Matrix storage for system identification

A computer program product for executing a method for performing parametric system identification on the basis of a least-squares method, the computer program product comprising a computer readable storage medium having computer readable program code embodied therewith, which, when executed, causes a computer to execute the steps of:

• preparing measurement values, observation models, and parameters;
• creating ID mapping in which the measurement values and values of the parameters are associated with the parameters;
• and
• storing the parameters in a matrix to calculate the parameters by using the least-squares method, wherein the matrix is divided into small matrices,
• in a case where a target small matrix among the small matrices is a diagonal small matrix, a component of an N×N sparse matrix is generated at a position of offset M from a principal diagonal and block sparse row (BSR) or diagonal (DIA) is selected if the ID mapping exists, and an N×N dense matrix is selected if the ID mapping does not exist, N being the number of parameters defined by values of the parameters, and M being a value of a parameter index,
• in a case where the target small matrix is not a diagonal small matrix and a term decided without the ID mapping exists, a dense matrix is stored, and
• in a case where the target small matrix is not a diagonal small matrix and a term decided without the ID mapping does not exist,
• if the same ID mapping is referred to, a block exists on the principal diagonal and BSR or DIA is selected, and
• if the same ID mapping is not referred to, elements are stored by using ELLPACK/ITPACK (ELL) if the number of parameters of a left element is larger than the number of parameters of a right element, and elements are stored by using compressed sparse row (CSR) if the number of parameters of the left element is not larger than the number of parameters of the right element.

A system for performing parametric system identification on the basis of a least-squares method through a process performed by a computer, the system comprising:

• storage means;
• measurement values, observation models, and data of parameters stored in the storage means;
• data of ID mapping stored in the storage means, the ID mapping associating the measurement values and values of the parameters with the parameters;
• and
• means for storing the parameters in a matrix to calculate the parameters by using the least-squares method, wherein the matrix is divided into small matrices,
• in a case where a target small matrix among the small matrices is a diagonal small matrix, a component of an N×N sparse matrix is generated at a position of offset M from a principal diagonal and block sparse row (BSR) or diagonal (DIA) is selected if the ID mapping exists, and an N×N dense matrix is selected if the ID mapping does not exist, N being the number of parameters defined by values of the parameters, and M being a value of a parameter index,
• in a case where the target small matrix is not a diagonal small matrix and a twin decided without the ID mapping exists, a dense matrix is stored, and
• in a case where the target small matrix is not a diagonal small matrix and a term decided without the ID mapping does not exist,
• if the same ID mapping is referred to, a block exists on the principal diagonal and BSR or DIA is selected, and
• if the same ID mapping is not referred to, elements are stored by using ELLPACK/ITPACK (ELL) if the number of parameters of a left element is larger than the number of parameters of a right element, and elements are stored by using compressed sparse row (CSR) if the number of parameters of the left element is not larger than the number of parameters of the right element.