[Cs5500] Reg: Clarifications on MMG playground

[Karl Lieberherr]

Hi Ravi:

the new purpose of the playground is that the students learn a bit
about numerical algorithms:

http://www.google.ch/url?sa=t&rct=j&q=floating+point+arithmetic+ACM+surveys&source=web&cd=6&ved=0CEkQFjAF&url=http%3A%2F%2Fciteseer.ist.psu.edu%2Fviewdoc%2Fdownload%3Fdoi%3D10.1.1.102.244%26rep%3Drep1%26type%3Dpdf&ei=-m7IToKYM-zF0AGHsvn0Dw&usg=AFQjCNFIP4OiauCFvvGxWXA_5bi3xuA18w

http://dl.acm.org/citation.cfm?id=103163

We want them to get close to the golden ratio (sqrt(5)-1)/2 but in
such a way that the playground can reliably detect whether they are
above or below.

Ideally we would like them to hit the first floating point number
below the golden ratio on the machine we use. Does the interface:
http://download.oracle.com/javase/1.4.2/docs/api/java/lang/Double.html
allow us to check this?

-- Karl

On Fri, Nov 18, 2011 at 6:16 PM, Ravishankar Rajagopal
<rajagopal.r at husky.neu.edu> wrote:
> Hello Professor,
>
> Karan and Madhu let me know today that the precision of the values of x,c
> and y for MMG would be limited to three digits. However based on our
> discussion, i had the impression that students would get more points the
> greater their precision is. Can you please let me know if there is a
> conflict in understanding here and which is right? Further they also let me
> know that students were complaining on some issues based on the wrong
> version of MMG.
>
> So I was planning to send a single mail asking students to pick up the
> latest version from sourceforge as well as clarify on the precision of the
> values of c,x and y.
>
> Further if it were true that the precision was 3, then the next question
> that would come up from students is what if they straight away start at the
> optimum value. Can you please clarify that as well?
>
> Regards,
> Ravi Rajagopal
>