Version 0.20

ADDED

• Added the remove(n,k) and bremove(n,k) methods, for removing all occurrences of the factor k from integer n. (acf1237)

IMPROVEMENTS

• Re-implemented the mod method based on the identity: x % y = x - y*floor(x/y), which always returns the exact result, assuming that x and y are rational numbers. (a622167)

FIXES

• Fixed a special case in modpow(x, y, z) when y < 0 and gcd(x, z) != 1. (f624550)
• Fixed a minor issue in broot(x, y) when y is not an integer. (f635807)
• Fixed a minor issue in bfpow(x, y) when y is a non-numeric scalar. (023b784)

---

All changes: 0.19...0.20

Version 0.17

ADDED

• isqrtrem(n) : the integer square root and the remainder.
• irootrem(n) : the integer nth root and the remainder.
• eta(n) : the Dirichlet eta function at n.
• beta(x,y) : the beta function (for x > 0 and y > 0).
• lambert_w(x) : the Lambert W function (for x >= -1/e)
• rad2deg(n) : converts radians to degrees
• deg2rad(n) : converts degrees to radians
• bessel_j(x,n) : the first order Bessel function: J_n(x).
• bessel_y(x,n) : the second order Bessel function: Y_n(x).

FIXES

• Fixed the accuracy in the method bernreal() for small n.
• Fixed a special case in the method irand() when the self value is 0.
  • irand(0) = 0

OTHER

• Minor optimization in the decimal expansion of non-integers.

Version 0.16

ADDED

• Added two methods for computing the nth-Harmonic number:
  • harmfrac(n)
  • harmreal(n)
• Added two methods for reseeding the pseudorandom number generator:
  • seed(n)
  • iseed(n)

IMPROVEMENTS

• Extended the method digits() to accept an optional base.
  • digits(4095, 16) #=> ('f', 'f', 'f')

BUG FIXES

• Using threads in some parts of the program, will no longer cause a segmentation fault.

However, Math::BigNum is not thread-safe, therefore using Math::BigNum objects inside threads, still causes a segmentation fault.

Version 0.15

ADDED

• Unimport support.

     no Math::BigNum;     # disables the `:constant` behavior in the current scope

• Added the new_int() and new_uint() methods for setting signed and unsigned native integers.

PERFORMANCE IMPROVEMENTS

• Noticeable optimizations in method is_pow().
  • Thanks to Dana Jacobsen - @danaj.
• Minor optimization in bernfrac() for values of n < 50.
• Faster algorithm in bernfrac() for values of n >= 50. (based on the Zeta function)
• Faster algorithm in bernreal(), using positive values of the Zeta function. (before, negative values were used)

BUG FIXES

• When a base is specified, special values, such "Inf" and "NaN", are no longer parsed specially.

     Math::BigNum->new('inf', 36) == 24171;
     Math::BigNum->new('nan', 36) == 30191;

OTHER

• Documentation improvements.

---

More changes: 0.14...0.15

Version 0.14

• Fixed a minor bug inside the method n->is_pow(p) when n is negative (for old versions of GMP).

Version 0.13

• Added the bernfrac method, powered by an improved version of Seidel's algorithm.
• Stricter validation of the precision specified as an import value.
• Minor bug-fix in the method is_pow for is_pow(-1, odd_n) -- it now returns true.
• Fixed some special cases in the method iroot.
• Fixed a special case in the method valuation when the second argument is 0.

Version 0.12

• Decreased the default floating-point precision from 256 bits, to 200 bits.
• Added support for specifying the default precision as an import value.

use Math::BigNum PREC => 1024;

• Fixed a minor-bug in the method is_pow for some special cases.
• Minor POD fixes.

Version 0.11

• Added the $n->parts method, which returns the numerator and the denominator as BigNum objects.

• Support for rational exponentiation with an integer power.

  (3/5)**3 == 27/125
  

• Added the $x->li method for computing the logarithm integral of $x.

• Added the $n->bernreal method, which returns the nth-Bernoulli number as a floating-point value.

• Added the $n->kronecker($k) method, which returns the Kronecker symbol (n|k).

• Added the following exportable functions:

  factorial(n)       # product of first n integers: n!
  primorial(n)       # product of primes <= n
  binomial(n,k)      # binomial coefficient
  fibonacci(n)       # nth-Fibonacci number
  lucas(n)           # nth-Lucas number
  ipow(a,k)          # integer exponentiation: int(a^k)
  

• Added the method $n->popcount, which returns the number of 1s in the binary representation of $n.

• Added the :all option to export anything that is exportable (functions + constants).

• Added the $n->valuation($k) method, which returns the number of times $n is divisible by $k.

• Added the $x->bfloat method, which truncates $x in-place to a floating-point value.

• Added the method $n->is_pow($k), which returns true iff a^k = n for some positive a.

• Added the method $x->lgrt, which returns the logarithmic-root of $x.

  100->lgrt        # solves for x in `x**x = 100` and returns: `3.59728...`
  

• Added the following floating-point methods:

  (b)fadd
  (b)fsub
  (b)fmul
  (b)fdiv
  (b)fmod
  (b)fpow
  

Improvements

• Extended the method float to accept an optional argument with the number of bits of precision.
• Optimized the $k->root($n) method for positive integer values of $n.
• Increased the default accuracy value for the method is_prime from 12 to 20.
• Increased the default precision of floating-point numbers from 128 to 256 bits.

Changes

• Changed the return values of the sign method from ("-", "", "+") to (-1, 0, 1).
• Some minor bug-fixes and documentation improvements.

Version 0.10

What's new

• Faster creation of MPFR objects from scalar integer values (aa3ee61).
• Conversion of scalar values to MPFR objects is done more strictly, by validating each conversion.
• Added the missing [b|i]mul methods in the Nan class (1b5792c).
• Minor overall performance improvements, by taking advantage of some commutative properties.
• Added the bln() method, which calculates the natural logarithm in-place (71bef20).
• Stricter definitions for the log and blog methods when one of the arguments is +/-Infinity (7140567).
• Require Math::GMPz >= 0.39 (for Rmpz_bin_si) (4de9f99).
• Use Math::GMPq's (>=0.41) _ulong_max() and _long_min() to determine the integer limits (4e7856b).

Version 0.09

What's new

• Better support for scalar values. (c2359d5)
• Better performance for idiv() and bidiv() when called with negative scalar integers.
• Faster creation of Math::BigNum objects with scalar integer values.
• Scalar values passed to (b)i* methods are now validated or converted correspondingly.