%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%% Copyright (C) 1996, 1998
%% Joergen Backelin and Alexander Podoplelov
%%
%% Bergman and Anick are distributed in the hope that they will be
%% useful, but WITHOUT ANY WARRANTY.  No author or distributor
%% accepts responsibility to anyone for the consequences of using them
%% or for whether they serve any particular purpose or work at all,
%% unless (s)he says so in writing.  Refer to the Bergman General Public
%% License for full details.

%% Everyone is granted permission to copy, modify and redistribute
%% bergman, but only under the conditions described in the
%% Bergman General Public License.   A copy of this license is
%% supposed to have been given to you along with bergman so you
%% can know your rights and responsibilities.  It should be in a
%% file named copyright.  Among other things, the copyright notice
%% and this notice must be preserved on all copies.
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%



          The protocol file for Anick-in-bergman module.
          Designed under Standard Lisp environment.


                            SPECIFICATION

 This document is a part of the bergman documentation (complementing
 its document protocol.txt) and should be read after the main
 conventions are known, like types and structures. The most relevant
 pre-requisit is types.txt of the bergman documentation.
 For instance, the type template "degree list" is only briefly
 described here, but in order to use it properly one should penetrate
 its structure more thoroughly.

 The purpose of the document is to describe all the functions
 and their arguments used in the Anick module. This description is
 divided into sections corresponding to bergman modules. Within a
 bergman  module section a list of functions is placed. Every
 description of a function has the following structure:
   FUNCTION (ARG1:TYPE1 ARG2:TYPE2 ... ARGn:TYPEn) : RESULT ; COMMENT
 where ARGi are identifiers and TYPEi are their types (see TYPE
 CONVENTION). Then the description of each argument and of the
 result follows. The COMMENT part is explained in "types.txt" in the
 bergman doc directory.


 The prefix of the (non user interface) procedures and macros
 concerning "Anick's resolution" is "an".

	                  TYPE CONVENTION

 We use the following abbreviations for internal types:

  * denotes that the type is inherited from bergman. For more
    information about such types, see "types.txt" in the bergman
    doc directory.

     Abbreviation          Description
  -------------------------------------------------------------------
  *  id                    (Standard lisp) identifier.
  *  string                (Standard lisp) string of ASCII characters.
  *  coeff  (non-zero)     Coefficient (not necessary an integer).
     chn                   Chain.
  *  qpol                  Quotient Polynomial.
     inputmatrix           Matrix of coefficients (in proper input form).
     matrix		   Matrix of coefficients in internal bergman
			   form. (Actually a sub-type of llgenredandcoeff).
  *  int                   (Standard lisp) integer.
     tp                    Tensor Polynomial.
     btnum                 Betti number.
  *  augmon                Augmented Monomial.
     -                     Void (No or arbitrary return, not to be used).
  *  bool                  Boolean: NIL or non-NIL (with NIL standing for
			   False).
  *  any                   Any Structure.
     zero                  Internal Representation of Zero.
  *  modestate             Type for mode settings.
     ---
  *  l@@@                  Prefix "l" stands for: "list of @@@".
  *  (a@@ . b@@)           Dotted pairs of objects of types a@@ and b@@.
  *  dl@@@		   Degree list (a special form of l(int . l@@@),
			   as explained in the bergman document types.txt).
  *  dp@@@                 Prefix "dp" stands for: "dotted pair (@@@ . @@@)".
  -------------------------------------------------------------------


                        TYPE EXPLANATION

     In order to let the reader understand internal types more clearly,
     we describe most of the complicated ones below. For the sake of
     formality we stick to C style for declaring data.

       o chn - the chain type:
	     struct {
	        augmon vx;    /* last vertex of the chain seen as a  */
		              /* path in the graph;                  */
                int    n;     /* length of the chain;                */
		chn    pchn;  /* the chain representing the unique   */
		              /* chain of length n-1 which is a left */
                              /* factor of the chain;                */
		tp     df;    /* the calculated differential of the  */
		              /* chain;                              */
		chn   *hchn;  /* list of prolongatings for the chain;*/
             }
       o tp -  tensor polynomial type: a list whose first element is
               NULL and the rest are of the following structure:
	      struct {
	        chn    chain; /* left factor in the tensor           */
		              /* multiplication;                     */
		augrattp   p; /* right factor in the tensor          */
		              /* multiplication;                     */
	      }             


			    MODES DESCRIPTION
     The following modes are used in anick module to configure printings
     and calculations:

     Keyword                 Description
  -------------------------------------------------------------------
     TENSPOL!-PRINTMODE        Either DISTRIBUTED or SEMIDISTRIBUTED
			       keyword, meaning the mode for printing
			       tensor polynomials.

