In mathematics, theHurwitz problem (named afterAdolf Hurwitz) is the problem of finding multiplicative relations betweenquadratic forms which generalise those known to exist between sums of squares in certain numbers of variables.
There are well-known multiplicative relationships between sums of squares in two variables
(known as theBrahmagupta–Fibonacci identity), and alsoEuler's four-square identity andDegen's eight-square identity. These may be interpreted as multiplicativity for the norms on thecomplex numbers (),quaternions (), andoctonions (), respectively.[1]: 1–3 [2]
The Hurwitz problem for the fieldK is to find general relations of the form
with thez being bilinear forms in thex andy: that is, eachz is aK-linear combination of terms of the formxi yj.[3]: 127
We call a tripleadmissible forK if such an identity exists.[1]: 125 Trivial cases of admissible triples include The problem is uninteresting forK ofcharacteristic 2, since over such fields every sum of squares is a square, and we exclude this case. It is believed that otherwise admissibility is independent of the field of definition.[1]: 137
Hurwitz posed the problem in 1898 in the special case and showed that, when coefficients are taken in, the only admissible values were[3]: 130 His proof extends to a field of anycharacteristicexcept 2.[1]: 3
The "Hurwitz–Radon" problem is that of finding admissible triples of the form Obviously is admissible. TheHurwitz–Radon theorem states that is admissible over any field where is the function defined forv odd, with and[1]: 137 [3]: 130
Other admissible triples include[1]: 138 and[1]: 137