home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!comp.vuw.ac.nz!canterbury.ac.nz!equinox.gen.nz!equinox!bignode!pete
- From: pete@bignode.equinox.gen.nz (Pete Moore)
- Subject: Re: Finding best mate
- Newsgroups: sci.math
- References: <your-email-addr-040193123900@microlab22.med.upenn.edu>
- Organization: Equinox Networks
- Reply-To: pete@bignode.equinox.gen.nz
- X-Newsreader: TIN [version 1.1 PL8]
- Message-ID: <pete.03mf@bignode.equinox.gen.nz>
- Date: 5 Jan 93 20:05:52 +1200
- Organization: Equinox Networks
- Lines: 48
-
- Jesse Goldman (your-email-addr@a1.mscf.upenn.edu) wrote:
- >Dear Mathematicians;
-
- > This weekend I saw an interview on T.V. with a mathematician who
- >stated that if a woman assumes she is to have 100 suitors in her life then
- >her best chances of finding an ideal mate would be to reject the first 37
- >and then choose the next one better than the preceeding 37.
-
- I've seen this referred to as the `secretary problem' - you interview 100
- candidates for a job as your secretary, and must decide at the end of each
- interview whether or not to hire the current candidate. If you decide no,
- the candidate leaves and can't be recalled, if yes, the interviews stop
- and any remaining candidates are sent away. The best strategy is to
- interview 37% (or 1/e more precisely) and send them away, then continue
- with the remaining candidates and select the first one who is better than
- all his/her predecessors. Of course, if the best candidate was actually in
- that first 37%, you are stuck with the last candidate!
-
- A couple of years ago I attended a lecture in which Paul Halmos presented
- a version of this problem as follows:
-
- I have a box containing 100 slips of paper. On each slip is a positive
- number (not necessarily an integer or even a rational). You may draw
- slips of paper one at a time and read the numbers thereon. At any point,
- you can stop and bet that you have just drawn the largest number in the
- box. I will give you 4 to 1 odds. Will you play?
-
- The lecture room (containing about a dozen professional mathematicians and
- an equal number of students) was completely silent - no-one was willing to
- accept Prof Halmos' offer!
- In fact, the simple stategy of examining & discarding the first 50, then
- betting on the first number thereafter which is bigger than all its
- predecessors, gives you a chance better than 1 in 4, since you will win if
- the 2nd-biggest number is in the first 50 (50% chance) and the biggest is
- in the 2nd 50 (50% chance) & there are other ways you can win also (e.g. if
- the 2 biggest numbers are both in the 2nd 50, provided that the biggest
- appears before the 2nd-biggest).
-
- The best strategy is again to examine and discard 1/e of the numbers and
- then bet on the first number (if any) which exceeds all its predecessors.
- You should play if you are offered any odds better than e to 1. Proving
- this is left as an exercise for the reader 8-)
- --
- +------------------- pete@bignode.equinox.gen.nz -------------------+
- | The effort to understand the universe is one of the very few things |
- | that lifts human life above the level of farce, and gives it some |
- | of the grace of tragedy - Steven Weinberg |
- +---------------------------------------------------------------------+
-