模組:Complex Number/Functions

模块文档[查看] [编辑] [历史] [清除缓存]

本模組定義了一些可供Module:Complex_Number系列模組使用的擴充函數。

使用條件

要能使用此函數，必須先輸入一個數字類別資料結構以及其專用的math程式庫

• 數字類別資料結構必須包含下列成員函數：
  • clean() : 削去過小的小數，如5.0000000000000015會被削成5。
  • metatable必須實作下列函數

    加法、減法、乘法、除法、取餘數、相等判斷、轉字串、相反數（加法反元素）

• 專用的math程式庫必須包含下列成員：
  • 基本函數：

    abs（絕對值）、 sgn（符號函數）、 arg（輻角）、 conjugate（共軛）、 sqrt（平方根）、 dot（內積）

  • 高斯符號：

    floor（無條件捨去）、 ceil（無條件進位）、 round（四捨五入）

  • 除法相關：

    mod、 div（除法）、 inverse（乘法反元素）

  • 指對數相關：

    exp（自然指数）、 log（自然對數）、 pow（冪）

  • 特殊成員

    re（返回實部）、nonRealPart（返回不含實部的複數）、 tovector（返回複數的數組形式）

  • 三角函數

    sin、 cos、 tan、 cot、 cis、 asin、 acos、 atan、 acot

  • 雙曲函數

    sinh、 cosh、 tanh、 coth、 asinh、 acosh、 atanh、 acoth

  • 常數

    e、 pi（圓周率）、 nan（不是數字）

使用方法

1. 初始化任何符合此擴充函數庫使用條件的數學庫
   • local 自訂函數庫名稱 = require("Module:Complex Number").函數庫名稱.init()

     以Module:Complex Number的cmath為例：

     local cmath = require("Module:Complex Number").cmath.init()

2. 初始化本擴充函數庫
   • 自訂函數庫名稱 = require("Module:Complex Number/Functions")._init(自訂函數庫名稱, 函數庫對應的數字建構函式)

     以上述之Module:Complex Number的cmath為例：

     cmath = require("Module:Complex Number/Functions")._init(cmath, cmath.constructor)

3. 使用擴充函數庫中的函數

   例如：

   print(cmath.factorial(5), cmath.sec(cmath.pi/4))

   輸出：120 　 　1.4142135623731

模組中的函數

三角函數擴充

擴充了原本未定義的三角函數

如sec（正割）、 csc（餘割）、 sech（雙曲正割）、 csch（雙曲餘割）、 asec（反正割）、 acsc（反餘割）、asech（反雙曲正割）、 acsch（反雙曲餘割）、 gd（古德曼函數） 、 cogd（餘古德曼函數）、 arcgd（反古德曼函數）

功能

輸入一個複數x，回傳其指定三角函數的值

range(x,min,max)

功能

只取函數的某一段

若x位於min,max區間內，則回傳x，否則回傳NaN

統計函數

定義了一些統計函數

如minimum（最小值）、 maximum（最大值）、 average（平均值）、 geoaverage（幾何平均值）、 var（變異數）、 σ（標準差）

功能

輸入一系列數字，回傳其指定的統計值

diff(function, x₀)

功能

輸入一個函數，計算該函數在x=x₀的導數

實作方式

數值微分#高階方法

integral(a, b, function, step)

功能

輸入一個函數，計算從a到b的定積分，並以step為求黎曼和的間距

實作方式

en:Boole's_rule

limit(x₀, way, function)

功能

輸入一個函數，計算從way方向向x₀逼近的極限。

其中，way=1為右極限、way=-1為左極限、way=0為不分方向的極限，若左極不等於右極回傳NaN

條件式

常數條件輸入

if(條件, 為真時, 為假時)、ifelse(條件1, 條件1為真, 條件2, 條件2為真, ... ,皆為假)

代表條件在傳入函數時已經完成計算

函數條件輸入

iff(條件函數, 為真時, 為假時)、ifelsef(條件函數1, 條件1為真, 條件函數2, 條件2為真, ... ,皆為假)

代表條件在傳入函數時尚未計算，判斷的當下才計算。所傳入的函數需要是無參數函數，若有參數也只會被忽略。用於定義遞迴下的條件

factorial(x)

功能

輸入一個複數x，回傳其階乘值

即factorial(x)$=n!$

實作方式

參考#gamma(x)

binomial(n,k)

功能

計算二項式係數

也可以理解為從n個元素中取出k個元素的方法數

gcd(a,b,c,...)

