power series method pade(m,n) off by 1?

[Daniel Friedan]

For p a power series in x, p.pade(m,n) requires p to be given up through and including O(x^(m+n+1)).

Shouldn't this be O(x^(m+n)) ?

The Padé approximant  P_m(x)/Q_n(x) contains m+n+1 unknown coefficients.  So p(x) should contain m+n+1 terms, x^0 ... x^(m+n).

Or am I missing something elementary?

Daniel

[Emmanuel Charpentier]

See my answer to the ask.sagemath question, whose the present question is a duplicate…

​