| 摘要 |
<P>PROBLEM TO BE SOLVED: To provide an error function arithmetic unit and method thereof that can speed up an operation of error function with a configuration suitable for integration at low cost. <P>SOLUTION: The error function arithmetic unit, which performs processing for finding an error function F<SB>M</SB>(T) where variables are T (positive real numbers) and degrees are M (M are integers ≥0), includes a coefficient table 10 and arithmetic section 150. The coefficient table 10 stores a value of F<SB>M+i</SB>(T<SB>0</SB>) /F<SB>M</SB>(T<SB>0</SB>) (i is at least one integer among 1 to n) found with respect to a predetermined T<SB>0</SB>or a value of constant times thereof, which are obtained by n-th polynomial approximation (n is a natural number) of the error function F<SB>M</SB>(T). The arithmetic section 150 reads out the value of F<SB>M+i</SB>(T<SB>0</SB>)/F<SB>M</SB>(T<SB>0</SB>) or that of constant times from the coefficient table 10, and at least part of processing for finding the value of the error function F<SB>M</SB>(T) is performed by the fixed point arithmetic using the value of F<SB>M+i</SB>(T<SB>0</SB>)/F<SB>M</SB>(T<SB>0</SB>) or that of constant times. <P>COPYRIGHT: (C)2005,JPO&NCIPI |