     TENSTERM!-PRINTMODE       A list of three strings which are beginning,
			       middle, and ending in printing of tensor
			       term.
     EDGE!-STRING              Equals a string which is printed between
			       edges while chain is printed. 


     EDGE!-STRING!-PRINTING    Bolean, meaning whether to use EDGE!-STRING
			       in chain printings

     BETTI		       True, when Betti numbers are to be calculated
     BETTI-FOREACHDEGREE       True, when Betti numbers are allowed to
			       be calculated and they are updated every
			       time the Groebner basis element is found.
  -------------------------------------------------------------------


			   MODULES DESCRIPTION


		BETTI NUMBERS
		(Source file: anbetti.sl)

		Procedures:

    anBETTIINIT () : - ;
          This procedure initializes variables which are used to collect
	  information about Betti Numbers. It should be called before
          Anick calculation.

    anbtNUMBEROFCHAINS (degree:int, chains:dlchn) : int ;
          This function accepts a homological degree of a chain,
	  a list of all chains, and returns the number of chains
	  of that degree in that list.

    ((anbtMEMBER (element:any, list:lany) : genint ;
          Returns the index of the element in the list.
	  The return value NIL corresponds to the absence of
	  the element in the list.))

    MEMBER2NIL (element:any, list:lany) : genint ; DESTRUCTIVE (2)
          Searches list for its first occurrence of element.
	  If found, the occurrence is (destructively) replaced by
	  NIL, and the number of the position of element in list is
	  returned. If element is not found in list, NIL is returned.

    anbtZEROPROLONG (c:lratcoeff, n:int) : lratcoeff ;
          Returns a list whose first n elements are NIL (representing
	  the coefficient zero), and which than continues with the list
	  ``tail''.

    anbtALIGN (m:llgenredandcoeff) : matrix ; DESTRUCTIVE
          This procedure destructively modifies the matrix "m" so
	  that all its lines are prolonged by zero coefficients and
	  have equal length, the length of its first input line.
	  (Consequently, no other input line may be longer than the
	  first one.) Besides aligning, to each line of the
	  matrix the INTEGERISE function is applied. The modified
	  matrix is returned.

    anbtMAKEMATRIXLINE (id:lchn, pol:tp) : lcoeff ;
          Returns a list of coefficients which are extracted from the pol.
	  Only coefficients of semi-distributed monomials where
	  corresponding qpol is equal to one are looked for.
	  The corresponding chains to those qpols are placed into
	  "id" list.  Extracted coefficients are arranged according to the
	  list "id" which is updated every time.

    anbtMAKEMATRIX (order:int, chains:lchn) : llcoeff ;
          This procedure accepts homological degree "order" and a list
	  of chains "lchn", and returns a "matrix" of coefficients of
	  all the semi-distributed tensor polynomials corresponding
	  to chains of that degree. The lines of its output "matrix"
	  may have different lengths, however.

    anBETTI (chains:lchn) : - ;
          The procedure accepts as its argument a list which is an
	  element of the anCHAINS degree list. More precisely,
	  its tail is a list of chains, but the first element equals
	  their homological degree. The procedure updates Betti
	  numbers which are storied in anCURRENTBETTINUMS for this
	  particular homological degree. anCURRENTBETTINUMS is some
	  tail of the anBETTINUMS list.

    anALLBETTI () : - ;
          This procedure applies the procedure anBETTI to all elements
	  of the anCHAINS list. It should be called if the calculation
	  of Betti numbers was not invoked during calculation of the Groebner
	  basis.

    PRINTBETTINUMBER (btnumber:btnum) : - ;
          Prints Betti number btnumber as follows:
	                 B(homological degree, total degree) = value

    PRINTBETTI () : - ;
          Prints all the calculated  Betti numbers.
          ((Applies the procedure PRINTBETTINUMBER to all elements of the
	  anBETTINUMS list.))