功能

計算a,b,c,....等數字的最大公因數，支援複數。

實作方式

輾轉相除法

lcm(a,b,c,...)

功能

計算a,b,c,....等數字的最小公倍數，支援複數。

實作方式

最小公倍數#计算方法

gamma(x)

功能

輸入一個複數x，回傳其Γ函數值

伽瑪函數的實作方式

精確度

有效數字14位

運算效率

平均一次運算耗時約0.3582毫秒（3.6×10⁻⁴ s、一秒可計算2,700+次），測試於2018年11月19日 (一) 06:39 (UTC)、2022年4月12日 (二) 17:54 (UTC)。

實作方式

• 共分成4個部分
  • 中間藍色部分是利用從零展開倒數伽瑪函數的泰勒級數定義

    展開至前30項

    ${\frac {1}{\Gamma (z)}}=z+\gamma z^{2}+\left({\frac {\gamma ^{2}}{2}}-{\frac {\pi ^{2}}{12}}\right)z^{3}+\cdots$

    $a_{n}={\frac {a_{2}a_{n-1}-\sum _{j=2}^{n-1}(-1)^{j}\,\zeta (j)\,a_{n-j}}{n-1}}$

  • 兩側橘紅色部分是利用中間藍色代Γ函數的遞迴關係式定義，並用For迴圈實作

    $\Gamma (x+1)=x\Gamma (x)$

    $\Gamma (x-1)={\frac {\Gamma (x)}{x-1}}$

  • 上下的綠色部分則是使用Robert H. Windschitl (2002) 所提出的公式近似

    $\Gamma (z)\approx {\sqrt {\frac {2\pi }{z}}}\left({\frac {z}{e}}{\sqrt {z\sinh {\frac {1}{z}}+{\frac {1}{810z^{6}}}}}\right)^{z}$

  • 最後黃色部分則是使用帶有斯特靈級數的斯特靈公式近似

    $n!={\sqrt {2\pi n}}\left({n \over e}\right)^{n}\left(1+{1 \over 12n}+{1 \over 288n^{2}}-{139 \over 51840n^{3}}-{571 \over 2488320n^{4}}+\cdots \right).$

    展開至前16項 （來源：（OEIS數列A001163）、（OEIS數列A001164））

  • 而背景透明標記 （灰白相間） 部分則為超出浮點數可儲存範圍，會溢位或出現inf或nan
  • 最左邊土黃色則是可能出現低於設計的精確度小數12位而回傳0

參考文獻

上述文档嵌入自Module:Complex Number/Functions/doc。 (编辑 | 历史)
编者可以在本模块的沙盒 (编辑 | 差异)和测试样例 (创建)页面进行实验。
本模块的子页面。

local p = {}
local math_lib
local to_number
function p._init(_math_lib, _to_number)
	local warp_funcs={"factorial","gamma","sec","csc","sech","csch","asec","acsc","asech","acsch","gd","cogd","arcgd",
		"gcd","lcm","range","binomial",'minimum','maximum','average','geoaverage','var','σ',
		'selectlist','for','summation','product','if','iff','ifelse','ifelsef','diff','integral', '∫', 'limit',
		'hide','exprs','lastexpr',--調整條目中定義不顯示的函數
		'randomseed','time','nil','null','call',
		'divisorsigma','findnext','findlast','divisor'
	}
	for i=1,#warp_funcs do
		_math_lib[ warp_funcs[i] ] = p['_' .. warp_funcs[i] ]
	end
	math_lib = _math_lib
	to_number = _to_number
	return _math_lib
end
function p._complex_number()
	return p._init(require("Module:Complex Number").cmath.init(), require("Module:Complex Number").cmath.init().toComplexNumber)
end

local noop_func = function()end

function p._binomial(cal_n,cal_k)
	local r_n=tonumber(tostring(cal_n))
	local r_k=tonumber(tostring(cal_k))
	if r_n and r_k then
		if r_n>0 and r_k>=0  then
			local f_n, f_k;
			_,f_n = math.modf(r_n);_,f_k = math.modf(r_k)
			if math.abs(f_n) < 1e-12 and math.abs(f_k) < 1e-12 then
				local result = 1
				if r_n == 0 then return result end
				while r_k>0 do
					result = result * r_n / r_k
					r_n = r_n - 1
					r_k = r_k - 1
				end
				return result
			end
		end
	end
	local n=to_number(cal_n)
	local k=to_number(cal_k)
	return p._factorial(n) * math_lib.inverse( p._factorial(k) ) * math_lib.inverse( p._factorial(n-k) )
end

