By memorizing a table of logarithms base 10, one can multiply by using addition. A log table could be just as easy (or hard) to memorize as a standard digit multiplication table, but be much more useful because the information is more concentrated in a range where it matters (values 1 through about 3). How big a table should one memorize? How many significant digits for each log? How should one round? I don't have a rigorous way of answering these questions. We try to keep the relative error low, but also keep entropy low. Here is a simple table, also giving the boundaries between each log. The plus or minus suffix gives the direction of the exact boundary, so a number at the given boundary should be rounded the opposite direction.
0 = | log | 1 |
1.1+ | ||
0.1 = | log | 1.26 |
1.4+ | ||
0.2 = | log | 1.6 |
1.8- | ||
0.3 = | log | 2 |
2.2+ | ||
0.4 = | log | 2.5 |
2.8+ | ||
0.5 = | log | 3.2 |
3.5+ | ||
0.6 = | log | 4 |
4.5- | ||
0.7 = | log | 5 |
5.6+ | ||
0.8 = | log | 6.3 |
7+ | ||
0.9 = | log | 8 |
9- | ||
1 = | log | 10 |
For example 2.8 < pi < 3.5, so log(pi)=0.5 ; 2.2 < e < 2.8, so log(e)=0.4 ; log (pi*e) = 0.4 + 0.5 = 0.9 = log (8), or pi*e = 8. Actual value 8.539734223 .
No comments :
Post a Comment