Loading ...
Sorry, an error occurred while loading the content.
 

absolute retracts in varieties of algebras

Expand Messages
  • p_ouwehand
    I have a simple characterization of the absolute retracts in a variety of algebras: An algebra A is an absolute retract if and only if it is (i) equationally
    Message 1 of 2 , May 15, 2005
      I have a simple characterization of the absolute retracts in a
      variety of algebras: An algebra A is an absolute retract if and only
      if it is (i) equationally compact, and (ii) algebraically closed
      (i.e. every finite set of equations satisfiable in some extension of
      A is already satisfiable in A). In particular, a finite algebra is an
      absolute retract if and only if it it is algebraically closed.

      From previous work, I know that in a congruence distributive variety
      every finite absolute retract is a product of maximal subdirectly
      irreducibles (i.e. those s.i.'s that have no proper essential
      extensions), and that in lattice varieties, the converse holds as
      well.

      Thus in a variety of lattices, the finite algebraically closed
      lattices are precisely products of maximal subdirectly irreducibles.
      E.g., the finite AC distributive lattices are exactly the finite
      Boolean lattices, a result apparently due to Schmid (1979).

      What I would like to know is the following: (1) Are the above results
      already known (and where can I find them)?, and (2) Are there any
      significant papers that study algebraically closed algebras in
      general (particularly the congruence distributive case)?
    • mhebert
      p_ouwehand asked: I have a simple characterization of the absolute retracts in a variety of algebras: An algebra A is an absolute retract if and only if
      Message 2 of 2 , May 15, 2005
        p_ouwehand    asked:
         
        " I have a simple characterization of the absolute retracts in a
        variety of algebras: An algebra A is an absolute retract if and only
         if it is (i) equationally compact, and (ii) algebraically closed
         (i.e. every finite set of equations satisfiable in some extension of
         A is already satisfiable in A). In particular, a finite algebra is an
        absolute retract if and only if it it is algebraically closed.
        [...]
        (1) Are the above results
         already known (and where can I find them)?, and (2) Are there any
        significant papers that study algebraically closed algebras in
         general (particularly the congruence distributive case)? "
         
        This characterization actually holds in the wider context of accessible categories with pushouts. (Those are more general than  locally presentable categories , which in turn are something a little bit more general than quasivarieties.)
         
        One of the problems here is the abundance of terminology for very close (or even equivalent) concepts.
        In particular an algebraically closed embeddings are also called a pure embeddings, and this has been studied quite a lot in locally presentable categories. An excellent general reference is
         
        Locally presentable and accessible categories, by J.Adamek and J. Rosicky, London MAth Soc Lecture Note Series 189, 1994. 
         
        More specific works on your question would be my
        (1) Algebraically closed and existentially closed substructures in categorical context, Theory and Applications of Categories 12 (2004), 269-298 ( pdf )
        and
        (2) K-purity and orthogonality, Theory and Applications of Categories 12 (2004), 355-371 (pdf)
         
        In case you are more familiar with the classical model-theoretic notations, a proof that "equationally compact" is equivalent to "every pure embedding is a retract" in a wide context is in
        Hodges, Model Theory1993 p. 528.
        That this is equivalent to "pure-injective" follows from the fact that pushouts preserve pure embeddings (and maybe by model-theoretic methods too). 
         
        Your result may not be mentioned explicitely in any of the above,  but it now reads as
        "Algebraically closed + pure-injectivity  <=>  Absolute retracts"
        which follows immediately from some of the  known equivalent definitions of purity, recalled in (1) and (2).
         
        Michel Hebert
         
        Let me know if a more precisely written proof is needed
         
         
         
         

         
         
        Fromunivalg@yahoogroups.com
        Tounivalg@yahoogroups.com
        Cc
        DateSun, 15 May 2005 08:27:39 -0000
        Subject[univalg] absolute retracts in varieties of algebras
          
        > I have a simple characterization of the absolute retracts in a
        > variety of algebras: An algebra A is an absolute retract if and only
        > if it is (i) equationally compact, and (ii) algebraically closed
        > (i.e. every finite set of equations satisfiable in some extension of
        > A is already satisfiable in A). In particular, a finite algebra is an
        > absolute retract if and only if it it is algebraically closed.
        >
        > From previous work, I know that in a congruence distributive variety
        > every finite absolute retract is a product of maximal subdirectly
        > irreducibles (i.e. those s.i.'s that have no proper essential
        > extensions), and that in lattice varieties, the converse holds as
        > well.
        >
        > Thus in a variety of lattices, the finite algebraically closed
        > lattices are precisely products of maximal subdirectly irreducibles.
        > E.g., the finite AC distributive lattices are exactly the finite
        > Boolean lattices, a result apparently due to Schmid (1979).
        >
        > What I would like to know is the following: (1) Are the above results
        > already known (and where can I find them)?, and (2) Are there any
        > significant papers that study algebraically closed algebras in
        > general (particularly the congruence distributive case)?
        >
        >
        >
        >
        >
        > Yahoo! Groups Links
        >
        > <*> To visit your group on the web, go to:
        > http://groups.yahoo.com/group/univalg/
        >
        > <*> To unsubscribe from this group, send an email to:
        > univalg-unsubscribe@yahoogroups.com
        >
        > <*> Your use of Yahoo! Groups is subject to:
        > http://docs.yahoo.com/info/terms/
        >
        >
        >
      Your message has been successfully submitted and would be delivered to recipients shortly.