    PRINTCURRENTBETTI () : - ;
          Prints Betti numbers for the last calculated Groebner basis
	  degree.
          ((Applies the procedure PRINTBETTINUMBER to all elements of the
	  anCURRENTBETTINUMS list.))
          
    anbtMacCorrect (deg:int) : bool ;
          This procedure takes a degree for updating Betti numbers in
	  a nice Macaulay form so that later it is possible to print them.
	  It affects only internal btMacMx structure and return either
	  T or NIL depending on whether the update actually changed
	  btMacMx.

    anbtMacSet (order:int, degree:int, value:int) : bool ;
          Sets the value of Betti number in Macaulay form of corresponding
	  "order" and "degree" to "value". Returns boolean value T.
          
    MACBETTIPRINT () : bool ;
          Prints Betti numbers in Macaulay form. Returns boolean value T.

    anExtractBettiModes () : lmodestate ;
        Returns the list of anick settings keywords and their values
	related to the module of Betti numbers.



                USER INTERFACE
                Source file: aninterf.sl)

		Variables:

    anOutChan
           This variable contains the output channel handler for
	   printing Anick related information into the result file.

    anDIFFTIME
           After the Anick resolution is calculated it contain the time
	   consumed by calculations for differentials in milliseconds.

    anBETTITIME
           After Anick resolution is calculated it contain the time
	   consumed by calculations for Betti numbers.

    anBETTINUMS
           This variable is a list of Betti numbers. Each Betti number
	   is of the structure ((order:int . degree:int) . value:int)

    ANICKRESOLUTIONOUTPUTFILE
           Is equal to the name of output file. In case it equals to NIL,
	   the output channel will be associated with the file name
	   "andifres.txt".

    anLASTCALCULATEDDEGREE
           Equals the last degree for which the anifResolutionFixDegree
	   procedure was called.


	        Procedures:

    SETTENSPOLPRTSTRINGS (strings:lstr) : lstr ;
          The list strings should consist of 3 strings which are
	  used to print tensor product. First string is printed
	  at the beginning, second - as a tensor multiplication sign
	  and the third - as the ending of thensor monomial.
	  The list of 3 strings of old settings is returned.

    GETTENSPOLPRTSTRINGS () : lstr ;
	  Returns a list of 3 strings which are used when tensor
	  monomials are printed as beginning, middle and ending
	  strings.

    SETEDGESTRING (strng:string) : genstring ;
          The string strng is taken for printing as edges when chain
	  vertexes are printed. The old value of it is returned.

    GETEDGESTRING () : - ;
          Returns the value of string which is printed between chain
	  vertexes.

    ResolutionInit () : - ;
          This procedure initializes all the variables concerned with
	  the Anick resolution. It needs information about Groebner
	  basis generators, therefore, it should be called after the
	  system has them.

    PRINTNEWCHAINSDIFF (chan:genint) : - ;
          Prints differentials for the new generated chains
	  to the channel chan

    PRINTCHAINSDIFF (chan:genint) : - ;
          Prints differentials for the all generated chains
	  to the channel chan

    PRTCHNDIFF (inchain:chn) : - ;
          Prints differential for the chain inchain in form
	     D (chain length, chain) = differential

    CALCBETTI (flag:bool) : - ;
          Depending on whether flag is non-NIL or not, switches on/off
	  the calculation of Betti numbers. It sets corresponding flags.
	  
    ONFLYBETTI (flag:bool) : - ;
          Depending on whether flag is non-NIL or not, switches on/off
	  the calculation of Betti numbers. Also, the feature of
	  calculating and printing them while Groebner basis is being
	  calculated degree by degree. It sets corresponding flags.

    GETBETTINUMS () : lbtnum;
          Returns the list of Betti numbers.

    GETBTORDER (btnumber:btnum) : int ;
          Returns the order of Betti number btnumber.

    GETBTDEGREE (btnumber:btnum) : int ;
          Returns the degree of Betti number btnumber.

    GETBTVALUE (btnumber:btnum) : int ;
          Returns the value of Betti number btnumber.


    anifResolutionFixDegree (dgr:int) : - ;
          For a given degree it calculates chains and the Anick
	  resolution differentials, and prints them into the result
	  file. It also updates related variables.


    ResolutionFixcDegEnd () : - ;
          Applies the procedure anifResolutionFixDegree for 
	  all the degrees which has to be fixed in anifResolutionFixDegree
	  sense.

    ResolutionDone () : - ;
          Closes the output stream used in Anick related procedures.

    ResolutionClear () : - ;
          Cleans up all the Anick related variables including Betti
	  numbers.

