Generate randomly distributed values
Sources
dspsrcs4
The Random Source block generates a frame of M values drawn from a uniform or Gaussian pseudorandom distribution, where you specify M in the Samples per frame parameter.
This reference page contains a detailed discussion of the following Random Source block topics:
When the Source type parameter is set to Uniform
,
the output samples are drawn from a uniform distribution whose minimum
and maximum values are specified by the Minimum and Maximum parameters,
respectively. All values in this range are equally likely to be selected.
A length-N vector specified for one or both of these parameters generates
an N-channel output (M-by-N matrix) containing a unique random distribution
in each channel.
For example, specify
Minimum = [ 0 0 -3 -3]
Maximum = [10 10 20
20]
to generate a four-channel output whose first and second columns contain random values in the range [0, 10], and whose third and fourth columns contain random values in the range [-3, 20]. When you specify only one of the Minimum and Maximum parameters as a vector, the block scalar expands the other parameter so it is the same length as the vector.
When the Source type parameter is set to Gaussian
,
you must also set the Method parameter, which
determines the method by which the block computes the output, and
has the following settings:
Ziggurat
— Produces
Gaussian random values by using the Ziggurat method.
Sum of uniform values
—
Produces Gaussian random values by adding and scaling uniformly distributed
random signals based on the central limit theorem. This theorem states
that the probability distribution of the sum of a sufficiently high
number of random variables approaches the Gaussian distribution. You
must set the Number of uniform values to sum parameter,
which determines the number of uniformly distributed random numbers
to sum to produce a single Gaussian random value.
For both settings of the Method parameter, the output samples are drawn from the normal distribution defined by the Mean and Variance parameters. A length-N vector specified for one or both of the Mean and Variance parameters generates an N-channel output (M-by-N frame matrix) containing a distinct random distribution in each column. When you specify only one of these parameters as a vector, the block scalar expands the other parameter so it is the same length as the vector.
The block's output can be either real or complex, as determined
by the Real
and Complex
options
in the Complexity parameter. These settings control
all channels of the output, so real and complex data cannot be combined
in the same output. For complex output with a Uniform
distribution,
the real and imaginary components in each channel are both drawn from
the same uniform random distribution, defined by the Minimum and Maximum parameters
for that channel.
For complex output with a Gaussian
distribution,
the real and imaginary components in each channel are drawn from normal
distributions with different means. In this case, the Mean parameter
for each channel should specify a complex value; the real component
of the Mean parameter specifies the mean of the
real components in the channel, while the imaginary component specifies
the mean of the imaginary components in the channel. When either the
real or imaginary component is omitted from the Mean parameter,
a default value of 0 is used for the mean of that component.
For example, a Mean parameter setting of [5+2i
0.5 3i]
generates a three-channel output with the following
means.
Channel 1 mean | real = 5 | imaginary = 2 |
Channel 2 mean | real = 0.5 | imaginary = 0 |
Channel 3 mean | real = 0 | imaginary = 3 |
For complex output, the Variance parameter, σ^{2}, specifies the total variance for each output channel. This is the sum of the variances of the real and imaginary components in that channel.
$${\sigma}^{2}={\sigma}_{\mathrm{Re}}^{2}+{\sigma}_{\mathrm{Im}}^{2}$$
The specified variance is equally divided between the real and imaginary components, so that
$$\begin{array}{c}{\sigma}_{\mathrm{Re}}^{2}=\frac{{\sigma}^{2}}{2}\\ {\sigma}_{\mathrm{Im}}^{2}=\frac{{\sigma}^{2}}{2}\end{array}$$
The Repeatability parameter determines whether or not the block outputs the same signal each time you run the simulation. You can set the parameter to one of the following options:
Repeatable
— Outputs
the same signal each time you run the simulation. The first time you
run the simulation, the block randomly selects an initial seed. The
block reuses these same initial seeds every time you rerun the simulation.
Specify seed
— Outputs
the same signal each time you run the simulation. Every time you run
the simulation, the block uses the initial seed(s) specified in the Initial
seed parameter. Also see Specifying the Initial Seed.
Not repeatable
—
Does not output the same signal each time you run the simulation.
Every time you run the simulation, the block randomly selects an initial
seed.
When you set the Repeatability parameter
to Specify seed
, you must set the Initial
seed parameter. The Initial seed parameter
specifies the initial seed for the pseudorandom number generator.
The generator produces an identical sequence of pseudorandom numbers
each time it is executed with a particular initial seed.
To specify the N initial seeds for an N-channel real-valued
output, Complexity parameter set to Real
,
provide one of the following in the Initial seed parameter:
Length-N vector of initial seeds — Uses each vector element as an initial seed for the corresponding channel in the N-channel output.
