TOPICS
Search

Dickman Function


DickmanFunction

The probability that a random integer between 1 and x will have its greatest prime factor <=x^alpha approaches a limiting value F(alpha) as x->infty, where F(alpha)=1 for alpha>1 and F(alpha) is defined through the integral equation

 F(alpha)=int_0^alphaF(t/(1-t))(dt)/t
(1)

for 0<=alpha<=1 (Dickman 1930, Knuth 1998), which is almost (but not quite) a homogeneous Volterra integral equation of the second kind. The function can be given analytically for 1/2<=alpha<=1 by

F(alpha)=1-int_alpha^1F(t/(1-t))(dt)/t
(2)
=1-int_alpha^1(dt)/t
(3)
=1+lnalpha
(4)

(Knuth 1998).

Amazingly, the average value of x such that p=n^x is

mu=lim_(n->infty)<x>
(5)
=lim_(n->infty)<(lnp)/(lnn)>
(6)
=int_0^1x(dF)/(dx)dx
(7)
=int_0^1F(t/(1-t))dt
(8)
=0.62432999,
(9)

which is precisely the Golomb-Dickman constant lambda, which is defined in a completely different way!

DickmanFunctionRho

The Dickman function can be solved numerically by converting it to a delay differential equation. This can be done by noting that t/(1-t) will become (1-t)/t=1/t-1 upon multiplicative inversion, so define rho(alpha)=F(1/alpha) to obtain

 rho(1/alpha)=int_0^alpharho(1/t-1)(dt)/t.
(10)

Now change variables under the integral sign by defining

t^'=1/t
(11)
dt^'=-(dt)/(t^2),
(12)

so

 (dt)/t=-(t^')/(dt^').
(13)

Plugging back in gives

 rho(1/alpha)=-int_infty^(1/alpha)rho(t^'-1)(dt^')/(t^').
(14)

To get rid of the 1/alphas, define u=1/alpha, so

 rho(u)=-int_infty^urho(t^'-1)(dt^')/(t^').
(15)

But by the first fundamental theorem of calculus,

 d/(dx)int_a^xf(x^')dx^'=f(x),
(16)

so differentiating both sides of equation (15) gives

 rho^'(u)=-(rho(u-1))/u.
(17)

This holds for 0<alpha<1, which corresponds to u>1. Rearranging and combining with an appropriate statement of the condition F(alpha)=1 for alpha>1 in the new variables then gives

 {rho(u)=1   for 0<=u<=1; urho^'(u)+rho(u-1)=0   for u>1.
(18)

The second-largest prime factor will be <=x^beta is given by an expression similar to that for F(alpha). It is denoted G(beta), where G(beta)=1 for beta>=1/2 and

 G(beta)=int_0^beta[G(t/(1-t))-F(t/(1-t))](dt)/t
(19)

for 0<=beta<=1/2.


See also

Buchstab Function, Golomb-Dickman Constant, Greatest Prime Factor, Prime Factor

Explore with Wolfram|Alpha

References

Dickman, K. "On the Frequency of Numbers Containing Prime Factors of a Certain Relative Magnitude." Arkiv för Mat., Astron. och Fys. 22A, 1-14, 1930.Knuth, D. E. The Art of Computer Programming, Vol. 2: Seminumerical Algorithms, 3rd ed. Reading, MA: Addison-Wesley, pp. 382-384, 1998.Norton, K. K. Numbers with Small Prime Factors, and the Least kth Power Non-Residue. Providence, RI: Amer. Math. Soc., 1971.Panario, D. "Smallest Components in Combinatorial Structures." Feb. 16, 1998. http://algo.inria.fr/seminars/sem97-98/panario.pdf.Ramaswami, V. "On the Number of Positive Integers Less than x and Free of Prime Divisors Greater than x^c." Bull. Amer. Math. Soc. 55, 1122-1127, 1949.Ramaswami, V. "The Number of Positive Integers <=X and Free of Prime Divisors >x^C, and a Problem of S. S. Pillai." Duke Math. J. 16, 99-109, 1949.

Referenced on Wolfram|Alpha

Dickman Function

Cite this as:

Weisstein, Eric W. "Dickman Function." From MathWorld--A Wolfram Web Resource. https://mathworld.wolfram.com/DickmanFunction.html

Subject classifications