Turn multiplication into linear algebra
An element of a field extension can act on the entire extension by multiplication. When the extension is finite-dimensional, this action is a linear map. Its trace and determinant turn an element of the larger field into elements of the base field, giving useful invariants without choosing a list of roots in advance.
Definition 1.2.1
(Trace and norm of an element)
Let be a finite field extension of degree . For , define the -linear map by . Define
These are elements of . A basis gives a matrix for , but changing the basis conjugates that matrix by an invertible matrix over . Trace and determinant are invariant under conjugation, so the definitions do not depend on the chosen basis. No separability or characteristic-zero assumption is needed for these definitions.
Properties from the maps
Theorem 1.2.2
(Basic properties of trace and norm)
For and , the trace is -linear and the norm is multiplicative:
For , and . If , then and .
Proof
(Matrices of multiplication maps)
Since , for , and , the trace is additive and -linear, and the determinant is multiplicative.
For , multiplication by is , so and , with the integer interpreted in . If , multiplication by has inverse , so and . The norm of zero is zero because .
Proof X-ray
Linearity and multiplicativity are different statements
Trace is linear because multiplication maps add. Norm is multiplicative because multiplication maps compose and determinants multiply. The determinant is not generally additive, so no analogous argument says . Basis independence is a separate change-of-coordinates fact; it should not be confused with either property.
Common mistake
(A base-field scalar has norm , not )
For the trace is and the norm is : the degree of the extension enters both formulas. In , for instance, and , not and .
A quadratic example
Worked example 1.2.3
(Trace and norm in )
Take with basis . For , multiplication by sends to and to . Thus its matrix, with columns given by these images, is
For example . When , its norm is nonzero and
The denominator is the product of the element and its conjugate. It is nonzero by the invertibility argument above; the displayed calculation also verifies the inverse directly.
Counterexample mode
Trace zero does not mean the element is zero
In this quadratic extension, but . Also , whereas . Thus trace can have a nontrivial kernel and norm need not be additive.
A characteristic-two warning
Worked example 1.2.4
(Trace of one in characteristic two)
Let , a field because the polynomial has no root in and is quadratic. The extension has basis and degree two. Therefore in characteristic two, even though . The formula must be read inside the base field, not as a positive real number. This example does not claim that the whole trace map is zero.
Worked example 1.2.5
(A cubic extension and a trace-zero generator)
Let with , and use the basis . Multiplication by sends , and , so
For the matrix is , with trace and determinant . Thus . The element generates and still has trace zero, so a zero trace says nothing about whether an element lies in the base field.
Exercises
Quick check
(Compute the trace and norm of over .)
Solution · Answer
The trace is and the norm is .
Quick check
(Verify that .)
Solution · Answer
Both factors on the left are , giving one. On the right, is a base-field scalar and its norm is .
Quick check
(In the characteristic-two example, compute the matrix of multiplication by in the stated basis, and its trace and norm.)
Solution · Answer
The relation is . The columns are and , so the matrix is . Its trace and determinant are both one in .
Quick check
(Why is for nonzero even if the extension is not assumed separable?)
Solution · Answer
Multiplication by is invertible with inverse multiplication by , so its determinant is nonzero. This uses the field property, not separability.