Can elegant perform dynamic Aperture optimization ?

[csun]

Hi Weiming Guo,

Thanks for your example. In the example, you are using dnux/dp and dnuy/dp to target the desired chromaticity. I am just curious why you don't use the linear chromaticity driving terms h11001 and h00111? I noticed that in the twiss parameter output file, the driving terms h11001 and h0011 are not close to zero even though the dnux/dp and dnuy/dp are already corrected to zero. My understanding is that those driving terms are related to the linear chromaticity by a pi factor, i.e. dnux/dp = h11001/pi. Here is the partial ouput of the twiss parameter file :
-------------------------------------------------------------------------------------------------------------------------------------------------------------------------
Step = 0 nux (1/(2$gp$r)) = 2.145948e+01 dnux/dp (1/(2$gp$r)) = 1.165833e-01
dnux/dp2 (1/(2$gp$r)) = -2.433317e+02 dnux/dp3 (1/(2$gp$r)) = -9.040604e+03 Ax (m) = 0.000000e+00
nuy (1/(2$gp$r)) = 8.196170e+00 dnuy/dp (1/(2$gp$r)) = -5.312905e-01 dnuy/dp2 (1/(2$gp$r)) = -2.594767e+02
dnuy/dp3 (1/(2$gp$r)) = 1.617910e+03 Ay (m) = 0.000000e+00 deltaHalfRange = 0.000000e+00
h11001 = 2.719532e+01 h00111 = 2.444541e+01 h20001 = 5.295725e-01
h00201 = 2.676596e+00 h10002 (1/m$a1/2$n) = 5.098531e-01 h21000 (1/m$a1/2$n) = 3.118103e+01
h30000 (1/m$a1/2$n) = 1.152045e+01 h10110 (1/m$a1/2$n) = 6.389664e+01 h10020 (1/m$a1/2$n) = 2.782511e+00
h10200 (1/m) = 9.747550e+00 h22000 (1/m) = 4.764831e+01 h11110 (1/m) = 2.478521e+02
h00220 (1/m) = 2.567794e+02 h31000 (1/m) = 1.122561e+01 h40000 (1/m) = 5.612807e+00
h20110 (1/m) = 8.477368e+01 h11200 (1/m) = 2.769764e+01 h20020 (1/m) = 1.117740e+01
--------------------------------------------------------------------------------------------------------------------------------------------------------------------------
Are those driving terms not precisely calculated in elegant? Thanks!

csun

[wguo]

Hi Sun,

I need your lattice file before I can tell if the results are correct.

Regards,

Weiming

[csun]

Attached are the files I use to generate twiss parameters. Thanks!

csun

[wguo]

Hi Sun,

I don't see any problem with the input files.

For linear chromaticity we use dnux/dp and dnuy/dp instead of h11001 and h00111 is because the h coefficients are calculated only at the quadrupole and sextupole locations (1/rho^2 is not counted either); therefore the result is not as precise as that obtained from twiss_command. The other quadrupole related terms have the similar approximation. If the magnets are long, one could set element_division to >1 to obtain better precision.

Regards,

Weiming

[michael_borland]

I plan to release the genetic optimization software for dynamic aperture and local momentum aperture in the near future. I post something when it is available.

--Michael

[csun]

Thanks, Weiming and Michael. I am looking forward to the release of the genetic optimization.

CSun

[Hwang, Ji-Gwang]

Hello, Dear all.

I'm also interested in optimization of dynamic aperture.
I trying to run the example. But It is not operate in windows OS.[It is desktop, Win7]
This problem is samely occur in my linux PC.[It is desktop, Ubuntu9]
I get some error message, such as "couldn't execute "csub": no such file or directory."
So, I delete the "csub" in scanTune files. but still it has a error, such as "unable to open K2 in mode r: no such file or directory."


If someone know that how to fix this error, please let me know.

Sincerely yours.

Hwang, Ji-Gwang

[michael_borland]

Ji-Gwang,

If you are using the files I uploaded for the tune scan and don't have the "csub" command, just edit the script to replace "exec csub elegant..." with "exec elegant..." It should work, but will be slow since there's no parallel execution of jobs.

--Michael

[wguo]

You can also try the elegant input file attached in my previous email. It doesn't need any shell commands.


Weiming