function p._factorial(cal_z)return p._gamma(to_number(cal_z) + 1)end
function p._sec(cal_z)return math_lib.inverse( math_lib.cos( to_number(cal_z) ) )end
function p._csc(cal_z)return math_lib.inverse( math_lib.sin( to_number(cal_z) ) )end
function p._sech(cal_z)return math_lib.inverse( math_lib.cosh( to_number(cal_z) ) )end
function p._csch(cal_z)return math_lib.inverse( math_lib.sinh( to_number(cal_z) ) )end
function p._asec(cal_z)return  math_lib.acos( math_lib.inverse( to_number(cal_z) ) )end
function p._acsc(cal_z)return  math_lib.asin( math_lib.inverse( to_number(cal_z) ) )end
function p._asech(cal_z)return  math_lib.acosh( math_lib.inverse( to_number(cal_z) ) )end
function p._acsch(cal_z)return  math_lib.asinh( math_lib.inverse( to_number(cal_z) ) )end
function p._gd(cal_z)return math_lib.atan( math_lib.tanh( to_number(cal_z) * 0.5 ) ) * 2 end
function p._arcgd(cal_z)return math_lib.atanh( math_lib.tan( to_number(cal_z) * 0.5 ) ) * 2 end
function p._cogd(cal_z)local x = to_number(cal_z); return - math_lib.sgn( x ) * math_lib.log( math_lib.abs( math_lib.tanh( x * 0.5 ) ) ) end

function p._range(val,vmin,vmax)
	local v_val, v_min, v_max = math_lib.tovector(to_number(val)), math_lib.tovector(to_number(vmin)), math_lib.tovector(to_number(vmax))
	for i=1,#v_val do
		local min_v, max_v = math.min(v_min[i],v_max[i]), math.max(v_min[i],v_max[i])
		if v_val[i] < min_v or v_val[i] > max_v then return to_number(math_lib.nan) end
	end
	return to_number(val)
end
function p._calc_diff(_value, _left, _right, _expr)
	local value = to_number(_value)
	local left = to_number(_left)
	local right = to_number(_right)
	local left_val = _expr(value + left)
	local right_val = _expr(value + right)
	return (left_val - right_val) / (left - right)
end
function p._diff_abramowitz_stegun(_x, _h, _f)
	local x = to_number(_x)
	local h = to_number(_h)
    local _1 = to_number(1) local _2 = to_number(2) local _8 = to_number(8) local _12 = to_number(12)
    --if tonumber(math_lib.abs(math_lib.nonRealPart(x))) > 1e-8 then
    --	--[[數值微分#複變的方法]] --(Martins, Sturdza et.al.)
    --	local _i = math_lib.i or _1
    --	return (math_lib.im(_f(x + _i * h))) / h
    --else
    	--[[數值微分#高階方法]] --(Abramowitz & Stegun, Table 25.2)
		return (- _f(x + _2*h) + _8*_f(x + h) - _8*_f(x - h) + _f(x - _2*h)) / (_12 * h)
	--end
end

function p._calc_from(_value, _left, _right, _expr)
	local value = to_number(_value)
	local left = to_number(_left)
	local left_val = _expr(value + left)
	return left_val - p._calc_diff(_value, _left, _right, _expr) * left
end
function p._get_sample_number(_value)
	local num = math_lib.re(math_lib.abs(to_number(_value)))
	return (tonumber(num)~=nil) and ((num < 1e-8) and 1 or num) or 1
end
function p._diff(_expr, _value)
	local func_type = type(noop_func)
	local value_str = tostring(_value)
	if value_str == 'nan' or value_str == '-nan'then
		error("計算失敗：無效的數值 ")
		return _expr
	end
	if value_str == 'inf' or value_str == '-inf'then
		error("計算失敗：不支援無窮微分 ")
		return _expr
	end
	local small_scale_pow = math_lib.floor(math_lib.log(p._get_sample_number(_value)) / math_lib.log(to_number(10)))-5
	local small_scale = math_lib.pow(10,small_scale_pow)
	if type(_expr) == func_type then
		return p._diff_abramowitz_stegun(_value, small_scale, _expr)
	else
		return 0
	end
end

