Hi,

Sage currently has a nice implementation of the Dickman rho function 
<https://sagemath.gitlab.io/documentation/html/en/reference/functions/sage/functions/transcendental.html>
 which 
is a solution to a specific differential delay equation (DDE).
In the context of ongoing work with integrals from sieve theory, I'm 
considering proposing two minor changes :

   1. allowing dickman_rho to return a rigorously proven RIF value (easy). 
   As it stands, it seems the precision is a constant multiple of abs_prec for 
   small values of the argument (for large values of the argument using 
   saddle-point method, the relative precision is O(1/x) but the constant is 
   non-explicit as far as I know, I don't intend to work this out)
   2. implementing the Buchstab function 
   <https://en.wikipedia.org/wiki/Buchstab_function>, which is another 
   solution to a differential delay equation, or more general solutions to DDE 
   of the type  x f'(x) + a f(x) + b f(x-1) = 0 using the same method as in 
   dickman_rho (Marsaglia-Zaman-Marsaglia).

Regarding 2. I'm not sure where this should belong: by default I'll propose 
a builtin "number theoretic function" with parameters a, b, in the same 
file as dickman_rho, but perhaps it should be part of a specific class, 
"DDESolution" ?

===
Sary

-- 
You received this message because you are subscribed to the Google Groups 
"sage-devel" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
To view this discussion visit 
https://groups.google.com/d/msgid/sage-devel/e906ac8f-2caa-4133-92f9-5b26b7b654abn%40googlegroups.com.

Reply via email to