| 摘要 |
A method is used for optimizing a parameter of an processing unit. The method is for providing an initial input F<SUB>i=0 </SUB>to be a value of the parameter, and detecting a corresponding initial output O(F<SUB>i=0</SUB>), determining step value DeltaF, providing a first input F<SUB>0</SUB>-DeltaF and a second input F<SUB>0</SUB>+DeltaF, detecting a first output O(F<SUB>0</SUB>-DeltaF) and a second output O(F<SUB>0</SUB>+DeltaF), determining whether the first output O(F<SUB>0</SUB>-DeltaF) is greater than the second output O(F<SUB>0</SUB>+DeltaF), determining a first seeking direction, providing other inputs F<SUB>i</SUB>, F<SUB>i+1</SUB>, F<SUB>i+2</SUB>, . . . F<SUB>i+n-1</SUB>, and then detecting corresponding other outputs O(F<SUB>i</SUB>), O(F<SUB>i+1</SUB>), O(F<SUB>i+2</SUB>), . . . O(F<SUB>i+n-1</SUB>), wherein n is a natural number, determining a stop point F<SUB>i</SUB>, stopping seeking along the first seeking direction at the stop point, identifying an extremum output, and identifying an input to be the optimized value of the parameter. An information storage apparatus for optimizing a parameter is also disclosed.
|