    CLEARRESOLUTION () : - ;
          Calls ResolutionDone and ResolutionClear procedures.

    ANICKDISPLAY () : - ;
          Prints the Betti numbers into the result file.

    CALCULATEANICKRESOLUTIONTOLIMIT (limdeg:int) : - ;
          Countinues the process of Anick resolution calculation up to
	  the degree limdeg. It maybe called after the Groebner basis is
	  calculated, in case it is finite and calculations stopped
	  at a lower degree than thatonto which the Anick resolution
	  should be calculated.

    ExtractAnickModes () : lmodestate ;
	  Returns the list of modestates.

    RestoreAnickModes (modeinfo:lmodestate) : - ;
	  Restores modestates supplied  by modeinfo.


		CHAINS
		(Source file: chrecords.sl)

		Variables:

    anEDGESTRING
          This string is printed between vertexes when chain is
	  printed. Default value is " " (space).


    anCHAINS
          contains the chains in a degree list, with respect to the
	  total degree. I.e., it is a list of sub-lists, one for each
          total degree. Each sub-list starts with the total degree
	  and then lists the chains of this total degree in
          increasing order (with respect to the procedure anLESSCHAIN).

    anNEWCHAINS
          contains the current total degree (cDeg) and the chains
          of this degree, ordered increasingly.


	((For reference, a couple of friendly bergman procedures are
	  mentioned:

    (CADR RIGHTMONFACTORS) gives the nontrivial vertices of the graph,
	as augmented monomials.

    (MONLISPOUT (PMon augmon)), (MONPRINT augmon) gives the lispform
	or prints the algebraic form of its argument, respectively. ))


		Macros:
 
    anChn2LowerChn (inchain:chn) : chn ;
          Gets the left factor chain of length one less than the
	  length of inchain.

    anChn2HiChns (inchain:chn) : chn ;
          Returns a list of all as yet known chains of length one
          more than the length of inchain, such that inchain is a
          left factor chain of them.

    anChn2LastVx (chain:chn) : augmon ;
          Gives the last vertex in the graph for chain, considered
	  as a path (with augmon vertices).

    anChn2Length (chain:chn) : int ;
          Returns the length of chain (the number of vertices in
	  the path graph corresponding to chain).

    anChn2Diff (chain:chn) : gentp ;
          Returns the result of the differential for chain if it
	  is calculated, or NIL otherwise.
          (Thus a calculated differential MUST NOT be represented by
          an empty list!)

    anSetDiff (chain:chn, newdiff:tp) : - ; DESTRUCTIVE (1)
          Sets the differential of chain to newdiff.


		Procedures:

    anPROLONGCHAIN (inchain:chn, intail:augmon) : (chn . augmon) ;
          inchain should be an n-chain and intail an augmon, such that
	  there is an (n+1)-chain "outchain" which is a left factor
	  of inchain * intail. Then (outchain . outtail) is returned,
	  where outtail fulfills
          inchain * intail = outchain * outtail.
	  (Else, an error may be signaled, or the result may be
	  arbitrary.)

    PRINTCHAIN (chain:chn) : - ;
          Prints chain. The vertices are printed as monomials
          (corresponding to the present ALGOUTMODE and minor output
	  mode settings). If PRINTEDGESTRING is ON, then between
	  vertices it prints the value of anEDGESTRING.

    anCONSTRUCTNEWCHAINS (inchain:chn, degdiff:int) : - ;
          Constructing new chains from inchain and a total
	  degree difference number degdiff. The new chains
	  represent the chains of length = 1 + length[inchain] and
	  total-degree = degdiff + totaldegree[inchain], and with
	  inchain as a left factor. They are sorted into their
	  proper positions on anNEWCHAINS.

    anCONSTRUCTZEROCHAINS () : - ;
          This routine should be called before the calculation of
          Anick's resolution to construct initial data for
          differential and integral recursive functions.

    anNEWGBELM2CHAIN (gbelm:augmon) : chn ;
          Returns a 1-chain corresponding to gbelm (which ordinarily
	  would be a leading monomial of a new Groebner basis).
	  Calculates and sets the differential of the output chain.

    anLESSCHAIN (chain1:chn, chain2:chn) : bool ;
          Returns NIL if chain1 doesn't precede chain2 in "monomial
	  deg-lex" order.

    anExtractChnModes () : lmodestate ;
        Returns the list of anick settings keywords and their values
	related to the module of chain structures.