function p._integral(_a,_b,func,step)
	local a = to_number(_a)
	local b = to_number(_b)
	local begin_str = tostring(_a)
	local stop_str = tostring(_b)
	if begin_str == 'nan' or begin_str == '-nan' or stop_str == 'nan' or stop_str == '-nan'then
		error("計算失敗：無效的迴圈 ")
		return func
	end
	if begin_str == 'inf' or begin_str == '-inf' or stop_str == 'inf' or stop_str == '-inf'then
		error("計算失敗：不支援無窮求積 ")
		return func
	end
	local n = tonumber(step) or 2000
	local f = (type(func)==type(noop_func))and func or (function()return to_number(func) or 0 end)
	local h = (b-a)/n
	local x0, xn = a, b
	local _2, _7, _12, _14, _32, _45 = 
		to_number(2), to_number(7), to_number(12), to_number(14), to_number(32), to_number(45)
	local i0, i1, i2 = 1, 2, 4
	local sumfxi0, sumfxi1, sumfxi2 = to_number(0), to_number(0), to_number(0)
	for i=1,n do -- Boole's rule
		if i0 > n-1 and i1 > n-2 and i2 > n-4 then break end
		local xi0, xi1, xi2 = a + i0 * h, a + i1 * h, a + i2 * h
		if i0 <= n-1 then sumfxi0 = sumfxi0 + f(xi0) end
		if i1 <= n-2 then sumfxi1 = sumfxi1 + f(xi1) end
		if i1 <= n-4 then sumfxi2 = sumfxi2 + f(xi2) end
		i0 = i0 + 2
		i1 = i1 + 4
		i2 = i2 + 4
	end
	return (_2 * h / _45) * (_7 * (f(x0) + f(xn)) + _32 * sumfxi0 + _12 * sumfxi1 + _14 * sumfxi2)
end
p['_∫'] = p._integral

function p._limit(_value, _way, _expr)
	local way = to_number(_way)
	local func_type = type(noop_func)
	local value_str = tostring(_value)
	if value_str == 'nan' or value_str == '-nan'then
		error("計算失敗：無效的數值 ")
		return _expr
	end
	if value_str == 'inf' or value_str == '-inf'then
		error("計算失敗：不支援無窮極限 ")
		return _expr
	end
	local small_scale_pow = tonumber(math_lib.re(math_lib.floor(math_lib.log(p._get_sample_number(_value)) / math_lib.log(to_number(10))))-6)
	local small_scale = math.pow(10,small_scale_pow)
	local small_scale_a = small_scale / 10
	local small_scale_c = small_scale * 10

