CWIGGLER generates vertical dispersion in helical mode

[Skamarokha]

It looks like CWIGGLER generates vertical dispersion if magnetic field is non-zero in both plains, POLE1_FACTOR = 0.25, POLE2_FACTOR = 0.75, SINUSOIDAL = 1, HELICAL = 1, FORCE_MATCHED=1, STEPS_PER_PERIOD=12 or more.

Also there is a mistake in CWIGGLER description - INTEGRATION_ORDER - Integration order (2, 4, or 6), but it only can be 2 or 4.

And I also suspect alpha_y mismatch from planar horizontal CWIGGLER. Changing STEPS_PER_PERIOD doesn't help much, in some cases it may even make things worse - increasing STEPS_PER_PERIOD (following the rule 4*n) can generate dispersion and worsen beta beatings.

[Skamarokha]

When CWIGGLER installed in center of symmetry there is asymmetry in alpha function behavior (FORCE_MATCH=1). Element WIGGLER behaves normally.

[michael_borland]

I'm not able to replicate these issues. Can you post some files that demonstrate the problems?

Thanks--Michael

[Skamarokha]

In archive there are 3 cases, one folder for each case. "result" - files with zero wiggler field, "result_" - files with settings for a case. In each folder there are plots of optical functions and difference of these functions in comparison to the case with turned off wiggler.

Case 1 - straight section with WIGGLER. This case represents how optical functions should behave. Alpha function behaves antisymmetrical in relation to the center of straight section, sum of alphay in any marker from left side with corresponding symmetrical marker from the right side is about 1e-15 with wiggler being off and on. Horizontal dispersion is zero.

Case 2 - straight section with CWIGGLER, SINUSOIDAL. Here we can observe sum of alphay at points of symmetry being about 1e-9 - not much, but the amplitude magnetic field is low as well. We can also observe small horizontal dispertion.

Case 3 - straight section with CWIGGLER, SINUSOIDAL, HELICAL. If we look at CWIGGLER-helical.png file we will see asymmetrical behavior of vertical dispersion. There is simillar asymmetry in horizontal dispersion, but value is smaller - 100 nm.

[michael_borland]

Mikhail,

Thanks for the detailed examples. The issues with the HELICAL=0 case can be solved by some fine-tuning. In particular, for CWIGGLER devices the transport matrix is determined through particle tracking. That has limited accuracy. You can improve the results for case 2 by adding this command at the top of the file:

Code: Select all

&global_settings
        tracking_matrix_step_factor = 0.25
&end

For case 3, there seems to be a bug, but I haven't quite tracked it down. More to come.

--Michael

[Skamarokha]

Thanks! I will try that.

[michael_borland]

Mikhail,

Regarding the issue with helical devices, it is not so much a bug as an unavoidable limitation. In a planar device, if we want a wiggler to generate zero orbit offset and zero dispersion, we need to start and end with half-length poles. The field is assumed to have the peak value at the start and end of the device. For a helical device, we can't do that simultaneously in both planes. The vertical plane has to have a delay of 90 deg at the start. To make an integer number of periods, we have to end that plane 270 deg early. That leads to an asymmetry that is reflected in the horizontal plane.

There are some possible workarounds that I'm looking into. However, the ultimate solution is to get the fields for the real magnet and determine the harmonics, rather than using SINUSOIDAL=1.

--Michael