🚀 go-pugleaf

In article <2341@njitgw.njit.edu> jfa0522@hertz.njit.edu (john f andrews ece) writes:
>In some article steve@nuchat.sccsi.com (Steve Nuchia) writes:
>BTW, can anyone recall the discussion on applying similar interpolation 
>to a sampled EKG some time ago? I believe someone wanted to reconstruct a
>complete EKG waveform from sampled data, and was actually testing these 
>digital techniques and reporting results here?  Perhaps I am incorrect but
>I beleive many of these ideas were tossed about then ...

This past summer I was working as a research engineer's assistant at a 
pacemaker company (which shall remain nameless) and we were converting a 
digitized EKG from one sample rate to another.  The original was sampled at
1000 samples/sec and the new digital scope needed to have it's rates as powers
of 2.  So we did a direct linear interpolation on the 1000 sample/sec tape
and reconstructed an EKG at 1024 samples/sec.  It was a trivial programming 
job once the algorithm was worked out properly.  Four people with engineering
PhDs came to me for math help  :)  (I will graduate with a BS in math in may.)

There was only one unusual result.  We converted 1000 to 1024 and then back to
1000 and did an error check.  At the nth power of 2 there were approximately
2n terms in error, evenly distributed around the power of 2 itself.  This was
not a particularly large error (10^-16) but the error doubled with every
power as well.

Perhaps someone here could explain that one for me. The engineers simply said
it was negligible and ignored it.

*******************************************************************************
Joshua C. Sasmor -- famous mathematician-to-be  *      jcs@lab4.math.ufl.edu
************************************************* brassman@maple.circa.ufl.edu
"The highest form of pure thought is in         * x9999bis@maple.circa.ufl.edu
 mathematics" -- Plato                          *
*******************************************************************************