                ANICK DIFFERENTIAL
		(Source file: diffint.sl)

	        Procedures:

    anDIFF (inchain:chn) : tp ;
          The chain inchain is processed according to the
	  algorithm of calculating the differential in the Anick
	  resolution. The result semi-distributed polynomial is
	  returned.

    andiINTEG (pol:tp) : tp ; DESTRUCTIVE
	  This procedure is used in the calculation of Anick differential
	  in order to complete the algorithm for calculating the Anick
	  differential.



	        GAUSS METHOD IMPLEMENTATION
		(Source file: gauss.sl)

	        Procedures:

    MPRINT (mat:matrix) : bool ;
          Prints mat as a matrix to standard output and returns T, if
	  that is not NIL. Else it prints nothing and returns NIL.

    CFLINEPRINT (ln:lcoeff) : bool ;
          Prints the list of coefficients ln in one line and returns
	  T if ln is non-NIL. Else it prints nothing and returns NIL.

    RANK (mat:inputmatrix) : int ;
          Returns the rank of mat calculated in present coefficient
	  field.

    INPUTMATRIX2REDANDMATRIX (mtrx:inputmatrix) : matrix ; DESTRUCTIVE
          Modifies the elements of mtrx so that their type is of
	  acceptance by further procedures.

    anRANK (mat:matrix) : int ; DESTRUCTIVE
          Returns the rank of mat.
	  ((Transform mat to triangular form by applying Gauss
	  method.))



		TENSOR POLYNOMIALS
		(Source files: tenspol.sl, t_inout.sl)


		Variables:

    SObeg
        This string is printed in the beginning of a tensor monomial.

    SOmid
        This string is printed for a tensor multiplication sign.

    SOend
        This string is printed in the end of a tensor monomial.

    anPrintSDPMode
        Contains the mode which is used for printing semi-distributed
	polynomials. The values are DISTRIBUTED and SEMIDISTRIBUTED.

	        Procedures:

    PRETTYPRINTSDP (pol:tp) : - ;
        Prints the semi-distributed polynomial pol in a form decided by
	the tensor polynomials print-mode. (See GETTENSORPOLSPRINTMODE).


    anSDP!_PRETTYPRINTSDP (pol:tp) : - ;
        Prints semi-distributed polynomial pol in a semi-distributed
	form.

    anDP!_PRETTYPRINTSDP (pol:tp) : - ;
        Prints semi-distributed polynomial pol in a distributed form.


    PRINTTENSORPOLSDISTRIBUTED () : - ;
        Selects distributed form for printing semi-distributed
	polynomials.

    PRINTTENSORPOLSSEMIDISTRIBUTED () : - ;
        Selects semi-distributed form for printing semi-distributed
	polynomials.

    anExtractTPModes () : lmodestate ;
        Returns the list of anick settings keywords and their values
	related to the module of tensor structures.

    anAssignSDPMode (var:id, value:id) : - ;
        Property function for selecting printing form for tensor
	polynomials.

    anAssignTMstrings (var:ld lst:lstr) : - ;
        Property function for assigning vaules to variables used
        in printing of the tensor term.

    ---

    anCONCATTENSPOLMON (pol:tp, mon:augmon) : tp ;
	Concatenates mon to each element of the tensor polynomial pol
	to the right side, reduces them by means of the current
	Groebner basis, and returns the result.

    anMAKETENSPOL (inchain:chn, mon:augmon) : tp ;
	Creates and returns the tensor polynomial which contains only
	one element: inchain tensor multiplied by mon.

    anDESTRUCTADDTENSPOLS (pol1:tp, pol2:tp, cf:coeff):tp; DESTRUCTIVE(1)
	Modifies pol1 so that pol1 = pol1+cf*pol2.
	Returns the modified pol1.


    anEXTRACTHIGHTENSMON (pol:tp) : (chn . augmon) ;
	Returns a dotted pair which represents a highest
	element in the deg-lex sense of pol.

    anPRINTQPOLORCONSTANT (pol:qpol) : - ;
        Prints a rational coefficient cf, a *, and then within
	parentheses the polynomial multiplied by the inverse of cf.
	If the polynomial is constant, prints only it as a
	rational coefficient.

    GETTENSORPOLSPRINTMODE () : id ;
        Returns the identifier which is a keyword either DISTRIBUTED
	or SEMIDISTRIBUTED. This identifier points out how the current
	semi-distributed polynomials are printed.

## End of the document ##
