Documentation errors

[denalepa1@…]

In Version 1.55.0 DOCUMENTATION there are two "typographical" errors. MATH/CONSTANTS documentation, the Table of math constants contains two errors:

1. The value of pi_minus_three is listed as 1.41593

it should be 0.141593

-=-=- -=-=-

2. The constant pow23_four_minus_pi is represented as 4^(2/3) - π

It should be (4-π)^3/2

2a. Note that 2/3 is NOT 3/2.

I assume that the constant is badly named and its name is inconsistent with its actual value. This is a deeper problem that the creators may wish to address.

[Steven Watanabe]

I think math is the right component, but a link to the page containing the problem would be helpful.

[John Maddock]

Paul, can I get you to check these out? Thanks!

[Paul A. Bristow]

Sorry for delay in response. Thanks for reporting this mistake, and not really just in the documentation :-(

Do you have a view of which constant is useful. I can't find where the constant called pow23_four_minus_pi is used.

It clearly should be the correct name/value combination.

Is either or both useful - especially to you?

pow23_four_minus_pi

or

pow32_four_minus_pi

(or both)

[John Maddock]

A quick grep shows it was used in the Rayleigh distribution at one point, but is commented out. Maybe we should just remove it - I doubt it has much general use?

[John Maddock]

On second thoughts, maybe

template <class RealType, class Policy>
inline RealType kurtosis(const rayleigh_distribution<RealType, Policy>& /*dist*/)

Should really use the constant? albeit we need to check both the name and the value to see which is correct ;)

[Paul A. Bristow]

The constant pow23_four_minus_pi does not seem generally useful and is not necessary for Rayleigh skewness or kurtosis.

So the constant has been removed and an explicit calculation used for the kurtosis excess(and skewness). This agrees with the constant calculated using NTL previously.