NMSIS-DSP  Version 1.2.0
NMSIS DSP Software Library
Matrix Functions

This set of functions provides basic matrix math operations. The functions operate on matrix data structures. For example, the type definition for the floating-point matrix structure is shown below: More...

Modules

 Householder transform of a vector
 Computes the Householder transform of a vector x.
 
 Matrix Addition
 Adds two matrices.
 
 Cholesky and LDLT decompositions
 Computes the Cholesky or LL^t decomposition of a matrix.
 
 Complex Matrix Multiplication
 Complex Matrix multiplication is only defined if the number of columns of the first matrix equals the number of rows of the second matrix. Multiplying an M x N matrix with an N x P matrix results in an M x P matrix.
 
 Complex Matrix Transpose
 Tranposes a complex matrix.
 
 Matrix Initialization
 Initializes the underlying matrix data structure. The functions set the numRows, numCols, and pData fields of the matrix data structure.
 
 Matrix Inverse
 Computes the inverse of a matrix.
 
 Matrix Multiplication
 Multiplies two matrices.
 
 QR decomposition of a Matrix
 Computes the QR decomposition of a matrix M using Householder algorithm.
 
 Matrix Scale
 Multiplies a matrix by a scalar. This is accomplished by multiplying each element in the matrix by the scalar. For example:
 
 Matrix Subtraction
 Subtract two matrices.
 
 Matrix Transpose
 Tranposes a matrix.
 
 Matrix Vector Multiplication
 Multiplies a matrix and a vector.
 

Detailed Description

This set of functions provides basic matrix math operations. The functions operate on matrix data structures. For example, the type definition for the floating-point matrix structure is shown below:

    typedef struct
    {
      uint16_t numRows;     // number of rows of the matrix.
      uint16_t numCols;     // number of columns of the matrix.
      float32_t *pData;     // points to the data of the matrix.
    } riscv_matrix_instance_f32;

There are similar definitions for Q15 and Q31 data types.

The structure specifies the size of the matrix and then points to an array of data. The array is of size numRows X numCols and the values are arranged in row order. That is, the matrix element (i, j) is stored at:

    pData[i*numCols + j]
Init Functions
There is an associated initialization function for each type of matrix data structure. The initialization function sets the values of the internal structure fields. Refer to riscv_mat_init_f32(), riscv_mat_init_q31() and riscv_mat_init_q15() for floating-point, Q31 and Q15 types, respectively.
Use of the initialization function is optional. However, if initialization function is used then the instance structure cannot be placed into a const data section. To place the instance structure in a const data section, manually initialize the data structure. For example:
riscv_matrix_instance_f32 S = {nRows, nColumns, pData};
riscv_matrix_instance_q31 S = {nRows, nColumns, pData};
riscv_matrix_instance_q15 S = {nRows, nColumns, pData};
where nRows specifies the number of rows, nColumns specifies the number of columns, and pData points to the data array.
Size Checking
By default all of the matrix functions perform size checking on the input and output matrices. For example, the matrix addition function verifies that the two input matrices and the output matrix all have the same number of rows and columns. If the size check fails the functions return:
    RISCV_MATH_SIZE_MISMATCH
Otherwise the functions return
    RISCV_MATH_SUCCESS
There is some overhead associated with this matrix size checking. The matrix size checking is enabled via the #define
    RISCV_MATH_MATRIX_CHECK
within the library project settings. By default this macro is defined and size checking is enabled. By changing the project settings and undefining this macro size checking is eliminated and the functions run a bit faster. With size checking disabled the functions always return RISCV_MATH_SUCCESS.