atwiggler implementation

[marlibgin2]

Hi I am trying to introduce an ID into a bare lattice without ... IDs. I have decided to use atwiggler, however every attempt to create a "stable" lattice seems to fail. This is the element I am introducing, if I interpret correctly the structure
it is 1m long, with an undulator period of 2cm, and a peak field of 0T. I have zeroed the field to simplify the case
IDh = atwiggler('IDh', 50*0.02, 0.02, 0, 3.0e9, 'GWigSymplecticPass');
I build an achromat as
achrRF_ID = cellcat( IDh, LS_ID, GS,MC1,GE, SS, ...
GS,UC1,GE, ...
GS,UC2,GE, ...
GS,UC3,GE, ...
GS,UC4,GE, ...
GS,UC5,GE, ShortStr_RF, ...
GS,flip(MC2),GE, ...
flip(LS_ID), IDh );
In the .m file that generates the achromat there is a calculation of the main parameters:
[RD, ~] = atlinopt(ring,0,1:numel(ring)+1);

Here I get a plethora of errors

ACHR_ID = m4U_250903_i01_04_01_01__ID
LENGTH CORRECTED
Loading MAX IV 3 GeV ring lattice from file: m4U_250903_i01_04_01_01__ID
Error using atpass
Required field 'R1' was not found in the element data structure
Error in linepass (line 90)
[Rout,lossinfo] = atpass(line,Rin,newlattice,1,refpts, ...

I am kind of lost.
If this does not ring a bell at all ... can I beg someone to pass me an example of a real lattice with an atwiggler element in the structure?
Thanks for your attention
Marco

[swhite2401]

@marlibgin2 I can provide a python example but I suppose this is not what you are looking for... maybe @oscarxblanco could help?

[marlibgin2]

Thank you very much Simon for the prompt reply, as you say I would prefer a matlab case to understand what I am doing wrong, but I may learn something from pyAT as well.
Marco

[oscarxblanco]

Hi @marlibgin2 ,
I don't see an error on the wiggler itself. I copied the line you wrote and there are no issues. And doing tracking with element pass, elempass(IDh,[0;10e-6;0;0;0;0],energy=3e9) also works.

However, about the ring construction I couldn't find any function in matlab called cellcat, so, I wouldn't be sure what is it doing. The error returned by atpass seems to come from a problem on the ring itself but you will need to check it is a 1D cell array (I mean an Nx1 matrix in matlab), and that all cells are AT elements.

I leave you here below a FODO example (not using cellcat), let me know if it works. The purpose of this example was checking any tunes difference between a FODO with wiggler set to Bmax=0, and without that wiggler.

clear;
%% FODO
the_length = 0.4;
QF = {atmultipole("QF", 0.5, [0 0],[0,1.2])};
drift_length = 0.5;
Dr = {atdrift("Dr", drift_length)};
HalfDr = {atdrift("Dr2", drift_length/2)};
Dr3 = {atdrift("Dr3", 0.00)};
Dr4 = {atdrift("Dr4", the_length)}; %0.4
QD = {atquadrupole("QD", 0.5, -1.2)};
Bend = {atsbend("Bend", 1, 2 * pi / 20)};

RP = {atringparam('fodo',1e9)};
FODOcell = [ ...
    RP, HalfDr, Bend, Dr, QF, Dr, Bend, Dr, QD, Dr3, Dr4 ...
    ];
FODOcell = atsetenergy(FODOcell,1e9);
%% get tune
[t,c] = tunechrom(FODOcell);
%% Wiggler
und = atwiggler('und',Dr4{1}.Length,Dr4{1}.Length/100,0,1000);
%% replace one element
uFODOcell = FODOcell;
uFODOcell{length(FODOcell)} = und;
%% get tune
[t1,c1] = tunechrom(uFODOcell);
%% get tune difference
dd = t1-t;

I would also suggest to update AT matlab because there was a missing step in the integrator which is solved in #1013.

