DFT#

Versioned name: DFT-7

Category: Signal processing

Short description: DFT operation performs the discrete complex-to-complex Fourier transformation of input tensor by specified dimensions.

Attributes:

No attributes available.

Inputs

  • 1: data - Input tensor of type T with data for the DFT transformation. Type of elements is any supported floating-point type. The last dimension of the input tensor must be equal to 2, that is the input tensor shape must have the form [D_0, D_1, ..., D_{N-1}, 2], representing the real and imaginary components of complex numbers in [:, ..., :, 0] and in [:, ..., :, 1] correspondingly. Required.

  • 2: axes - 1D tensor of type T_IND specifying dimension indices where DFT is applied, and axes is any unordered list of indices of different dimensions of input tensor, for example, [0, 4], [4, 0], [4, 2, 1], [1, 2, 3], [-3, 0, -2]. These indices should be integers from -(r - 1) to (r - 2) inclusively, where r = rank(data). A negative axis a is interpreted as an axis r - 1 + a. Other dimensions do not change. The order of elements in axes attribute matters, and is mapped directly to elements in the third input signal_size. Required.

    Note

    The following constraint must be satisfied: rank(data) >= len(axes) + 1 and input_shape[-1] == 2 and (rank(data) - 1) not in axes and (-1) not in axes.

  • 3: signal_size - 1D tensor of type T_SIZE describing signal size with respect to axes from the input axes. If signal_size[i] == -1, then DFT is calculated for full size of the axis axes[i]. If signal_size[i] > input_shape[: r - 1][axes[i]], then input data are zero-padded with respect to the axis axes[i] at the end. Finally, signal_size[i] < input_shape[: r - 1][axes[i]], then input data are trimmed with respect to the axis axes[i]. More precisely, if signal_size[i] < input_shape[: r - 1][axes[i]], the slice 0: signal_size[i] of the axis axes[i] is considered. Optional, with default value `[input_shape[: r - 1][a] for a in axes]`.

    Note

    If the input signal_size is specified, the size of signal_size must be the same as the size of axes.

Outputs

  • 1: Resulting tensor with elements of the same type as input data tensor. The shape of the output is calculated as follows: if the input signal_size is not specified, the shape of output is the same as the shape of data. Otherwise, output_shape[axis] = input_shape[axis] for axis not in axes, and if signal_size[i] == -1, then output_shape[: r - 1][axes[i]] = input_shape[: r - 1][axes[i]], else output_shape[: r - 1][axes[i]] = signal_size[i].

Types

  • T: floating-point type.

  • T_IND: int64 or int32.

  • T_SIZE: int64 or int32.

Detailed description: DFT performs the discrete Fourier transformation of input tensor with respect to specified axes. Calculations are performed according to the following rules.

For simplicity, assume that an input tensor A has the shape [B_0, ..., B_{k-1}, M_0, ..., M_{r-1}, 2], axes=[k,...,k+r-1], and signal_size=[S_0,...,S_{r-1}].

Let D be an input tensor A, taking into account the signal_size, and, hence, D has the shape [B_0, ..., B_{k-1}, S_0, ..., S_{r-1}, 2].

Next, put

\[X[j_0,\dots,j_{k-1},j_k,\dots,j_{k+r-1}]=D[j_0,\dots,j_{k-1},j_k,\dots,j_{k+r-1},0]+iD[j_0,\dots,j_{k-1},j_k,\dots,j_{k+r-1},1]\]

for all indices j_0,...,j_{k+r-1}, where i is an imaginary unit, that is X is a complex tensor.

Then the discrete Fourier transform is the tensor \(Y\) of the same shape as the tensor \(X\), such that

\[Y[n_0,\dots,n_{k-1},m_0,\dots,m_{r-1}]=\sum\limits_{j_0=0}^{S_0-1}\cdots\sum\limits_{j_{r-1}=0}^{S_{r-1}-1}X[n_0,\dots,n_{k-1},j_0,\dots,j_{r-1}]\exp\left(-2\pi i\sum\limits_{q=0}^{r-1}\frac{m_qj_q}{S_q}\right)\]

for all indices n_0,...,n_{k-1}, m_0,...,m_{r-1}, and the result of the operation is the real tensor Z with the shape [B_0, ..., B_{k-1}, S_0, ..., S_{r-1}, 2] and such that

\[Z[n_0,\dots,n_{k-1},m_0,\dots,m_{r-1}, 0]=Re Y[n_0,\dots,n_{k-1},m_0,\dots,m_{r-1}],\]
\[Z[n_0,\dots,n_{k-1},m_0,\dots,m_{r-1}, 1]=Im Y[n_0,\dots,n_{k-1},m_0,\dots,m_{r-1}].\]

Calculations for the generic case of axes and signal sizes are similar.

