How far off the edge of the table can we reach by stacking <i>n</i> identical blocks of length 1? A classical solution achieves an overhang of 1/2<i>H</i><inf><i>n</i></inf>, where <i>H</i><inf><i>n</i></inf> = Σ<sup><i>n</i></sup><inf><i>i</i>=1</inf> 1/<i>i</i> ~ ln <i>n</i> is the <i>n</i><sup>th</sup> harmonic number, by stacking all the blocks one on top of another with the <i>i</i><sup>th</sup> block from the top displaced by 1/2<i>i</i> beyond the block below. This solution is widely believed to be optimal. We show that it is exponentially far from optimal by giving explicit constructions with an overhang of Ω(<i>n</i><sup>1/3</sup>). We also prove some upper bounds on the overhang that can be achieved. The stability of a given stack of blocks corresponds to the feasibility of a <i>linear program</i> and so can be efficiently determined.
[1]
J. F. Hall.
Fun with stacking blocks
,
2005
.
[2]
George Minchin Minchin.
A treatise on statics with applications to physics
,
1884
.
[3]
Paul B. Johnson.
Leaning Tower of Lire
,
1955
.
[4]
Martin Gardner,et al.
Sixth book of mathematical games from Scientific American
,
1971
.
[5]
Leonard Eisner,et al.
Leaning Tower of The Physical Reviews
,
1959
.
[6]
Alexander Schrijver,et al.
Theory of linear and integer programming
,
1986,
Wiley-Interscience series in discrete mathematics and optimization.
[7]
Frank Potter,et al.
Mad about physics : braintwisters, paradoxes, and curiosities
,
2001
.
[8]
Richard M. Sutton.
A Problem of Balancing
,
1955
.