Single scalar — Uses the scalar to generate N random values, which it uses as the seeds for the N-channel output.
To specify the initial seeds for an N-channel complex-valued
output, Complexity parameter set to Complex
,
provide one of the following in the Initial seed parameter:
Length-N vector of initial seeds — Uses each vector element as an initial seed for generating N channels of real random values. The block uses pairs of adjacent values in each of these channels as the real and imaginary components of the final output, as illustrated in the following figure.
Single scalar — Uses the scalar to generate N random values, which it uses as the seeds for generating N channels of real random values. The block uses pairs of adjacent values in each of these channels as the real and imaginary components of the final output, as illustrated in the following figure.
The Sample time parameter value, T_{s},
specifies the random sequence sample period when the Sample
mode parameter is set to Discrete
.
In this mode, the block generates the number of samples specified
by the Samples per frame parameter value, M,
and outputs this frame with a period of M*T_{s}.
When Sample mode is set to Continuous
,
the block is configured for continuous-time operation, and the Sample
time and Samples per frame parameters
are disabled. Note that many DSP System Toolbox™ blocks
do not accept continuous-time inputs.
Only some of the parameters described below are visible in the dialog box at any one time.
The distribution from which to draw the random values, Uniform
or Gaussian
.
For more information, see Distribution Type.
The method by which the block computes the Gaussian random values, Ziggurat
or Sum
of uniform values
. This parameter is enabled when Source
type is set to Gaussian
. For
more information, see Distribution Type.
The minimum value in the uniform distribution. This parameter
is enabled when you select Uniform
from
the Source type parameter. Tunable in
Simulation mode only.
The maximum value in the uniform distribution. This parameter
is enabled when you select you select Uniform
from
the Source type parameter. Tunable in
Simulation mode only.
The number of uniformly distributed random values to sum to
compute a single number in a Gaussian random distribution. This parameter
is enabled when the Source type parameter is
set to Gaussian
, and the Method parameter
is set to Sum of uniform values
. For more
information, see Distribution Type.
The mean of the Gaussian (normal) distribution. This parameter
is enabled when you select Gaussian
from
the Source type parameter. Tunable in
Simulation mode only.
The variance of the Gaussian (normal) distribution. This parameter
is enabled when you select Gaussian
from
the Source type parameter. Tunable in
Simulation mode only.
The repeatability of the block output: Not repeatable
, Repeatable
,
or Specify seed
. In the Repeatable
and Specify
seed
settings, the block outputs the same signal every
time you run the simulation. For details, see Output Repeatability.
The initial seed(s) to use for the random number generator when
you set the Repeatability parameter to Specify
seed
. For details, see Specifying the Initial Seed. Tunable in
Simulation mode only.
When you select this check box, block inherits the sample mode, sample time, output data type, complexity, and signal dimensions of the signal from the downstream block. When you select this check box, the Sample mode, Sample time, Samples per frame, Output data type, and Complexity parameters are disabled.
Suppose you want to back propagate a 1-D vector. The output of the Random Source block is a 1-D vector of length M, where length M is inherited from the downstream block. When the Minimum, Maximum, Mean, or Variance parameter specifies N channels, the 1-D vector output contains M/N samples from each channel. An error occurs in this case when M is not an integer multiple of N.
Suppose you want to back propagate a M-by-N signal. When N>1, your signal has N channels. When N = 1, your signal has M channels. The value of the Minimum, Maximum, Mean, or Variance parameter can be a scalar or a vector of length equal to the number of channels. You can specify these parameters as either row or column vectors, except when the signal is a row vector. In this case, the Minimum, Maximum, Mean, or Variance parameter must also be specified as a row vector.
The sample mode, Continuous
or Discrete
.
This parameter is enabled when the Inherit output port attributes check
box is cleared.
The sample period, T_{s}, of the random output sequence. The output frame period is M*T_{s}. This parameter is enabled when the Inherit output port attributes check box is cleared.
The number of samples, M, in each output frame.
This parameter is enabled when the Inherit output port attributes check box is cleared.
The data type of the output, single-precision or double-precision. This parameter is enabled when the Inherit output port attributes check box is cleared.
The complexity of the output, Real
or Complex
.
This parameter is enabled when the Inherit output port attributes check
box is cleared.
Double-precision floating-point
Single-precision floating-point
Discrete Impulse | DSP System Toolbox |
Maximum | DSP System Toolbox |
Minimum | DSP System Toolbox |
Signal From Workspace | DSP System Toolbox |
Standard Deviation | DSP System Toolbox |
Variance | DSP System Toolbox |
Constant | Simulink |
Random Number | Simulink |
Signal Generator | Simulink |
rand | MATLAB |
randn | MATLAB |
RandStream | MATLAB |