Re: [LUG] Nature on supporting Open Source software

[Tom via list <list@xxxxxxxxxxxxx>]

On 03/07/2019 11:44, maceion@xxxxxxxxx wrote:

On 03/07/2019 08:00, Adrian Midgley wrote:

https://www.nature.com/articles/d41586-019-02046-0?utm_source=Nature+Briefing&utm_campaign=a651c61001-briefing-dy-20190702&utm_medium=email&utm_term=0_c9dfd39373-a651c61001-44185525


Adrian Midgley   http://defoam.net/  http://photo.defoam.net/

Thanks. Extremely interesting.
Such documentation analysis and thinking, if available on a 'analysis before action' brief many years ago (before any desktop computers- slide rule and 3 significant figures were the best available) might have allowed checking of a 'apparent' non reproducible fault in design calculations and saved some lives. Five of us in a party got 5 very different answers to the same input before calculation yet each calculation was 'correct arithmetically'.  The figures after the significant figures were very important if extended. Only using 12 figure logs did we approach the same answer.

-- 
Regards
Eion MacDonald

The article and some associated links are quite interesting.

As you state being correct arithmetically does not always work. Error propagation through calculations is always worth looking at - some ways of solving the same things can go massively wrong if you don't watch your error bars. Good unit testing is worth applying in most things!

Tom te tom te tom

-- 
The Mailing List for the Devon & Cornwall LUG
https://mailman.dclug.org.uk/listinfo/list
FAQ: http://www.dcglug.org.uk/listfaq