	if type(_expr) == func_type then
		if math_lib.re(math_lib.abs(way)) < 1e-10 then
			local left = p._calc_from(_value, -small_scale, -small_scale_a, _expr)
			local right = p._calc_from(_value, small_scale_a, small_scale, _expr)
			if math_lib.re(math_lib.abs(left - right)) < small_scale_c then
				return (left + right) / 2
			else
				return math_lib.nan
			end
		else
			return (math_lib.re(way) > 0) and p._calc_from(_value, small_scale_a, small_scale, _expr) or
				p._calc_from(_value, -small_scale, -small_scale_a, _expr)
		end
	else
		return _expr
	end
end
p._nil = "nil"
p._null = "nil"
function p._if(_expr, _true_expr, _false_expr)
	return p._ifelse_func(false, _expr, _true_expr, _false_expr)
end
function p._iff(_expr, _true_expr, _false_expr)
	return p._ifelse_func(true, _expr, _true_expr, _false_expr)
end
function p._ifelse(...)
	return p._ifelse_func(false, ...)
end
function p._ifelsef(...)
	return p._ifelse_func(true, ...)
end
function p._ifelse_func(is_func, ...)
	local func = noop_func
	local exprlist = {...}
	local last_else = #exprlist % 2 == 1
	local max_num = (last_else and (#exprlist - 1) or #exprlist) / 2
	for i=1,max_num do
		local _expr = exprlist[i * 2 - 1]
		local expr = (type(_expr) == type(func)) and _expr() or _expr
		local expr_true = exprlist[i * 2]
		local _chk_flag = math_lib.abs(to_number(expr)) > 1e-14;
		if _chk_flag then 
			return (type(expr_true) == type(func) and is_func) and expr_true() or expr_true 
		end
	end
	if last_else then
		local expr_false = exprlist[#exprlist]
		return (type(expr_false) == type(func) and is_func) and expr_false() or expr_false 
	end
	local _expr = exprlist[1]
	return (type(_expr) == type(func)) and _expr() or _expr
end
function p._for(_start,_end,_step,_expr)
	local _begin = to_number(_start);
	local _stop = to_number(_end);
	local _do_step = to_number(_step);
	local check_loop = (_stop - _begin) / _do_step
	local begin_str = tostring(_begin)
	local stop_str = tostring(_stop)
	if math_lib.re(math_lib.abs(_do_step))<=1e-14 or math_lib.re(check_loop) < 0 or
		begin_str == 'nan' or begin_str == '-nan' or begin_str == 'inf' or begin_str == '-inf' or 
		stop_str == 'nan' or stop_str == '-nan' or stop_str == 'inf' or stop_str == '-inf'then 
			error("計算失敗：無效的迴圈 ")
			return _expr
	end
	if type(_expr) == type(noop_func) then
		local it = _begin
		local init = to_number(0)
		while math_lib.re(it) <= math_lib.re(_stop) do
			init = _expr(to_number(it),to_number(init))
			it = it + _do_step
		end
		return init
	else
		return _expr
	end
end
function p._summation(_start,_end,_expr)
	local _begin = to_number(_start);
	local _stop = to_number(_end);
	local _do_step = to_number(1);
	local check_loop = (_stop - _begin) / _do_step
	local begin_str = tostring(_begin)
	local stop_str = tostring(_stop)
	if math_lib.re(math_lib.abs(_do_step))<=1e-14 or math_lib.re(check_loop) < 0 or 
		begin_str == 'nan' or begin_str == '-nan' or stop_str == 'nan' or stop_str == '-nan'then
			error("計算失敗：無效的迴圈 ")
			return _expr
		end
	if begin_str == 'inf' or begin_str == '-inf' or stop_str == 'inf' or stop_str == '-inf'then
		error("計算失敗：不支援無窮求和 ")
		return _expr
	end
	local func_type = type(noop_func)
	local it = _begin
	local init = to_number(0)--空和
	while math_lib.re(it) <= math_lib.re(_stop) do
		init = init + ((type(_expr) == func_type) and _expr(to_number(it)) or to_number(_expr))--累加
		it = it + _do_step
	end
	return init
end

function p._product(_start,_end,_expr)
	local _begin = to_number(_start);
	local _stop = to_number(_end);
	local _do_step = to_number(1);
	local check_loop = (_stop - _begin) / _do_step
	local begin_str = tostring(_begin)
	local stop_str = tostring(_stop)
	if math_lib.re(math_lib.abs(_do_step))<=1e-14 or math_lib.re(check_loop) < 0 or 
		begin_str == 'nan' or begin_str == '-nan' or stop_str == 'nan' or stop_str == '-nan'then
			error("計算失敗：無效的迴圈 ")
			return _expr
	end
	if begin_str == 'inf' or begin_str == '-inf' or stop_str == 'inf' or stop_str == '-inf'then
		error("計算失敗：不支援無窮求積 ")
		return _expr
	end
	local func_type = type(noop_func)
	local it = _begin
	local init = to_number(1)--空積
	while math_lib.re(it) <= math_lib.re(_stop) do
		init = init * ((type(_expr) == func_type) and _expr(to_number(it)) or to_number(_expr))--累乘
		it = it + _do_step
	end
	return init
end

function p._randomseed(_seed)
	local seed = tonumber(tostring(_seed)) or (os.time() * os.clock())
	math.randomseed(math.floor(seed))
	return to_number(seed)
end

function p._time()
	return to_number(os.time())
end

function p._call(func, ...)
	if type(func) == type(noop_func) then return func(...)end
	return func
end

function p._hide(...)
	local input_args = {...}
	return to_number(input_args[#input_args])
end
p._exprs = p._hide
p._lastexpr = p._hide

