|Re: [eigen] Matrix2i mean|
[ Thread Index |
| More lists.tuxfamily.org/eigen Archives
- To: eigen@xxxxxxxxxxxxxxxxxxx
- Subject: Re: [eigen] Matrix2i mean
- From: Julian Kent <jkflying@xxxxxxxxx>
- Date: Wed, 6 Nov 2019 19:04:56 +0100
- Dkim-signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=20161025; h=mime-version:references:in-reply-to:from:date:message-id:subject:to; bh=NgcRWLo/nWBTQBELTaBEyDZO8enuISLAUPL2gsogaLA=; b=HxkvxjWtv7OCFfAaMQQOaBAJ8uZLjvvoZlAaaTXuOsfsm3Mbr1UOXHmG/tWqa1KvwS lju1JqTRa1x3+EuKLZPgkejJMNlTuvEhbfBjEnH+C5IoiyAkArhSlpRatK0YbzzU0l1N LKr3yobMSF2/YZOIY0zHBmJsqnffT1ZXEXaNYSgQmhuB1ESVU4h/wYb3meqq8T84RzMr sIEDd7a852ob/U4fagDEc+MV6tTI7UPzK3QJRTFNdt0temhzqJKr/dEabBi/zlXQcRdU lVFZU5t/UhxywmflgyWSJ7Wb0vonmc4nO6APjF9GgrVdP/XFD9O18UNcuxbLCxNDtdgD 0sqQ==
If it is about floating point accuracy rather than integer overflow, I've had excellent results using Kahan Summation. It gives bounded error over any problem size, and fixed a bunch of problems for me which needed accurate dot products.
Perhaps it would make sense to have an accurate_mean, accurate_sum and accurate_dot which fix the various problems that integers and floats/doubles have for accumulation style problems?