'rp'

   Running Performance Predictor

   Given several running performances, this program will predict
   performance at distances other than those given. The model used is
   that average speed 'v' depends on distance 's' as follows:

         v = v0 - v1 * log(s)

   where v0 and v1 are constants

             v0 ~ top speed

             v1 = speed loss per order of magnitude of s

   If you would plot your average speed versus distance on semi-log
   paper, the points would fall on a straight line (for a given
   level of exertion: e.g., "all out" or "training"). The level of
   fit is surprisingly good for casual as well as world-class runners.
   It is easy to extrapolate or interpolate to distances of interest
   once the best-fit line is found.

   Least-squares analysis is done on the input data, which consist of
   a file of 'events' where an event is a distance-time pair (e.g.
   1 mile  4:00). There must be at least two events at different
   distances to define the line! Multiple events at a given distance are
   OK. Also, the more widely the distances are separated, the more reliable
   the results will be. A one-mile time and a 10k time work well; a
   five-mile and 10k don't.

   With more than two events, the
   correlation score (0-100) gives a measure
   of the consistency of the data.

   Note that the constants are only relatively constant: more interval
   work, say, would increase v0, and better aerobic conditioning might
   reduce v1. Data spanning months may be inconsistent if fitness
   is changing rapidly.


   rp takes three files as input:

      distance units file: default 'dunits'
      prediction distances file: default 'dlist'
      event file: specified on command line

   The function and format of each file is described below:

      dunits - This file contains pairs of the alphanumeric
               representation and value (in meters) of the distance
               units that rp knows about. E.g.,

                     meters       1
                     kilometers   1000
                     miles        1609.4
                     marathon     42195
                     laps         400
                     hops         0.5
                     skips        1.2

      dlist - This file contains a list of the distances that you
              want rp to predict performance at.
              Distance units must match those given in dunit.rsc,
              although non-ambiguous abbreviations are allowed. E.g.,

                     800 meters
                     1000 me
                     1 mile
                     10 k
                     0.5 mar
                     1 mar
                     94000 hops

       - This file contains the data on
                     which rp will base its model. A data point
                     is a distance and a time, e.g.,

                     * This is a comment
                     * 1977 Data for Steve Ovett
                     400 meters     45.6     * Crystal Palace, June 18
                     1 mile         4:00     * Crystal Palace, June 18
                     0.5 mar        1:02:48  * Brighton, August 3

   Time strings are entered in the natural way, hh:mm:ss, mm:ss, or ss.s

   rp usage:

   rp [-12] [-q] [-u ] [-l ] []

      where:
               -u and -l override the default files

               The default function is to predict performance at the
               distances given in the distance file. Output is in
               tabular form.

               Alternatively:

               -12 -> simulates the 12-minute fitness test given to
                      Air Force trainees to assess basic condition
                      (see Dr. Kenneth Cooper's books)

                -q -> you will be prompted to input a 'standard run,'
                      say, your typical training run. The program
                      will print out a list of 'equivalent' runs, i.e.,
                      runs at the same level of effort in some sense.
                      This might be useful in planning for those long
                      20 mile training runs in preparation for a
                      marathon.

                rp with no arguments will read events from stdin
-------------------------------------------------------------------------------
Joe Knapp       jmk@asr2.att.com    jmk@cbnews.att.com   att!asr2!jmk
AT&T Bell Labs, Columbus, OH    (614)860-3547