Previous: Bit-Counting, Up: Example Definitions [Contents][Index]
A somewhat limited sine function could be defined as follows, using the well-known Taylor series expansion for ‘sin(x)’:
(defmath mysin ((float (anglep x))) (interactive 1 "mysn") (setq x (to-radians x)) ; Convert from current angular mode. (let ((sum x) ; Initial term of Taylor expansion of sin. newsum (nfact 1) ; "nfact" equals "n" factorial at all times. (xnegsqr :"-(x^2)")) ; "xnegsqr" equals -x^2. (for ((n 3 100 2)) ; Upper limit of 100 is a good precaution. (working "mysin" sum) ; Display "Working" message, if enabled. (setq nfact (* nfact (1- n) n) x (* x xnegsqr) newsum (+ sum (/ x nfact))) (if (~= newsum sum) ; If newsum is "nearly equal to" sum, (break)) ; then we are done. (setq sum newsum)) sum))
The actual sin function in Calc works by first reducing the problem
to a sine or cosine of a nonnegative number less than ‘pi/4’. This
ensures that the Taylor series will converge quickly. Also, the calculation
is carried out with two extra digits of precision to guard against cumulative
round-off in ‘sum’. Finally, complex arguments are allowed and handled
by a separate algorithm.
(defmath mysin ((float (scalarp x))) (interactive 1 "mysn") (setq x (to-radians x)) ; Convert from current angular mode. (with-extra-prec 2 ; Evaluate with extra precision. (cond ((complexp x) (mysin-complex x)) ((< x 0) (- (mysin-raw (- x))) ; Always call mysin-raw with x >= 0. (t (mysin-raw x)))))) (defmath mysin-raw (x) (cond ((>= x 7) (mysin-raw (% x (two-pi)))) ; Now x < 7. ((> x (pi-over-2)) (- (mysin-raw (- x (pi))))) ; Now -pi/2 <= x <= pi/2. ((> x (pi-over-4)) (mycos-raw (- x (pi-over-2)))) ; Now -pi/2 <= x <= pi/4. ((< x (- (pi-over-4))) (- (mycos-raw (+ x (pi-over-2))))) ; Now -pi/4 <= x <= pi/4, (t (mysin-series x)))) ; so the series will be efficient.
where mysin-complex is an appropriate function to handle complex
numbers, mysin-series is the routine to compute the sine Taylor
series as before, and mycos-raw is a function analogous to
mysin-raw for cosines.
The strategy is to ensure that ‘x’ is nonnegative before calling
mysin-raw. This function then recursively reduces its argument
to a suitable range, namely, plus-or-minus ‘pi/4’. Note that each
test, and particularly the first comparison against 7, is designed so
that small roundoff errors cannot produce an infinite loop. (Suppose
we compared with ‘(two-pi)’ instead; if due to roundoff problems
the modulo operator ever returned ‘(two-pi)’ exactly, an infinite
recursion could result!) We use modulo only for arguments that will
clearly get reduced, knowing that the next rule will catch any reductions
that this rule misses.
If a program is being written for general use, it is important to code it carefully as shown in this second example. For quick-and-dirty programs, when you know that your own use of the sine function will never encounter a large argument, a simpler program like the first one shown is fine.
Previous: Bit-Counting, Up: Example Definitions [Contents][Index]