6.75 asympt ignores order on first use (10.1.00)

6.75.1 John S Robinson

Here is a quirk discovered by a colleague:

asympt does not seem to honour the Order setting the first time it is called for a particular expression:

 
> restart: R:=sqrt(r^2-a^2); 
 
                                     2    2 
                          R := sqrt(r  - a )
 

Series expansion for large r converted to an algebraic expression. Note that Order is ignored.

I think this is well described as a ’quirk’; Maple seems to need to do one series expansion of the expression before it starts to honour the value of Order. If you skip/execute the a0 line below then the a1 line ignores/honours the Order option. If you cange the - in the a0 line to +, it works differently, so it seems to be something to do with remembering the series for that particular expression.

> a0:=asympt(1/sqrt(x^2-a^2),x); 
 
                          2          4           6 
                         a          a           a        1 
        a0 := 1/x + 1/2 ---- + 3/8 ---- + 5/16 ---- + O(----) 
                          3          5           7        9 
                         x          x           x        x 
 
> a1:=asympt(1/R,r,4); 
 
                                      2 
                                     a        1 
                    a1 := 1/r + 1/2 ---- + O(----) 
                                      3        5 
                                     r        r 
 
> a2:=asympt(1/R,r,4); 
 
                                      2 
                                     a        1 
                    a2 := 1/r + 1/2 ---- + O(----) 
                                      3        5 
                                     r        r 
 
> a3:=asympt(1/R,r); 
 
                                2          4 
                               a          a        1 
              a3 := 1/r + 1/2 ---- + 3/8 ---- + O(----) 
                                3          5        7 
                               r          r        r
 

Without executing the a0 line we get:

> restart: R:=sqrt(r^2-a^2); 
> a1:=asympt(1/R,r,4); 
 
                                2          4 
                               a          a        1 
              a1 := 1/r + 1/2 ---- + 3/8 ---- + O(----) 
                                3          5        7 
                               r          r        r 
 
> a2:=asympt(1/R,r,4); 
 
                                      2 
                                     a        1 
                    a2 := 1/r + 1/2 ---- + O(----) 
                                      3        5 
                                     r        r 
 
> a3:=asympt(1/R,r); 
 
                          2          4           6 
                         a          a           a        1 
        a3 := 1/r + 1/2 ---- + 3/8 ---- + 5/16 ---- + O(----) 
                          3          5           7        9 
                         r          r           r        r
 

This is not a great problem, but I am intrigued as to why it happens, and if there is a better anser than ’do it twice’

6.75.2 Robert Israel(18.1.00)

This is actually a bug in "series" (which "asympt" calls). I don’t have any advice other than to call it twice. The reason it works the second time, I think, is that "series" consults its remember table to see if it has computed the same series before to this or higher order; if it has done so, it truncates the series appropriately.