> # Excerpt from a MUG posting
> #
> # Steve Newby <newby@voicenet.com> wrote:
> #
> # In my work, I often have to solve a series of equations and then refer
> # back to those solutions later in the worksheet (ie those solutions
> # become inputs into other equations). The procedure works just fine.
> # 
> # However, when the worksheet is saved and then reevaluated a few days
> # later, there is no way to guarantee that Maple outputs those solutions
> # in the same ORDER. I am then left with no alternative but to sift
> # through through the worksheet, editing as necessary, to make sure that
> # the references still work.
> #
> #From: Ivan Huerta <ihuerta@mat.puc.cl>
> #Subject: Constant Irritant
> #To: maple-list@daisy.uwaterloo.ca
> #Date: Tue, 14 May 96 13:20:16 INVCL
> #
> #Hi:  
> #
> #Some alternatives ..
> #
> #Define a  lexorder comparison of expressions of the type:  var = value
> #according to var
> #
> compare_lhs:= proc(a,b) ;
>     lexorder( lhs(a), lhs(b) ); 
> end;
> #
> #then use instead
> #
> sort_solve:= proc(set_eqs,set_vars) ;
>        sort( [op(solve(set_eqs,set_vars))], compare_lhs);
>   end;
> #  
> #The function sort_solve gives the solutions in a list sorted by variable name,
> #which is always the same.
> #
> #Another possibility is to index solutions by variable name 
> #by means of a table. This solution allows to easily 
> #"loop" over all values of a variable var for different solution sets. 
> #
> table_solve:= proc(set_eqs,set_vars)
>           local sol,i,t;
>           sol:= solve(set_eqs,set_vars);
>           for i from 1 to nops(sol)   do t[lhs(sol[i])]:=rhs( sol[i]) od;
>            eval(t);
>            end;
> #           
> #If
> # 
>  sol:=table_solve(set_eqs,set_vars);
> #
> #then 
> #
> sol[var];
> #
> #gives the value of var in the solution set sol
> #
> #Of course one needs to do some argument checking to make these routines more 
> #robust.
> #
> #Examples:
> #
> eq1:= xx+ y + Zz = 1;
> eq2 :=xx -y + Zz = 1;
> eq3 :=xx +y + 2*Zz = 0;
> set_eqs := {eq1,eq2,eq3};
> set_vars := {xx,y,Zz};
> solve(set_eqs,set_vars);
> #
> #                           {y = 0, Zz = -1, xx = 2}
> #
> sort_solve(set_eqs,set_vars); #result is independent of run
> #
> #                           [Zz = -1, xx = 2, y = 0]
> #
> sol:=table_solve( set_eqs,set_vars);                    
> #
> #                             sol := table([
> #                                         xx = 2
> #                                         Zz = -1
> #                                         y = 0
> #                                     ])
> #
> sol[y];
> #
> #                                      0
> #
> sol[xx];
> #
> #                                      2
> #
> #Hope this helps, 
> #
> #Saludos,
> #
> #Ivan Huerta
> #Facultad de Matematicas
> #Pontificia Universidad Catolica de Chile
> #
> #=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-
> #
> #Date: Tue, 14 May 1996 12:05:50 -0700 (PDT)
> #From: Jim Gunson <jimg@Kwantlen.BC.CA>
> #To: maple-list@daisy.uwaterloo.ca, newby@voicenet.com
> #Subject: Constant Irritant
> #
> #I have shared your frustration. In addition I wish that solve would 
> #accept lists of equations and variables, instead of insisting on sets. 
> #
> #In working with system of equations in linear algebra, I have students 
> #use "genmatrix", "gausselim" and "backsub". These give a solution in 
> #vector form so that if the variables are specified as [x,y], a solution 
> #[1,2] means x=1, y =2. If your equations are always linear. I commend 
> #this approach.
> #
> #An alternative is the code below, which will reorder the solutions into a 
> #LIST in the order of the variables. This works only for single solutions. 
> #If the solution may be a set, then the use of "map" will allow the 
> #reordering procedure to be applied to each solution.
> # 
> #getinorder takes a solution set sols and list of variables and sort
> #the solutions in the order given by that of vars.
>  getinorder:= proc(sols,vars)
>    local novars,newsol,var,sol;
>    novars:=nops(vars);
>    newsol:=op({});  
>    for var in vars do
>      for sol in sols do
>        if op(1,sol)=var 
>          then newsol:=newsol,var=op(2,sol):fi;
>      od;
>    od;
>    RETURN([newsol]):
>   end:
> # 
> #solinoder solves a system sys for a list of variables vars, and
> #returns a list of solutions.
> #
> solinorder:=proc(sys,vars)local sols;
>    sols:=getinorder(solve(sys,{op(vars)}),vars);
> end;
> #
> solinorder := proc(sys, vars)
> local sols;
>    sols := getinorder(solve(sys, {op(vars)}), vars)
> end
> #
> eq1:=2*x+3*y=2;eq2:=x-y=3;
> solinorder({eq1,eq2},[x,y]);
> #
> #                        eq1 := 2 x + 3 y = 2
> #
> #                          eq2 := x - y = 3
> #
> #                        [x = 11/5, y = -4/5]
> #
> #-----------------------------------------------------------------------------
> #Name:  Jim Gunson                         | email: jimg@kwantlen.bc.ca
> #Title: Mathematics Instructor             | phone: (604) 599-2201
> #Kwantlen University College, BC, Canada   | fax:   (604) 599-2279