vmmath.java

linux下建立JAVA虚拟机的源码KAFFE

   * <li>If the first argument is negative zero and the second argument is a
   * negative finite odd integer, or the first argument is negative infinity
   * and the second argument is a positive finite odd integer, then the result
   * is negative infinity.</li>
   * <li>If the first argument is less than zero and the second argument is a
   * finite even integer, then the result is equal to the result of raising
   * the absolute value of the first argument to the power of the second
   * argument.</li>
   * <li>If the first argument is less than zero and the second argument is a
   * finite odd integer, then the result is equal to the negative of the
   * result of raising the absolute value of the first argument to the power
   * of the second argument.</li>
   * <li>If the first argument is finite and less than zero and the second
   * argument is finite and not an integer, then the result is NaN.</li>
   * <li>If both arguments are integers, then the result is exactly equal to
   * the mathematical result of raising the first argument to the power of
   * the second argument if that result can in fact be represented exactly as
   * a double value.</li>
   *
   * </ul><p>(In the foregoing descriptions, a floating-point value is
   * considered to be an integer if and only if it is a fixed point of the
   * method {@link #ceil(double)} or, equivalently, a fixed point of the
   * method {@link #floor(double)}. A value is a fixed point of a one-argument
   * method if and only if the result of applying the method to the value is
   * equal to the value.) This is accurate within 1 ulp, and is semi-monotonic.
   *
   * @param a the number to raise
   * @param b the power to raise it to
   * @return a<sup>b</sup>
   */
  public static native double pow(double a, double b);

  /**
   * Get the IEEE 754 floating point remainder on two numbers. This is the
   * value of <code>x - y * <em>n</em></code>, where <em>n</em> is the closest
   * double to <code>x / y</code> (ties go to the even n); for a zero
   * remainder, the sign is that of <code>x</code>. If either argument is NaN,
   * the first argument is infinite, or the second argument is zero, the result
   * is NaN; if x is finite but y is infinite, the result is x. This is
   * accurate within the limits of doubles.
   *
   * @param x the dividend (the top half)
   * @param y the divisor (the bottom half)
   * @return the IEEE 754-defined floating point remainder of x/y
   * @see #rint(double)
   */
  public static native double IEEEremainder(double x, double y);

  /**
   * Take the nearest integer that is that is greater than or equal to the
   * argument. If the argument is NaN, infinite, or zero, the result is the
   * same; if the argument is between -1 and 0, the result is negative zero.
   * Note that <code>Math.ceil(x) == -Math.floor(-x)</code>.
   *
   * @param a the value to act upon
   * @return the nearest integer &gt;= <code>a</code>
   */
  public static native double ceil(double a);

  /**
   * Take the nearest integer that is that is less than or equal to the
   * argument. If the argument is NaN, infinite, or zero, the result is the
   * same. Note that <code>Math.ceil(x) == -Math.floor(-x)</code>.
   *
   * @param a the value to act upon
   * @return the nearest integer &lt;= <code>a</code>
   */
  public static native double floor(double a);

  /**
   * Take the nearest integer to the argument.  If it is exactly between
   * two integers, the even integer is taken. If the argument is NaN,
   * infinite, or zero, the result is the same.
   *
   * @param a the value to act upon
   * @return the nearest integer to <code>a</code>
   */
  public static native double rint(double a);

  /**
   * <p>
   * Take a cube root. If the argument is NaN, an infinity or zero, then
   * the original value is returned.  The returned result must be within 1 ulp
   * of the exact result.  For a finite value, <code>x</code>, the cube root
   * of <code>-x</code> is equal to the negation of the cube root
   * of <code>x</code>. 
   * </p>
   * <p>
   * For a square root, use <code>sqrt</code>.  For other roots, use
   * <code>pow(a, 1 / rootNumber)</code>.
   * </p>
   *
   * @param a the numeric argument
   * @return the cube root of the argument
   * @see #sqrt(double)
   * @see #pow(double, double)
   */
  public static native double cbrt(double a);

  /**
   * <p>
   * Returns the hyperbolic cosine of the given value.  For a value,
   * <code>x</code>, the hyperbolic cosine is <code>(e<sup>x</sup> + 
   * e<sup>-x</sup>)/2</code>
   * with <code>e</code> being <a href="#E">Euler's number</a>.  The returned
   * result must be within 2.5 ulps of the exact result.
   * </p>
   * <p>
   * If the supplied value is <code>NaN</code>, then the original value is
   * returned.  For either infinity, positive infinity is returned.
   * The hyperbolic cosine of zero must be 1.0.
   * </p>
   * 
   * @param a the numeric argument
   * @return the hyperbolic cosine of <code>a</code>.
   * @since 1.5
   */
  public static native double cosh(double a);

