[Search for users] [Overall Top Noters] [List of all Conferences] [Download this site]

Conference rusure::math

Title:Mathematics at DEC
Moderator:RUSURE::EDP
Created:Mon Feb 03 1986
Last Modified:Fri Jun 06 1997
Last Successful Update:Fri Jun 06 1997
Number of topics:2083
Total number of notes:14613

22.0. "120 degree angles in lattice" by HARE::STAN () Tue Jan 31 1984 11:24

Another problem I got from George Berzsenyi is the following:

Are there three points in a 3-dimensional lattice that form
a triangle with integer sides and having a 120 degree angle?

(A lattice refers to those points with integer coordinates in a
standard rectangular coordinate system.)

I believe it is known that in a 2-dimensional lattice, no triangle can have
an angle of 60 degrees or 120 degrees.

I found an angle of 120 degrees in the 3-dimensional lattice
(though the triangle did not have integer sides).

Let A=(0,0,0), B=(1,1,18), C=(4,1,15), then angle ACB is 120 degrees.

I have not yet found such a triangle where the lengths of all
the sides are integers.
T.RTitleUserPersonal
Name
DateLines
22.1Bringing a whole new dimension to the problem!ZFC::DERAMOHello my node::name is ...Sun Dec 13 1987 13:2323
    In four dimensions, there is
    
    A (0,0,0,0)    BC = 14
    B (10,0,0,0)   AC =  6
    C (-3,5,1,1)   AB = 10
    
    The angle at the origin is 120 degrees.
    
    If you put the 120 degree angle at the origin (point A)
    and point B at (a,b,c) and point C at (d,e,f) then you
    get the four equations in integer variables:
    
       r^2 = a^2 + b^2 + c^2
       s^2 = d^2 + e^2 + f^2
       t^2 = (a-d)^2 + (b-e)^2 + (c-f)^2
       t^2 = r^2 + rs + s^2                 [this from the law of cosines]
    
    I am about to program a search for solutions with the absolute values
    of a-f less than or equal to 100.  I doubt that I will find anything.
    Before I start, has anyone done this kind of search already?
    
    Dan
    
22.2Impossibility proofHERON::BUCHANANMon Dec 14 1987 10:4455
    What about the following argument, using your notation in .1, Dan.
    i.e. looking for integer soln to
       (i) r^2 = a^2 + b^2 + c^2
       (ii) s^2 = d^2 + e^2 + f^2
       (iii) t^2 = (a-d)^2 + (b-e)^2 + (c-f)^2
       (iv) t^2 = p^2 + p.q + q^2
    with r,s,t all > 0.
    
    If I've got 4 integers W, X, Y, Z such that
    W^2 = X^2 + Y^2 + Z^2
    then EITHER  W,X,Y,Z all even
         OR W and X (say) odd, and the other two integers, Y and Z even
    
    In the problem in hand, by looking at (i) and (ii) in
    this way, there are 4 cases without loss of generality:
    
    Case 1) a,b,c,d,e,f all even.   In this case we can divide everything
    by 2 repeatedly until we reach one of the other cases (Or all of
    a,b,c,d,e,f are zero, in which case r,s,t are all zero).   So ignore
    this case.
    
    Case 2) a and d odd, b,c,e,f even
    Case 3) a and e odd, b,c,d,f even
    In either of these cases r and s are both odd.   However, if we
    combine (iii) and (iv) and then substitute out r^2 and s^2, we have
    
    (v) r.s = -2(a.d + b.e + c.f)
    
    so at least one of r and s must be even
    
    Thus we are forced to conclude...
    Case 4) a is odd b,c,d,e and f all even.
    r is odd and s is even.
    
    Substitute in (ii) and (v) with r = 2R + 1, s = 2S, d = 2D, 
    e = 2E, f = 2F...
    
    (ii') S^2 = D^2 + E^2 + F^2
    (v') -R.S -(S/2) = a.D + b.E + c.F
    
    so by (v') S is even so by (ii') D,E and F are even also
    
    so substitute S = 2S', D = 2D', E = 2E' and F = 2F'
    
    (ii'') S'^2 = D'^2 + E'^2 + F'^2
    (v'') -R.S' -(S'/2) = a.D' + b.E' + c.F'
    
    But this is where we were last time round.   i.e. S,D,E and F have
    as a factor 2^n for all n. i.e. S,D.E.F are all 0.
    
    CONCLUSION: The proposed construction is impossible in three
    dimensions.
    
    Cheers,
    Andy Buchanan.
22.3case closedZFC::DERAMOHello my node::name is ...Mon Dec 14 1987 11:2622
     I'm impressed!

     In case anyone is confused by some of your points:

>>    If I've got 4 integers W, X, Y, Z such that
>>    W^2 = X^2 + Y^2 + Z^2
>>    then EITHER  W,X,Y,Z all even
>>         OR W and X (say) odd, and the other two integers, Y and Z even

    ... because the only squares mod 4 are 0 and 1.

>>    Case 2) a and d odd, b,c,e,f even
>>    Case 3) a and e odd, b,c,d,f even

     ... because the odd one in d-e-f is either in the same
     coordinate as the odd one in a-b-c or it isn't.

     Everything else was very straightforward.  That's neat.  Not
     only do we all learn something, but I don't have to write
     the search code, either! :^)

     Dan