Ci field ($N>d^i$)

A field is $C_i$ if every homogeneous polynomial of degree $d\ge 1$ in more than $d^i$ variables has a nonzero zero. Algebraically closed fields are $C_0$, finite fields are $C_1$, and the Lang-Nagata theorem for Ci fields controls function-field extensions.

A finitely generated field extension of transcendence degree $s$ of a $C_i$ field is $C_{i+s}$. In particular a function field of a curve over a C1 field is $C_2$. The theorem includes stability under finite algebraic extensions when $s=0$.

A $C_2$ field is a Ci field with the bound $N>d^2$. Function fields of curves over finite fields are examples. The extra-variable construction in surjectivity of reduced norms over C2 fields makes this polynomial-solvability property useful for division algebras.

For a central division algebra $D$ of degree $d$ over a C2 field, the form $\mathrm{Nrd}(X)-a{T}^d$ has degree $d$ in $d^2+1$ variables. Its nontrivial zero must have $T\ne 0$, because the reduced norm of a division algebra is anisotropic. Dividing $X$ by $T$ gives norm $a$. Zero is the norm of zero.

A $C_1$ field is a Ci field with the bound $N>d$. The Chevalley-Warning theorem gives finite-field examples; the Tsen theorem gives function fields of curves over algebraically closed fields. A $C_1$ field has trivial Brauer group.

A central division algebra of degree $d>1$ has an anisotropic reduced norm of degree $d$ in $d^2>d$ variables. That contradicts the C1 field property. Thus only degree one is possible, and every central simple algebra is split.

For a degree-$m$ extension $L/K$, substitute a $K$-basis of $L$ into a homogeneous degree-$d$ polynomial in $N>d$ variables over $L$, then take its field norm. The resulting degree-$md$ polynomial has $mN>md$ variables over $K$. A nonzero zero over a C1 field gives a nonzero zero of the original form because the norm of a field element vanishes only at zero. Separability is unnecessary.

The function field of a curve over an algebraically closed field is a C1 field. Consequently its Brauer group is zero. This is the input for cyclic splitting by extension of constants over a finite constant field.

For the function field $K$ of a curve over a finite full constant field $k$, extend constants to $\overline{k}$. The resulting curve function field is $C_1$ by the Tsen theorem, so every central simple algebra splits there. The finite collection of coefficients of a splitting isomorphism descends to some finite constant extension $k^{\prime }/k$. Regularity makes $K{k}^{\prime }/K$ cyclic, giving a finite cyclic splitting field.