[Issue 95167] XIRR() function gives Err:502

[bu...@apache.org]

https://bz.apache.org/ooo/show_bug.cgi?id=95167

--- Comment #13 from Ariel Constenla-Haile <ar...@apache.org> ---
(In reply to Driss from comment #12)
> In the function Code the Newton's Method was used. The convergence of this
> Method depends amongst others on the start value fResultRate, which is the
> third argument of XIRR (guessed rate with default Value= 0,1). In case of
> divergence one get the err502. Changing the guessed rate can help to get
> convergence and avoid err502.
> 
> I will try to improve the code. If you have any idea or hints how to avoid
> divergence of Newton's method it would be great.
> 
> May be the third argument for XIRR  should not be given by the user but
> calculated in the code in a manner that divergence will be avoid as far as
> possible?

While searching for a solution try to keep compatibility with the current use;
if the function is defined in the standard, follow its specification; try to
see what MS Office does, for example
https://support.office.com/en-nz/article/XIRR-function-c3cdd4d8-359c-482a-a8ba-7a0052f5c053

Regina is on Cc, sure she can give you a better advice.

-- 
You are receiving this mail because:
You are on the CC list for the issue.