Example:

There is no signal_size input (4D input tensor):

<layer ... type="DFT" ... >
    <input>
        <port id="0">
            <dim>1</dim>
            <dim>320</dim>
            <dim>320</dim>
            <dim>2</dim>
        </port>
        <port id="1">
            <dim>2</dim> <!-- axes input contains [1, 2] -->
        </port>
    <output>
        <port id="2">
            <dim>1</dim>
            <dim>320</dim>
            <dim>320</dim>
            <dim>2</dim>
        </port>
    </output>
</layer>

There is no signal_size input (3D input tensor):

<layer ... type="DFT" ... >
    <input>
        <port id="0">
            <dim>320</dim>
            <dim>320</dim>
            <dim>2</dim>
        </port>
        <port id="1">
            <dim>2</dim> <!-- axes input contains [0, 1] -->
        </port>
    <output>
        <port id="2">
            <dim>320</dim>
            <dim>320</dim>
            <dim>2</dim>
        </port>
    </output>
</layer>

There is signal_size input (4D input tensor):

<layer ... type="DFT" ... >
    <input>
        <port id="0">
            <dim>1</dim>
            <dim>320</dim>
            <dim>320</dim>
            <dim>2</dim>
        </port>
        <port id="1">
            <dim>2</dim> <!-- axes input contains [1, 2] -->
        </port>
        <port id="2">
            <dim>2</dim> <!-- signal_size input contains [512, 100] -->
        </port>
    <output>
        <port id="3">
            <dim>1</dim>
            <dim>512</dim>
            <dim>100</dim>
            <dim>2</dim>
        </port>
    </output>
</layer>

There is signal_size input (3D input tensor):

<layer ... type="DFT" ... >
    <input>
        <port id="0">
            <dim>320</dim>
            <dim>320</dim>
            <dim>2</dim>
        </port>
        <port id="1">
            <dim>2</dim> <!-- axes input contains [0, 1] -->
        </port>
        <port id="2">
            <dim>2</dim> <!-- signal_size input contains [512, 100] -->
        </port>
    <output>
        <port id="3">
            <dim>512</dim>
            <dim>100</dim>
            <dim>2</dim>
        </port>
    </output>
</layer>

There is signal_size input (5D input tensor, -1 in signal_size, unsorted axes):

<layer ... type="DFT" ... >
    <input>
        <port id="0">
            <dim>16</dim>
            <dim>768</dim>
            <dim>580</dim>
            <dim>320</dim>
            <dim>2</dim>
        </port>
        <port id="1">
            <dim>3</dim> <!-- axes input contains  [3, 1, 2] -->
        </port>
        <port id="2">
            <dim>3</dim> <!-- signal_size input contains [170, -1, 1024] -->
        </port>
    <output>
        <port id="3">
            <dim>16</dim>
            <dim>768</dim>
            <dim>1024</dim>
            <dim>170</dim>
            <dim>2</dim>
        </port>
    </output>
</layer>

There is signal_size input (5D input tensor, -1 in signal_size, unsorted axes, the second example):

<layer ... type="DFT" ... >
    <input>
        <port id="0">
            <dim>16</dim>
            <dim>768</dim>
            <dim>580</dim>
            <dim>320</dim>
            <dim>2</dim>
        </port>
        <port id="1">
            <dim>3</dim> <!-- axes input contains  [3, 0, 2] -->
        </port>
        <port id="2">
            <dim>3</dim> <!-- signal_size input contains [258, -1, 2056] -->
        </port>
    <output>
        <port id="3">
            <dim>16</dim>
            <dim>768</dim>
            <dim>2056</dim>
            <dim>258</dim>
            <dim>2</dim>
        </port>
    </output>
</layer>