PROBLEM CORNER



A (growing) compilation of interesting little mathematics problems.
Feel free to send answers or pose some problems yourself!
(Just email them to me at branicky@lids.mit.edu. And please note the source!)
Induction Fun
Poser: Michael S. Branicky; Source: Donald Knuth

0-1 Determinants
Poser: Michael S. Branicky; Source: Mitchell Trott

Envelope Problem
Poser: Michael S. Branicky; Source: Thomas Cover

Parity Problem
Poser: Michael S. Branicky; Source: Paul G. Kwiat


Induction Fun

Poser: Michael S. Branicky, 3 May 1996

Source: Donald Knuth, Stanford University

Scenario: Here is a nice problem which teaches a valuable lesson about induction. Consider the recursion

K(0) = 1,
K(n+1) = 1 + min( 2 * K([n/2]), 3 * K([n/3]) ), n >= 0,

where [x] denotes the integer part of x. For convenience, the first several values of K(n) are given below.

Problem: Prove or disprove K(n) >= n.

Start-up:

   n: 0 1 2 3 4 5 6 7 8  9
K(n): 1 3 3 4 7 7 7 9 9 10



0-1 Determinants

Poser: Michael S. Branicky, 3 May 1996

Source: Mitchell Trott, M.I.T.

Scenario: Given an n by n matrix A, where entry

aij = { 0, with probability 1/2; 1, with probabilty 1/2,

independently, for all i, j.

Problem: What is Pr[det A = 0]?


Envelope Problem

Poser: Michael S. Branicky, 29 April 1996

Source: Thomas Cover, Stanford University

Scenario: Secretly, two real numbers are placed into two identical evelopes and sealed. You are presented with the two envelopes, can choose one to open, and, based on the number you see, may trade evelopes or not. The incentive: if the ultimate envelope you choose contains the number x, you receive x dollars (a negative x means you owe that amount). Assume you want to maximize x.

Problem: Is there a strategy by which you can recieve the maximum more than 1/2 of time? If so, demonstrate one. If not, prove none exists.


Parity Problem

Poser: Michael S. Branicky, 5 October 1995

Source: Paul G. Kwiat, Los Alamos Labs

Scenario: Secretly, that is without personal or mixed knowledge, to the foreheads of each of N people are attached a symbol (either a 0 or a 1) and a bandage covering it. The N people are then allowed to meet and discuss their respective "strategies" for the upcoming event: they are to go into a room, remove their bandages, and---without communicating to each other in any way---each secretly register a guess as to the "parity" of the entire group (i.e., a 0 if the total number of 1's is even, a 1 if it is odd). The "group answer" is computed by majority [strictly greater than one-half; half-half is undefined] vote of the registered guesses.

Problem: Is there a strategy by which the group can be correct more than 1/2 of time? If so, demonstrate one. If not, prove none exists.



Created before: 1997-09-09
Last modified: 1999-09-04
mb@ieee.org (Michael Branicky)