In algebra, Pfister's sixteen-square identity is a non-bilinear identity of form
(
+
x
2
+
x
3
2
+ ⋯ +
16
1
y
y
3
+ ⋯ +
16
It was first proven to exist by H. Zassenhaus and W. Eichhorn in the 1960s,[1] and independently by Albrecht Pfister[2] around the same time.
There are several versions, a concise one of which is
{\displaystyle {\begin{aligned}&\scriptstyle {z_{1}={\color {blue}{x_{1}y_{1}-x_{2}y_{2}-x_{3}y_{3}-x_{4}y_{4}-x_{5}y_{5}-x_{6}y_{6}-x_{7}y_{7}-x_{8}y_{8}}}+u_{1}y_{9}-u_{2}y_{10}-u_{3}y_{11}-u_{4}y_{12}-u_{5}y_{13}-u_{6}y_{14}-u_{7}y_{15}-u_{8}y_{16}}\\&\scriptstyle {z_{2}={\color {blue}{x_{2}y_{1}+x_{1}y_{2}+x_{4}y_{3}-x_{3}y_{4}+x_{6}y_{5}-x_{5}y_{6}-x_{8}y_{7}+x_{7}y_{8}}}+u_{2}y_{9}+u_{1}y_{10}+u_{4}y_{11}-u_{3}y_{12}+u_{6}y_{13}-u_{5}y_{14}-u_{8}y_{15}+u_{7}y_{16}}\\&\scriptstyle {z_{3}={\color {blue}{x_{3}y_{1}-x_{4}y_{2}+x_{1}y_{3}+x_{2}y_{4}+x_{7}y_{5}+x_{8}y_{6}-x_{5}y_{7}-x_{6}y_{8}}}+u_{3}y_{9}-u_{4}y_{10}+u_{1}y_{11}+u_{2}y_{12}+u_{7}y_{13}+u_{8}y_{14}-u_{5}y_{15}-u_{6}y_{16}}\\&\scriptstyle {z_{4}={\color {blue}{x_{4}y_{1}+x_{3}y_{2}-x_{2}y_{3}+x_{1}y_{4}+x_{8}y_{5}-x_{7}y_{6}+x_{6}y_{7}-x_{5}y_{8}}}+u_{4}y_{9}+u_{3}y_{10}-u_{2}y_{11}+u_{1}y_{12}+u_{8}y_{13}-u_{7}y_{14}+u_{6}y_{15}-u_{5}y_{16}}\\&\scriptstyle {z_{5}={\color {blue}{x_{5}y_{1}-x_{6}y_{2}-x_{7}y_{3}-x_{8}y_{4}+x_{1}y_{5}+x_{2}y_{6}+x_{3}y_{7}+x_{4}y_{8}}}+u_{5}y_{9}-u_{6}y_{10}-u_{7}y_{11}-u_{8}y_{12}+u_{1}y_{13}+u_{2}y_{14}+u_{3}y_{15}+u_{4}y_{16}}\\&\scriptstyle {z_{6}={\color {blue}{x_{6}y_{1}+x_{5}y_{2}-x_{8}y_{3}+x_{7}y_{4}-x_{2}y_{5}+x_{1}y_{6}-x_{4}y_{7}+x_{3}y_{8}}}+u_{6}y_{9}+u_{5}y_{10}-u_{8}y_{11}+u_{7}y_{12}-u_{2}y_{13}+u_{1}y_{14}-u_{4}y_{15}+u_{3}y_{16}}\\&\scriptstyle {z_{7}={\color {blue}{x_{7}y_{1}+x_{8}y_{2}+x_{5}y_{3}-x_{6}y_{4}-x_{3}y_{5}+x_{4}y_{6}+x_{1}y_{7}-x_{2}y_{8}}}+u_{7}y_{9}+u_{8}y_{10}+u_{5}y_{11}-u_{6}y_{12}-u_{3}y_{13}+u_{4}y_{14}+u_{1}y_{15}-u_{2}y_{16}}\\&\scriptstyle {z_{8}={\color {blue}{x_{8}y_{1}-x_{7}y_{2}+x_{6}y_{3}+x_{5}y_{4}-x_{4}y_{5}-x_{3}y_{6}+x_{2}y_{7}+x_{1}y_{8}}}+u_{8}y_{9}-u_{7}y_{10}+u_{6}y_{11}+u_{5}y_{12}-u_{4}y_{13}-u_{3}y_{14}+u_{2}y_{15}+u_{1}y_{16}}\\&\scriptstyle {z_{9}=x_{9}y_{1}-x_{10}y_{2}-x_{11}y_{3}-x_{12}y_{4}-x_{13}y_{5}-x_{14}y_{6}-x_{15}y_{7}-x_{16}y_{8}+x_{1}y_{9}-x_{2}y_{10}-x_{3}y_{11}-x_{4}y_{12}-x_{5}y_{13}-x_{6}y_{14}-x_{7}y_{15}-x_{8}y_{16}}\\&\scriptstyle {z_{10}=x_{10}y_{1}+x_{9}y_{2}+x_{12}y_{3}-x_{11}y_{4}+x_{14}y_{5}-x_{13}y_{6}-x_{16}y_{7}+x_{15}y_{8}+x_{2}y_{9}+x_{1}y_{10}+x_{4}y_{11}-x_{3}y_{12}+x_{6}y_{13}-x_{5}y_{14}-x_{8}y_{15}+x_{7}y_{16}}\\&\scriptstyle {z_{11}=x_{11}y_{1}-x_{12}y_{2}+x_{9}y_{3}+x_{10}y_{4}+x_{15}y_{5}+x_{16}y_{6}-x_{13}y_{7}-x_{14}y_{8}+x_{3}y_{9}-x_{4}y_{10}+x_{1}y_{11}+x_{2}y_{12}+x_{7}y_{13}+x_{8}y_{14}-x_{5}y_{15}-x_{6}y_{16}}\\&\scriptstyle {z_{12}=x_{12}y_{1}+x_{11}y_{2}-x_{10}y_{3}+x_{9}y_{4}+x_{16}y_{5}-x_{15}y_{6}+x_{14}y_{7}-x_{13}y_{8}+x_{4}y_{9}+x_{3}y_{10}-x_{2}y_{11}+x_{1}y_{12}+x_{8}y_{13}-x_{7}y_{14}+x_{6}y_{15}-x_{5}y_{16}}\\&\scriptstyle {z_{13}=x_{13}y_{1}-x_{14}y_{2}-x_{15}y_{3}-x_{16}y_{4}+x_{9}y_{5}+x_{10}y_{6}+x_{11}y_{7}+x_{12}y_{8}+x_{5}y_{9}-x_{6}y_{10}-x_{7}y_{11}-x_{8}y_{12}+x_{1}y_{13}+x_{2}y_{14}+x_{3}y_{15}+x_{4}y_{16}}\\&\scriptstyle {z_{14}=x_{14}y_{1}+x_{13}y_{2}-x_{16}y_{3}+x_{15}y_{4}-x_{10}y_{5}+x_{9}y_{6}-x_{12}y_{7}+x_{11}y_{8}+x_{6}y_{9}+x_{5}y_{10}-x_{8}y_{11}+x_{7}y_{12}-x_{2}y_{13}+x_{1}y_{14}-x_{4}y_{15}+x_{3}y_{16}}\\&\scriptstyle {z_{15}=x_{15}y_{1}+x_{16}y_{2}+x_{13}y_{3}-x_{14}y_{4}-x_{11}y_{5}+x_{12}y_{6}+x_{9}y_{7}-x_{10}y_{8}+x_{7}y_{9}+x_{8}y_{10}+x_{5}y_{11}-x_{6}y_{12}-x_{3}y_{13}+x_{4}y_{14}+x_{1}y_{15}-x_{2}y_{16}}\\&\scriptstyle {z_{16}=x_{16}y_{1}-x_{15}y_{2}+x_{14}y_{3}+x_{13}y_{4}-x_{12}y_{5}-x_{11}y_{6}+x_{10}y_{7}+x_{9}y_{8}+x_{8}y_{9}-x_{7}y_{10}+x_{6}y_{11}+x_{5}y_{12}-x_{4}y_{13}-x_{3}y_{14}+x_{2}y_{15}+x_{1}y_{16}}\end{aligned}}}
are set equal to zero, then it reduces to Degen's eight-square identity (in blue).