function p._selectlist(x,...) 
	local input_args = {...}
	local y = input_args[1]
	local z = input_args[2]
	local id_x = tonumber(tostring(x)) or 0
	if type(y)==type({}) and #y >= id_x and id_x>0 then
		if id_x <= 0 then id_x = id_x + #y + 1 end
		return y[id_x] or tonumber('nan') 
	elseif type(z)==type({}) and #z >= id_x then
		local id_y = tonumber(tostring(y)) or 0
		if id_x <= 0 then id_x = id_x + #z + 1 end
		if type(z[id_x])==type({}) and #(z[id_x]) >= id_y and id_y>0 then
			if id_y <= 0 then id_y = id_y + #(z[id_x]) + 1 end
			return (z[id_x][id_y]) or tonumber('nan')
		end
	end
	id_x = tonumber(tostring(x)) or 0
	if id_x <= 0 then id_x = id_x + #input_args + 1 end
	return input_args[id_x] or tonumber('nan') 
end
function p._minimum(...) return p.minmax('min', ...) end
function p._maximum(...) return p.minmax('max', ...) end
function p._average(...) return p.minmax('avg', ...) end
function p._geoaverage(...) return p.minmax('gavg', ...) end
function p._var(...) return p.minmax('var', ...) end
p['_σ'] = function(...) return p.minmax('σ',...) end
function p.minmax(mode,...)
	local tonumber_lib = to_number or tonumber
	local lib_math = math_lib or math
	local args, tester = { ... }, {tonumber("nan")}
	local sum, prod, count, sumsq, sig = tonumber_lib(0), tonumber_lib(1), 0, tonumber_lib(0), (mode =='var'or mode=='σ')
	local mode_map = {}
	local non_nan
	for i=1,#args do
		local got_number, calc_number = tonumber(tostring(args[i])) or tonumber("nan"), tonumber_lib(args[i])
		if calc_number then sum, prod, count = calc_number + sum, calc_number * prod, count + 1 end
		if sig == true then 
			local x_2 = calc_number * calc_number
			if lib_math.dot then
				x_2 = lib_math.dot(calc_number, lib_math.conjugate(calc_number))
			end
			sumsq = sumsq + x_2 
		end
		mode_map[args[i]] = (mode_map[args[i]]or 0) + 1
		if tostring(got_number):lower()~="nan" and type(non_nan) == type(nil) then
			tester[1], non_nan = got_number, got_number
		else tester[#tester + 1] = got_number end
	end
	local modes={min=math.min,max=math.max,sum=function()return sum end,prod=function()return prod end,count=function()return count end,
		avg=function()return sum*tonumber_lib(1/count) end,
		gavg=function()return lib_math.pow(prod,tonumber_lib(1/count)) end,
		var=function()return sumsq*tonumber_lib(1/count)-sum*sum*tonumber_lib(1/(count*count)) end,
		['σ']=function()return lib_math.sqrt(sumsq*tonumber_lib(1/count)-sum*sum*tonumber_lib(1/(count*count))) end,
		mode=function()
			local max_count, mode_data = 0, ''
			for mkey, mval in pairs(mode_map) do
				if mval > max_count then
					max_count = mval
					mode_data = mkey
				end
			end
			return mode_data
		end
	}
	if tostring(tester[1]):lower()=="nan" and mode:sub(1,1)=='m' then
		local error_msg = ''
		for i=1,#args do if error_msg~=''then error_msg = error_msg .. '、 ' end
			error_msg = error_msg .. tostring(args[i])
		end
		error("給定的數字 " .. error_msg .." 無法比較大小") 
	end
	if type(modes[mode]) ~= type(tonumber) then
		error("未知的統計方式 '" .. mode .."' ") 
	end
	return modes[mode](unpack(tester))
end

local function fold(func, ...)
	-- Use a function on all supplied arguments, and return the result. The function must accept two numbers as parameters,
	-- and must return a number as an output. This number is then supplied as input to the next function call.
	local vals = {...}
	local count = #vals -- The number of valid arguments
	if count == 0 then return
		-- Exit if we have no valid args, otherwise removing the first arg would cause an error.
		nil, 0
	end 
	local ret = table.remove(vals, 1)
	for _, val in ipairs(vals) do
		ret = func(ret, val)
	end
	return ret, count
end