  /**
   * <p>
   * Returns <code>e<sup>a</sup> - 1.  For values close to 0, the
   * result of <code>expm1(a) + 1</code> tend to be much closer to the
   * exact result than simply <code>exp(x)</code>.  The result must be within
   * 1 ulp of the exact result, and results must be semi-monotonic.  For finite
   * inputs, the returned value must be greater than or equal to -1.0.  Once
   * a result enters within half a ulp of this limit, the limit is returned.
   * </p>   
   * <p>
   * For <code>NaN</code>, positive infinity and zero, the original value
   * is returned.  Negative infinity returns a result of -1.0 (the limit).
   * </p>
   * 
   * @param a the numeric argument
   * @return <code>e<sup>a</sup> - 1</code>
   * @since 1.5
   */
  public static native double expm1(double a);

  /**
   * <p>
   * Returns the hypotenuse, <code>a<sup>2</sup> + b<sup>2</sup></code>,
   * without intermediate overflow or underflow.  The returned result must be
   * within 1 ulp of the exact result.  If one parameter is held constant,
   * then the result in the other parameter must be semi-monotonic.
   * </p>
   * <p>
   * If either of the arguments is an infinity, then the returned result
   * is positive infinity.  Otherwise, if either argument is <code>NaN</code>,
   * then <code>NaN</code> is returned.
   * </p>
   * 
   * @param a the first parameter.
   * @param b the second parameter.
   * @return the hypotenuse matching the supplied parameters.
   * @since 1.5
   */
  public static native double hypot(double a, double b);

  /**
   * <p>
   * Returns the base 10 logarithm of the supplied value.  The returned
   * result must within 1 ulp of the exact result, and the results must be
   * semi-monotonic.
   * </p>
   * <p>
   * Arguments of either <code>NaN</code> or less than zero return
   * <code>NaN</code>.  An argument of positive infinity returns positive
   * infinity.  Negative infinity is returned if either positive or negative
   * zero is supplied.  Where the argument is the result of
   * <code>10<sup>n</sup</code>, then <code>n</code> is returned.
   * </p>
   *
   * @param a the numeric argument.
   * @return the base 10 logarithm of <code>a</code>.
   * @since 1.5
   */
  public static native double log10(double a);

  /**
   * <p>
   * Returns the natural logarithm resulting from the sum of the argument,
   * <code>a</code> and 1.  For values close to 0, the
   * result of <code>log1p(a)</code> tend to be much closer to the
   * exact result than simply <code>log(1.0+a)</code>.  The returned
   * result must be within 1 ulp of the exact result, and the results must be
   * semi-monotonic.
   * </p>
   * <p>
   * Arguments of either <code>NaN</code> or less than -1 return
   * <code>NaN</code>.  An argument of positive infinity or zero
   * returns the original argument.  Negative infinity is returned from an
   * argument of -1.
   * </p>
   *
   * @param a the numeric argument.
   * @return the natural logarithm of <code>a</code> + 1.
   * @since 1.5
   */
  public static native double log1p(double a);

  /**
   * <p>
   * Returns the hyperbolic sine of the given value.  For a value,
   * <code>x</code>, the hyperbolic sine is <code>(e<sup>x</sup> - 
   * e<sup>-x</sup>)/2</code>
   * with <code>e</code> being <a href="#E">Euler's number</a>.  The returned
   * result must be within 2.5 ulps of the exact result.
   * </p>
   * <p>
   * If the supplied value is <code>NaN</code>, an infinity or a zero, then the
   * original value is returned.
   * </p>
   * 
   * @param a the numeric argument
   * @return the hyperbolic sine of <code>a</code>.
   * @since 1.5
   */
  public static native double sinh(double a);

  /**
   * <p>
   * Returns the hyperbolic tangent of the given value.  For a value,
   * <code>x</code>, the hyperbolic tangent is <code>(e<sup>x</sup> - 
   * e<sup>-x</sup>)/(e<sup>x</sup> + e<sup>-x</sup>)</code>
   * (i.e. <code>sinh(a)/cosh(a)</code>)
   * with <code>e</code> being <a href="#E">Euler's number</a>.  The returned
   * result must be within 2.5 ulps of the exact result.  The absolute value
   * of the exact result is always less than 1.  Computed results are thus
   * less than or equal to 1 for finite arguments, with results within
   * half a ulp of either positive or negative 1 returning the appropriate
   * limit value (i.e. as if the argument was an infinity).
   * </p>
   * <p>
   * If the supplied value is <code>NaN</code> or zero, then the original
   * value is returned.  Positive infinity returns +1.0 and negative infinity
   * returns -1.0.
   * </p>
   * 
   * @param a the numeric argument
   * @return the hyperbolic tangent of <code>a</code>.
   * @since 1.5
   */
  public static native double tanh(double a);

}