props.txt

particle filter 粒子滤波器 matlab工具箱

Help on properties of the linear time-variant expression class (pfltv).

'expression'
The expression supplied by the user.

'str'
A string representing the expression.

'exprsize'
The size of the expression, [m n], where m is the number of rows and n is the number of columns
of the generated output. A 0 means that the particular size couldn't be determined and has to be set manually, if
this property is needed (which isn't the case most of the time).

'gradx'
The gradient of the expression with respect to x.

'gradw'
The gradient of the expression with respect to w.

'xvars'
A cell array containing the names of the states.
Not used in this class and just included for consistency.

'uvars'
A cell array containing the name of the elements forming u(t).
Not used in this class and just included for consistency.

'wvars'
A cell array containing the name of the elements forming w(t).
Not used in this class and just included for consistency.

'islinear'
Is true for linear expressions and false otherwise.
Since this class is linear, this property is set to 'true'.

'evalvar'
Some data objects, such as pfltv that use matrix multiplication, don't support
xvars, uvars and wvars. These objects need to know what variable(s) to use in the
evaluation. 'evalvar' contains this information. The variables x, t, w and u
are represented by the numbers 1, 2, 3 and 4 respectively.
evalvar=1 means that x will be used in the evaluation.
evalvar=[1 3] means that [x; u] will be used (x and u are, like always,
row vectors or scalars).
Example - the eval command of pfltv ('expression' is a matrix):
evalvar=1 returns expression*x
evalvar=[1 4 3] returns expression*[x; w; u]
Note that pfltv doesn't support empty evalvar vectors. For that
kind of functionallity, see pfarray.

'varsize'
Only needed when evaluating multiple variables, ie when 'evalvar' is a vector.
When differentiating an expression based on multiple variables, we need to know
the size of each variable (column vector) in order to extract the right columns of the
matrix. 'varsize' must be a vector of 4 elements representing the size of
[x t u w]. Not that if a particular variable isn't evaluated (ie its index is
not present in the evalvar vector), its size does not matter and can be set to 0.
Example: x=[1 2 3]', w=[1 2]', u=3 and evalvar=[1 4 3], the 'varsize' argument
must be set to [3 x 2 1], where x can be anything (preferably 1, since t is always
a scalar).
The gradient with respect to w will be a 6x6 matrix where column 4 and 5 will be
extracted from the expression matrix, and the remaining columns will contain zeros.

'description'
A short description of the object.