--[[
Fold arguments by selectively choosing values (func should return when to choose the current "dominant" value).
]]
local function binary_fold(func, ...)
	local value = fold((function(a, b) if func(a, b) then return a else return b end end), ...)
	return value
end
local Reciprocal_gamma_coeff = {1,0.577215664901532860607,-0.655878071520253881077,-0.0420026350340952355290,0.166538611382291489502,-0.0421977345555443367482,-0.00962197152787697356211,0.00721894324666309954240,-0.00116516759185906511211,-0.000215241674114950972816,0.000128050282388116186153,-0.0000201348547807882386557,-1.25049348214267065735e-6,1.13302723198169588237e-6,-2.05633841697760710345e-7,6.11609510448141581786e-9,5.00200764446922293006e-9,-1.18127457048702014459e-9,1.04342671169110051049e-10,7.78226343990507125405e-12,-3.69680561864220570819e-12,5.10037028745447597902e-13,-2.05832605356650678322e-14,-5.34812253942301798237e-15,1.22677862823826079016e-15,-1.18125930169745876951e-16,1.18669225475160033258e-18,1.41238065531803178156e-18,-2.29874568443537020659e-19,1.71440632192733743338e-20}
--https://oeis.org/A001163 、 https://oeis.org/A001164
local stirling_series_coeff = {1,0.0833333333333333333333333,0.00347222222222222222222222,-0.00268132716049382716049383,-0.000229472093621399176954733,0.000784039221720066627474035,0.0000697281375836585777429399,-0.000592166437353693882864836,-0.0000517179090826059219337058,0.000839498720672087279993358,0.0000720489541602001055908572,-0.00191443849856547752650090,-0.000162516262783915816898635,0.00640336283380806979482364,0.000540164767892604515180468,-0.0295278809456991205054407,-0.00248174360026499773091566,0.179540117061234856107699,0.0150561130400264244123842,-1.39180109326533748139915,-0.116546276599463200850734}
function p._gamma_high_imag(cal_z)
	local z = to_number(cal_z)
	if z ~= nil and math_lib.abs(math_lib.nonRealPart(z)) > 2 then
		local inv_z = math_lib.inverse(z)
		return math_lib.sqrt((math_lib.pi * 2) * inv_z) * math_lib.pow(z * math_lib.exp(-1) *
			math_lib.sqrt( (z * math_lib.sinh(inv_z) ) + math_lib.inverse(to_number(810) * z * z * z * z * z * z) ),z)
	end
	return nil
end
function p._gamma_morethen_lua_int(cal_z)
	local z = to_number(cal_z) - to_number(1)
	local lua_int_term = 18.1169 --FindRoot[ Factorial[ x ] == 2 ^ 53, {x, 20} ]
	if math_lib.abs(z) > (lua_int_term - 1) or (math_lib.re(z) < 0 and math_lib.abs(math_lib.nonRealPart(z)) > 1 ) then
		local sum = 1
		for i = 1, #stirling_series_coeff - 1 do
			local a, n = to_number(z), tonumber(i) local y, k, f = to_number(1), n, to_number(a)
			while k ~= 0 do 
				if k % 2 == 1 then y = y * f end 
				k = math.floor(k / 2); f = f * f
			end
			sum = sum + stirling_series_coeff[i + 1] * math_lib.inverse(y)
		end
		return math_lib.sqrt( (2 * math.pi) * z ) * math_lib.pow( z * math.exp(-1), z ) * sum
	end
	return nil
end
function p._gamma_abs_less1(cal_z)
	local z = to_number(cal_z)
	if (math.abs(math_lib.re(z)) <=1.001) then
		if math_lib.abs(math_lib.nonRealPart(z)) < 1e-14 and ( (math.abs(math_lib.re(z) - 1) < 1e-14) or (math.abs(math_lib.re(z) - 2) < 1e-14) ) then return to_number(1)end
		return math_lib.inverse(p._recigamma_abs_less1(z))
	end
	return nil
end
function p._recigamma_abs_less1(z)
	local result = to_number(0)
	for i=1,#Reciprocal_gamma_coeff do
		result = result + Reciprocal_gamma_coeff[i] * math_lib.pow(z,i)
	end
	return result
end
function p._gamma(cal_z)
	local z = to_number(cal_z)
	if math_lib.abs(math_lib.nonRealPart(z)) < 1e-14 and ((math_lib.re(z) < 0 or math.abs(math_lib.re(z)) < 1e-14)
		and math.abs(math.floor(math_lib.re(z)) - math_lib.re(z)) < 1e-14 ) then return tonumber("nan") end
	local pre_result = p._gamma_morethen_lua_int(z) or p._gamma_high_imag(z) or p._gamma_abs_less1(z)
	if pre_result then return pre_result end
	local real_check = math_lib.re(z)
	local loop_count = math.floor(real_check)
	local start_number, zero_flag = z - loop_count, false
	if math_lib.abs(start_number) <= 1e-14 then start_number = to_number(1);zero_flag = true end
	local result = math_lib.inverse(p._recigamma_abs_less1(start_number))
	if math_lib.abs(math_lib.nonRealPart(z)) < 1e-14 and ((math_lib.re(z) > 1e-14 )and math.abs(math.floor(math_lib.re(z)) - math_lib.re(z)) < 1e-14 ) then result = to_number(1)  end
	local j = to_number(start_number)
	for i=1,math.abs(loop_count) do
		if loop_count > 0 then result = result * j else result = result * math_lib.inverse(j-1) end
		if zero_flag==true and loop_count > 0 then zero_flag=false else if loop_count > 0 then j = j + 1 else j = j - 1 end end
	end
	if math_lib.abs(math_lib.nonRealPart(z)) < 1e-14 and ((math_lib.re(z) > 1e-14 )and math.abs(math.floor(math_lib.re(z)) - math_lib.re(z)) < 1e-14 ) then return math_lib.floor(result) end
	return result
end