[marlibgin2]

Thank you Oscar, indeed you are right cellcat is an internal function we have made here at max iv, so non-standard

function catArray = cellcat(varargin)

N = 0;
for i = 1:numel(varargin)
N = numel(varargin{i}) + N;
end
catArray = cell(1,N);

N = 1;
for i = 1:numel(varargin)
if ~iscell(varargin{i})
catArray{N} = varargin{i};
N = N+1;
else
for j = 1:numel(varargin{i})
catArray{N} = varargin{i}{j};
N = N+1;
end
end
end

I have tried also another route, to define the R1, R2, T1, T2 fields in the IDh element, since it seemed to complain about their non existence. Result: Matlab crashed!

So stepped back and tried to run your FODO case

clear;
%% FODO
the_length = 0.4;
QF = {atmultipole("QF", 0.5, [0 0],[0,1.2])};
drift_length = 0.5;
Dr = {atdrift("Dr", drift_length)};
HalfDr = {atdrift("Dr2", drift_length/2)};
Dr3 = {atdrift("Dr3", 0.00)};
Dr4 = {atdrift("Dr4", the_length)}; %0.4
QD = {atquadrupole("QD", 0.5, -1.2)};
Bend = {atsbend("Bend", 1, 2 * pi / 20)};

RP = {atringparam('fodo',1e9)};
FODOcell = [ ...
RP, HalfDr, Bend, Dr, QF, Dr, Bend, Dr, QD, Dr3, Dr4 ...
];
FODOcell = atsetenergy(FODOcell,1e9);
%% get tune
[t,c] = tunechrom(FODOcell);
%% Wiggler
und = atwiggler('und',Dr4{1}.Length,Dr4{1}.Length/100,0,1000);
%% replace one element
uFODOcell = FODOcell;
uFODOcell{length(FODOcell)} = und;
%% get tune
[t1,c1] = tunechrom(uFODOcell);
%% get tune difference
dd = t1-t;
**_**Error using atpass
Required field 'R1' was not found in the element data structure

Error in linepass (line 90)
[Rout,lossinfo] = atpass(line,Rin,newlattice,1,refpts, ...

Error in xorbit_dp (line 21)
RMATf = linepass(ring,RMATi,args{:});

Error in findorbit/xfindorbit (line 53)
orb0=xorbit_dp(ring,dp,varargs{:});

Error in frequency_control (line 31)
[varargout{1:nargout}]=func(ring,varargs{:},'is_6d',is_6d);

Error in findorbit (line 33)
[orbs,orbitin]=frequency_control(@xfindorbit,ring,varargin{:});

Error in tunechrom/xtunechrom (line 58)
[~,orbitin]=findorbit(ring,dpargs{:},varargs{:});

Error in frequency_control (line 31)
[varargout{1:nargout}]=func(ring,varargs{:},'is_6d',is_6d);

Error in tunechrom (line 43)
[varargout{1:nargout}]=frequency_control(@xtunechrom,ring,allargs{:});_

Is there something dodgy with my general implementation of AT ?
investigating

[marlibgin2]

note: und = atwiggler('und',Dr4{1}.Length,Dr4{1}.Length/100,0,1000);
shouldn't the last term be 1e9? atwiggler help gives me "energy in eV". I tried this but fails in the same way for me

[oscarxblanco]

@marlibgin2 , with respect to the energy, it is unrelated to the issue, but you are right ! Energy should be higher.

[oscarxblanco]

@marlibgin2 , if you don't mind to have a double copy of AT matlab and in order to avoid unnecessary modifications on your system I would suggest to clone again the project (which is not heavy), do the installation (cd atmat; atpath; atmexall) and try it out. It should give you no errors.

I don't think this is related to the properties R1,R2,T1,T2. It seems to me there is an incongruity between the compiled executable and the matlab ring construction, which we should be able to trace back in case there are backwards incompatibilities.

If you prefer not to clone the project again. I would suggest to take note on the current git commit you are, then pull the newest master, and compile it with atmex.

I have used a lot of git jargon in this message, let me know if it is clear. Otherwise I could describe it in more detail.

[marlibgin2]

I think it is a sensible thing to do Oscar, I would try it and let you know. Git jargon is fine. Thank you

[marlibgin2]

Hola Oscar, I followed your advice and cloned the latest at-master, ported my lattice generator and it seems that atwiggler works with no errors. So, yes, we need to check what is going on in our environment, maybe we need to update the code, re-compile also. But it is comforting. Thank you so much for your help and time!

Marco