characterization of prime ideals


This entry gives a number of equivalentMathworldPlanetmathPlanetmathPlanetmathPlanetmathPlanetmath http://planetmath.org/node/5865characterizations of prime ideals in rings of different generality.

We start with a general ring R.

Theorem 1.

Let R be a ring and P⊊R a two-sided idealMathworldPlanetmath. Then the following statements are equivalent:

  1. 1.

    Given (left, right or two-sided) ideals I,J of P such that the product of ideals I⁢J⊆P, then I⊆P or J⊆P.

  2. 2.

    If x,y∈R such that x⁢R⁢y⊆P, then x∈P or y∈P.

Proof.
  • •

    “1⇒2”:

    Let x,y∈R such that x⁢R⁢y⊆P. Let (x) and (y) be the (left, right or two-sided) ideals generated by x and y, respectively. Then each element of the product of ideals (x)⁢R⁢(y) can be expanded to a finite sum of productsPlanetmathPlanetmathPlanetmathPlanetmath each of which contains or is a factor of the form ±x⁢r⁢y for a suitable r∈R. Since P is an ideal and x⁢R⁢y⊆P, it follows that (x)⁢R⁢(y)⊆P. Assuming statement 1, we have (x)⊆P, R⊆P or (y)⊆P. But P⊊R, so we have (x)⊆P or (y)⊆P and hence x∈P or y∈P.

  • •

    “2⇒1”:

    Let I,J be (left, right or two-sided) ideals, such that the product of ideals I⁢J⊆P. Now R⁢J⊆J or I⁢R⊆I (depending on what type of ideal we consider), so I⁢R⁢J⊆I⁢J⊆P. If I⊆P, nothing remains to be shown. Otherwise, let i∈I∖P, then i⁢R⁢j⊆P for all j∈J. Since i∉P we have by statement 2 that j∈P for all j∈J, hence J⊆P.

∎

There are some additional properties if our ring is commutativePlanetmathPlanetmathPlanetmath.

Theorem 2.

Let R a commutative ring and P⊊R an ideal. Then the following statements are equivalent:

  1. 1.

    Given ideals I,J of P such that the product of ideals I⁢J⊆P, then I⊆P or J⊆P.

  2. 2.
  3. 3.

    The set R∖P is a subsemigroup of the multiplicative semigroup of R.

  4. 4.

    Given x,y∈R such that x⁢y∈P, then x∈P or y∈P.

  5. 5.

    The ideal P is maximal in the set of such ideals of R which do not intersect a subsemigroup S of the multiplicative semigroup of R.

Proof.
  • •

    “1⇒2”:

    Let x¯,y¯∈R/P be arbitrary nonzero elements. Let x and y be representatives of x¯ and y¯, respectively, then x∉P and y∉P. Since R is commutative, each element of the product of ideals (x)⁢(y) can be written as a product involving the factor x⁢y. Since P is an ideal, we would have (x)⁢(y)⊆P if x⁢y∈P which by statement 1 would imply (x)⊆P or (y)⊆P in contradictionMathworldPlanetmathPlanetmath with x∉P and y∉P. Hence, x⁢y∉P and thus x¯⁢y¯≠0.

  • •

    “2⇒3”:

    Let x,y∈R∖P. Let π:R→R/P be the canonical projection. Then π⁢(x) and π⁢(y) are nonzero elements of R/P. Since π is a homomorphismMathworldPlanetmathPlanetmathPlanetmathPlanetmathPlanetmathPlanetmathPlanetmathPlanetmath and due to statement 2, π⁢(x)⁢π⁢(y)=π⁢(x⁢y)≠0. Therefore x⁢y∉P, that is R∖P is closed under multiplication. The associative property is inherited from R.

  • •

    “3⇒4”:

    Let x,y∈R such that x⁢y∈P. If both x,y were not elements of P, then by statement 3 x⁢y would not be an element of P. Therefore at least one of x,y is an element of P.

  • •

    “4⇒1”:

    Let I,J be ideals of R such that I⁢J⊆P. If I⊆P, nothing remains to be shown. Otherwise, let i∈I∖P. Then for all j∈J the product i⁢j∈I⁢J, hence i⁢j∈P. It follows by statement 4 that j∈P, and therefore J⊆P.

  • •

    “4⇒5”:

    The condition 4 that the set  S=R∖P  is a multiplicative semigroup.  Now P is trivially the greatest ideal which does not intersect S.

  • •

    “5⇒4”:

    We presume that P is maximal of the ideals of R which do not intersect a semigroup S and that  x⁢y∈P.  Assume the contrary of the assertion, i.e. that  x∉P  and  y∉P.  Therefore, P is a proper subsetMathworldPlanetmathPlanetmath of both  (P,x)  and  (P,y).  Thus the maximality of P implies that

    (P,x)∩S≠{},(P,y)∩S≠{}.

    So we can choose the elements s1 and s2 of S such that

    s1=p1+r1⁢x+n1⁢x,s2=p2+r2⁢y+n2⁢y,

    where  p1,p2∈P,   r1,r2∈R  and  n1,n2∈ℤ.  Then we see that the product

    s1⁢s2=(p1+r2⁢y+n2⁢y)⁢p1+(r1⁢x+n1⁢x)⁢p2+(r1⁢r2+n2⁢r1+n1⁢r2)⁢x⁢y+(n1⁢n2)⁢x⁢y

    would belong to the ideal P.  But this is impossible because s1⁢s2 is an element of the multiplicative semigroup S and P does not intersect S.  Thus we can conclude that either x or y belongs to the ideal P.

∎

If R has an identity elementMathworldPlanetmath 1, statements 2 and 3 of the preceding theorem become stronger:

Theorem 3.

Let R be a commutative ring with identity element 1. Then an ideal P of R is a prime idealMathworldPlanetmathPlanetmath if and only if R/P is an integral domainMathworldPlanetmath. Furthermore, P is prime if and only if R∖P is a monoid with identity element 1 with respect to the multiplication in R.

Proof.

Let P be prime, then 1∉P since otherwise P would be equal to R. Now by theorem 2 R/P is a cancellation ring. The canonical projection π:R→R/P is a homomorphism, so π⁢(1) is the identity element of R/P. This in turn implies that the semigroup R∖P is a monoid with identity element 1. ∎

Title characterization of prime ideals
Canonical name CharacterizationOfPrimeIdeals
Date of creation 2013-03-22 15:22:01
Last modified on 2013-03-22 15:22:01
Owner GrafZahl (9234)
Last modified by GrafZahl (9234)
Numerical id 9
Author GrafZahl (9234)
Entry type Result
Classification msc 13C05
Classification msc 16D25
Synonym characterisation of prime ideals
Related topic LocalizationMathworldPlanetmath
Related topic QuotientRingModuloPrimeIdeal