local function findGcd(a, b)
	local r, oldr = to_number(b), to_number(a)
	while math_lib.abs(r) > 1e-6 do local mod_val = oldr % r oldr, r = to_number(r), mod_val end
	if math_lib.abs(math_lib.nonRealPart(oldr)) < 1e-14 and (math_lib.re(oldr) < 0 ) then oldr = -oldr end
	return oldr
end

function p._gcd(...)
	local result, count = fold(findGcd, ...)
	return result
end

function p._lcm(...)
	local function findLcm(_a, _b)
		local a, b = to_number(_b), to_number(_a)
		return math_lib.abs(a * b) / findGcd(a, b)
	end
	local result, count = fold(findLcm, ...)
	return result
end

function p._findnext(func, x)
	local it = to_number(x) + 1
	if type(func) ~= type(noop_func) then
		if math_lib.abs(to_number(func)) < 1e-14 then
			return to_number("inf")
		else
			return it
		end
	end
	local checker = func(it)
	while math_lib.abs(to_number(checker)) < 1e-14 do
		it = it + 1
		checker = func(it)
	end
	return it
end

function p._findlast(func, x)
	local it = to_number(x) - 1
	if type(func) ~= type(noop_func) then
		if math_lib.abs(to_number(func)) < 1e-14 then
			return to_number("-inf")
		else
			return it
		end
	end
	local checker = func(it)
	while math_lib.abs(to_number(checker)) < 1e-14 do
		it = it - 1
		checker = func(it)
	end
	return it
end

local function get_divisor_combination(n)
	local lib_factor = require('Module:Factorization')
	local is_complex = math_lib.abs(math_lib.nonRealPart(n)) > 1e-14
	local factors
	if math_lib.abs(math_lib.floor(n)-n) < 1e-14 then
		factors = (lib_factor[is_complex and "_gaussianFactorization" or "_factorization"])(
			is_complex and n or tonumber(tostring(n))
		)
	else factors = {[n]=1} end
	local gened=require('Module:Combination').getCombinationGenerator()
	gened:init(factors,0)
	return gened:findSubset()
end
function p._divisor(_n, _x)
	local x = to_number(_x)
	local n = to_number(_n)
	if math_lib.abs(n) < 1e-14 then return 0 end
	local combination = get_divisor_combination(n)
	local divisors = {}
	for i=1,#combination do
		local divisor = to_number(1)
		for j=1,#(combination[i]) do
			divisor = divisor * to_number(combination[i][j])
		end
		divisors[#divisors+1] = divisor
	end
	table.sort(divisors, function(a,b) return math_lib.abs(a) < math_lib.abs(b) end)
	return divisors[math_lib.abs(x)] or 0
end

function p._divisorsigma(_x, _n)
	local x = to_number(1), n
	if _n == nil then 
		n = to_number(_x)
	else
		x = to_number(_x)
		n = to_number(_n)
	end
	if math_lib.abs(n) < 1e-14 then return 0 end

	local is_complex = math_lib.abs(math_lib.nonRealPart(n)) > 1e-14
	if not is_complex then n = math_lib.abs(n) end
	local combination = get_divisor_combination(n)
	local sum = to_number(0)
	for i=1,#combination do
		local divisor = to_number(1)
		for j=1,#(combination[i]) do
			divisor = divisor * to_number(combination[i][j])
		end
		sum = sum + math_lib.pow(divisor, x)
	end
	return sum
end

return p