transforms of specific functions

For periodic functions, both the Fourier transform and the DTFT comprise only a discrete set of frequency components (Fourier series), and the transforms diverge at those frequencies. One common practice (not discussed above) is to handle that divergence via Dirac delta and Dirac comb functions. But the same spectral information can be discerned from just one cycle of the periodic function, since all the other cycles are identical. Similarly, finite-duration functions can be represented as a Fourier series, with no actual loss of information except that the periodicity of the inverse transform is a mere artifact.

It is common in practice for the duration of s(•) to be limited to the period, P or N.  But these formulas do not require that condition.

s(t) transforms (continuous-time)

	Continuous frequency	Discrete frequencies
Transform	${\displaystyle S(f)\,\triangleq \,\int _{-\infty }^{\infty }s(t)\cdot e^{-i2\pi ft}\,dt}$	${\displaystyle \overbrace {{\frac {1}{P}}\cdot S\left({\frac {k}{P}}\right)} ^{S[k]}\,\triangleq \,{\frac {1}{P}}\int _{-\infty }^{\infty }s(t)\cdot e^{-i2\pi {\frac {k}{P}}t}\,dt\equiv {\frac {1}{P}}\int _{P}s_{P}(t)\cdot e^{-i2\pi {\frac {k}{P}}t}\,dt}$
Inverse	${\displaystyle s(t)=\int _{-\infty }^{\infty }S(f)\cdot e^{i2\pi ft}\,df}$	${\displaystyle \underbrace {s_{P}(t)=\sum _{k=-\infty }^{\infty }S[k]\cdot e^{i2\pi {\frac {k}{P}}t}} _{\text{Poisson summation formula (Fourier series)}}\,}$

s(nT) transforms (discrete-time)

	Continuous frequency	Discrete frequencies
Transform	${\displaystyle \underbrace {{\frac {1}{T}}S_{\frac {1}{T}}(f)\,\triangleq \,\sum _{n=-\infty }^{\infty }s(nT)\cdot e^{-i2\pi fnT}} _{\text{Poisson summation formula (DTFT)}}}$	${\displaystyle {\begin{aligned}\overbrace {{\frac {1}{T}}S_{\frac {1}{T}}\left({\frac {k}{NT}}\right)} ^{S[k]}\,&\triangleq \,\sum _{n=-\infty }^{\infty }s(nT)\cdot e^{-i2\pi {\frac {kn}{N}}}\\&\equiv \underbrace {\sum _{n}s_{P}(nT)\cdot e^{-i2\pi {\frac {kn}{N}}}} _{\text{DFT}}\,\end{aligned}}}$
Inverse	${\displaystyle s(nT)=T\int _{\frac {1}{T}}{\frac {1}{T}}S_{\frac {1}{T}}(f)\cdot e^{i2\pi fnT}\,df}$ ${\displaystyle \sum _{n=-\infty }^{\infty }s(nT)\cdot \delta (t-nT)=\underbrace {\int _{-\infty }^{\infty }{\frac {1}{T}}\ S_{\frac {1}{T}}(f)\cdot e^{i2\pi ft}\,df} _{\text{inverse Fourier transform}}\,}$	${\displaystyle {\begin{aligned}s_{P}(nT)&=\overbrace {{\frac {1}{N}}\sum _{k}S[k]\cdot e^{i2\pi {\frac {kn}{N}}}} ^{\text{inverse DFT}}\\&={\tfrac {1}{P}}\sum _{k}S_{\frac {1}{T}}\left({\frac {k}{P}}\right)\cdot e^{i2\pi {\frac {kn}{N}}}\end{aligned}}}$