source: MML/trunk/at/atphysics/findsyncorbit.m @ 5

function [orbit, varargout] = findsyncorbit(RING,dCT,varargin);
%FINDSYNCORBIT finds closed orbit, synchronous with the RF cavity 
% and momentum deviation dP (first 5 components of the phase space vector)
% by numerically solving  for a fixed point 
% of the one turn map M calculated with LINEPASS
%
%       (X, PX, Y, PY, dP2, CT2 ) = M (X, PX, Y, PY, dP1, CT1)
%    
%    under constraints CT2 - CT1 =  dCT = C(1/Frev - 1/Frev0) and dP2 = dP1 , where 
%    Frev0 = Frf0/HarmNumber is the design revolution frequency
%    Frev  = (Frf0 + dFrf)/HarmNumber is the imposed revolution frequency
%
% IMPORTANT!!!  FINDSYNCORBIT imposes a constraint (CT2 - CT1) and
%    dP2 = dP1 but no constraint on the value of dP1, dP2
%    The algorithm assumes time-independent fixed-momentum ring 
%    to reduce the dimensionality of the problem.
%    To impose this artificial constraint in FINDSYNCORBIT 
%    PassMethod used for any element SHOULD NOT 
%    1. change the longitudinal momentum dP (cavities , magnets with radiation)
%    2. have any time dependence (localized impedance, fast kickers etc).
%
%
% FINDSYNCORBIT(RING,dCT) is 5x1 vector - fixed point at the 
%               entrance of the 1-st element of the RING (x,px,y,py,dP)
%
% FINDSYNCORBIT(RING,dCT,REFPTS) is 5-by-Length(REFPTS)
%     array of column vectors - fixed points (x,px,y,py,dP)
%     at the entrance of each element indexed in REFPTS array. 
%     REFPTS is an array of increasing indexes that  select elements 
%     from the range 1 to length(RING)+1. 
%     See further explanation of REFPTS in the 'help' for FINDSPOS
%
% FINDSYNCORBIT(RING,dCT,REFPTS,GUESS) - same as above but the search
%     for the fixed point starts at the initial condition GUESS
%     Otherwise the search starts from [0 0 0 0 0 0]'.
%     GUESS must be a 6-by-1 vector;
%
% [ORBIT, FIXEDPOINT] = FINDSYNCORBIT( ... )
%     The optional second return parameter is 
%     a 6-by-1 vector of initial conditions 
%     on the close orbit at the entrance to the RING.  
%
% See also FINDORBIT, FINDORBIT4, FINDORBIT6.
%
if ~iscell(RING)
   error('First argument must be a cell array');
end

d = 1e-9;       % step size for numerical differentiation
max_iterations = 20;
J = zeros(5);

if nargin==4
    if isnumeric(varargin{2}) & all(eq(size(varargin{2}),[6,1]))
       Ri=varargin{2};
   else
       error('The last argument GUESS must be a 6-by-1 vector');
   end
else
    Ri = zeros(6,1);
end

D5 = d*eye(5);  
%B5 = zeros(5,5);
RMATi = zeros(6,6);
theta5 = [0 0 0 0  dCT]';



%Fist iteration
RMATi= Ri*ones(1,6) + [D5 zeros(5,1); zeros(1,6)];
RMATf = linepass(RING,RMATi);
Rf = RMATf(:,6);
% compute the transverse part of the Jacobian 
J5 =  [RMATf([1:4,6],1:5)-RMATf([1:4,6],6)*ones(1,5)]/d;

% Replace matrix inversion with \
%B5 = inv(diag([1 1 1 1 0]) - J5);
% Ri_next = Ri +  [B5* (Rf([1:4,6])-[Ri(1:4);0]-theta5) ; 0];
Ri_next = Ri +  [(diag([1 1 1 1 0]) - J5)\(Rf([1:4,6])-[Ri(1:4);0]-theta5) ; 0];
change = norm(Ri_next - Ri);
Ri = Ri_next;

itercount = 1;


while (change>eps) & (itercount < max_iterations)
   
   RMATi= Ri*ones(1,6) + [D5 zeros(5,1); zeros(1,6)];    
   RMATf = linepass(RING,RMATi,'reuse');

   Rf = RMATf(:,6);
   % compute the transverse part of the Jacobian 
   J5 =  [RMATf([1:4,6],1:5)-RMATf([1:4,6],6)*ones(1,5)]/d;
   % Replace matrix inversion with \
   %B5 = inv(diag([1 1 1 1 0]) - J5);
   %Ri_next = Ri +  [B5*(Rf([1:4,6])-[Ri(1:4);0]-theta5); 0];
   Ri_next = Ri +  [(diag([1 1 1 1 0]) - J5)\(Rf([1:4,6])-[Ri(1:4);0]-theta5); 0];
   change = norm(Ri_next - Ri);
   Ri = Ri_next;
   itercount = itercount+1;
   
end;


if(nargin<3) | (varargin{1}==(length(RING)+1))
    % return only the fixed point at the entrance of RING{1}
    orbit=Ri(1:5,1);
else            % 3-rd input argument - vector of reference points along the Ring
                % is supplied - return orbit            
   orb6 = linepass(RING,Ri,varargin{1},'reuse'); 
   orbit = orb6(1:5,:); 
end

if nargout==2
    varargout{1}=Ri;
end