{\displaystyle {\begin{aligned}&u_{1}={\tfrac {\left(ax_{1}^{2}+x_{2}^{2}+x_{3}^{2}+x_{4}^{2}+x_{5}^{2}+x_{6}^{2}+x_{7}^{2}+x_{8}^{2}\right)x_{9}-2x_{1}\left(bx_{1}x_{9}+x_{2}x_{10}+x_{3}x_{11}+x_{4}x_{12}+x_{5}x_{13}+x_{6}x_{14}+x_{7}x_{15}+x_{8}x_{16}\right)}{c}}\\&u_{2}={\tfrac {\left(x_{1}^{2}+ax_{2}^{2}+x_{3}^{2}+x_{4}^{2}+x_{5}^{2}+x_{6}^{2}+x_{7}^{2}+x_{8}^{2}\right)x_{10}-2x_{2}\left(x_{1}x_{9}+bx_{2}x_{10}+x_{3}x_{11}+x_{4}x_{12}+x_{5}x_{13}+x_{6}x_{14}+x_{7}x_{15}+x_{8}x_{16}\right)}{c}}\\&u_{3}={\tfrac {\left(x_{1}^{2}+x_{2}^{2}+ax_{3}^{2}+x_{4}^{2}+x_{5}^{2}+x_{6}^{2}+x_{7}^{2}+x_{8}^{2}\right)x_{11}-2x_{3}\left(x_{1}x_{9}+x_{2}x_{10}+bx_{3}x_{11}+x_{4}x_{12}+x_{5}x_{13}+x_{6}x_{14}+x_{7}x_{15}+x_{8}x_{16}\right)}{c}}\\&u_{4}={\tfrac {\left(x_{1}^{2}+x_{2}^{2}+x_{3}^{2}+ax_{4}^{2}+x_{5}^{2}+x_{6}^{2}+x_{7}^{2}+x_{8}^{2}\right)x_{12}-2x_{4}\left(x_{1}x_{9}+x_{2}x_{10}+x_{3}x_{11}+bx_{4}x_{12}+x_{5}x_{13}+x_{6}x_{14}+x_{7}x_{15}+x_{8}x_{16}\right)}{c}}\\&u_{5}={\tfrac {\left(x_{1}^{2}+x_{2}^{2}+x_{3}^{2}+x_{4}^{2}+ax_{5}^{2}+x_{6}^{2}+x_{7}^{2}+x_{8}^{2}\right)x_{13}-2x_{5}\left(x_{1}x_{9}+x_{2}x_{10}+x_{3}x_{11}+x_{4}x_{12}+bx_{5}x_{13}+x_{6}x_{14}+x_{7}x_{15}+x_{8}x_{16}\right)}{c}}\\&u_{6}={\tfrac {\left(x_{1}^{2}+x_{2}^{2}+x_{3}^{2}+x_{4}^{2}+x_{5}^{2}+ax_{6}^{2}+x_{7}^{2}+x_{8}^{2}\right)x_{14}-2x_{6}\left(x_{1}x_{9}+x_{2}x_{10}+x_{3}x_{11}+x_{4}x_{12}+x_{5}x_{13}+bx_{6}x_{14}+x_{7}x_{15}+x_{8}x_{16}\right)}{c}}\\&u_{7}={\tfrac {\left(x_{1}^{2}+x_{2}^{2}+x_{3}^{2}+x_{4}^{2}+x_{5}^{2}+x_{6}^{2}+ax_{7}^{2}+x_{8}^{2}\right)x_{15}-2x_{7}\left(x_{1}x_{9}+x_{2}x_{10}+x_{3}x_{11}+x_{4}x_{12}+x_{5}x_{13}+x_{6}x_{14}+bx_{7}x_{15}+x_{8}x_{16}\right)}{c}}\\&u_{8}={\tfrac {\left(x_{1}^{2}+x_{2}^{2}+x_{3}^{2}+x_{4}^{2}+x_{5}^{2}+x_{6}^{2}+x_{7}^{2}+ax_{8}^{2}\right)x_{16}-2x_{8}\left(x_{1}x_{9}+x_{2}x_{10}+x_{3}x_{11}+x_{4}x_{12}+x_{5}x_{13}+x_{6}x_{14}+x_{7}x_{15}+bx_{8}x_{16}\right)}{c}}\end{aligned}}}
The identity shows that, in general, the product of two sums of sixteen squares is the sum of sixteen rational squares.
Incidentally, the
also obey,
No sixteen-square identity exists involving only bilinear functions since Hurwitz's theorem states an identity of the form
bilinear functions of the
However, the more general Pfister's theorem (1965) shows that if the
are rational functions of one set of variables, hence has a denominator, then it is possible for all
[3] There are also non-bilinear versions of Euler's four-square and Degen's eight-square identities.