I tried to code the Salamin-Brent iteration. I calculated with exact numbers and the procedure returned the numerical value with the .n() function. The iteration was very very slow, for 20 iterations it is approximately 900 secs. (In Maple the same code is 0.06 secs). Can I modify the numerical precision in every step with some method? (Because in every step the number of correct digits is doubling )
Revision history [back]
 | 1 | initial version |
computing the digist of Pi
| | 2 | retagged |
computing the digist of Pi
I tried to code the Salamin-Brent iteration. I calculated with exact numbers and the procedure returned the numerical value with the .n() function. The iteration was very very slow, for 20 iterations it is approximately 900 secs. (In Maple the same code is 0.06 secs). Can I modify the numerical precision in every step with some method? (Because in every step the number of correct digits is doubling )
