--- /dev/null
+// expt().
+
+// General includes.
+#include "cl_sysdep.h"
+
+// Specification.
+#include "cl_real.h"
+
+
+// Implementation.
+
+#include "cl_R.h"
+#include "cl_rational.h"
+#include "cl_integer.h"
+#include "cl_I.h"
+
+// Methode:
+// Für y>0:
+// a:=x, b:=y.
+// Solange b gerade, setze a:=a*a, b:=b/2. [a^b bleibt invariant, = x^y.]
+// c:=a.
+// Solange b:=floor(b/2) >0 ist,
+// setze a:=a*a, und falls b ungerade, setze c:=a*c.
+// Ergebnis c.
+// Für y=0: Ergebnis 1.
+// Für y<0: (/ (expt x (- y))).
+
+// Assume y>0.
+inline const cl_R expt_pos (const cl_R& x, const cl_I& y)
+{
+ if (rationalp(x)) {
+ DeclareType(cl_RA,x);
+ return expt(x,y); // x rational -> schnellere Routine
+ } else {
+ DeclareType(cl_F,x);
+ var cl_F a = x;
+ var cl_I b = y;
+ while (!oddp(b)) { a = square(a); b = b >> 1; }
+ var cl_F c = a;
+ until (eq(b,1))
+ { b = b >> 1;
+ a = square(a);
+ if (oddp(b)) { c = a * c; }
+ }
+ return c;
+ }
+}
+
+const cl_R expt (const cl_R& x, const cl_I& y)
+{
+ if (eq(y,0)) { return 1; } // y=0 -> Ergebnis 1
+ var cl_boolean y_negative = minusp(y);
+ var cl_I abs_y = (y_negative ? -y : y); // Betrag von y nehmen
+ var cl_R z = expt_pos(x,abs_y); // (expt x (abs y))
+ return (y_negative ? recip(z) : z); // evtl. noch Kehrwert nehmen
+}