Issue with at.load_madx conversion

[elafmusa]

Hi everyone,

I am trying to use at.load_madx to convert a MADX sequence file (available here) as follows:

ring = load_madx("z_ballistic.seq", use="fccee_p_ring", energy = 45.6e9)
ring.enable_6d()
print(ring.envelope_parameters())

However, I get the error from the envelope parameters print: array must not contain infs or NaNs. Additionally, during conversion, I see the message: "MADX type Collimator is ignored and replaced by a Drift."

As an alternative, I used xconverters to convert the same file, and it worked without errors:

from xconverters import convert_lattices
xlat = convert_lattices.from_madx_seqfile('z_ballistic.seq', seq_name='fccee_p_ring', particle='electron', energy=45.6, dependencies=False)
ring = convert_lattices.to_pyat(xlat)

ring.enable_6d()
print(ring.envelope_parameters())

Frac. tunes (6D motion): [0.75791762 0.37388092 0.02895448]
Energy: 4.560000e+10 eV
Energy loss / turn: 3.908958e+07 eV
Mode emittances: [ 8.56422504e-10 -1.44757890e-37 2.18746556e-06]
Damping partition numbers: [1.00045508 1.00001076 1.99953415]
Damping times: [0.70521964 0.70553298 0.35285247] s
Energy spread: 0.000391665
Bunch length: 0.00558646 m
Cavities voltage: 79200000.0 V
Synchrotron phase: 2.62542 rd
Synchrotron frequency: 95.7474 Hz

I would like to confirm that I am using at.load_madx correctly, or if the Collimator handling is causing this issue.

Thank you in advance,
Elaf

[lfarv]

Hello @elafmusa

Thanks for reporting the problem. Could you post your z_balistic.seq file, so that I can look at it ?

[elafmusa]

Dear @lfarv

Thank you for your answer, please find the lattice file in this link:

https://syncandshare.desy.de/index.php/s/WsRXr5jqmjLynfm

[lfarv]

Thanks ! Could you also send the correctly translated lattice as a .mat file ? I’ll try to understand where the differences are with the result of load_madx…Le 2 nov. 2024 à 08:51, elafmusa ***@***.***> a écrit : Dear @lfarv Thank you for your answer, please find the lattice file in this link: https://syncandshare.desy.de/index.php/s/WsRXr5jqmjLynfm —Reply to this email directly, view it on GitHub, or unsubscribe.You are receiving this because you were mentioned.Message ID: ***@***.***>

[elafmusa]

Dear @lfarv ,

Please find the .mat file in this link:

https://syncandshare.desy.de/index.php/s/oS7rBWzXg5KZTjR

Best regards,
Elaf

[lfarv]

@elafmusa : there was indeed a problem in load_madx, which is corrected in #851.

After correction, there are still 3 small differences between the result of load_madx and your .mat file:

• the lattice in the .mat file has radiation in sextupoles. This is not the default behavior of enable_6d. Is this due to xconverters, or did you explicitly set it?
• load_madx keeps the rf frequency as it is specified in the RFCAVITY elements. This leads to an unstable machine. Setting it to the nominal frequency by calling ring.set_rf_frequency() brings it to exactly the same value as in the .mat file. Was the frequency set by xconverters ?
• xconverters adds two additional markers at the beginning and end of the lattice, load_madx does not.

Finally, after ensuring that radiating elements are the same, and that the rf frequency is the same, both lattices are identical. Here is how I got the two lattice:

seqring=at.load_madx(seqname, use="fccee_p_ring", energy = 45.6e9)
seqring.enable_6d()         # enable radiation
seqring.set_rf_frequency()  # set rf frequency to nominal

and

matring = at.load_mat(matname)[1:-1]                         # Remove the beginning and end markers
matring.enable_6d(sextupole_pass="StrMPoleSymplectic4Pass")  # remove radiation in sextupoles

Thanks for reporting the problem. I tried to use xconverters to check, but I get an error when importing convert_lattices…

[elafmusa]

Dear @lfarv,

Thank you very much!

As for the converted lattice using xconverters i run exactly the following lines:

from xconverters import convert_lattices
xlat = convert_lattices.from_madx_seqfile('z_ballistic.seq', seq_name='fccee_p_ring', particle='electron', energy=45.6, dependencies=False)
ring = convert_lattices.to_pyat(xlat)
ring.enable_6d(dipole_pass='BndMPoleSymplectic4RadPass',
            quadrupole_pass='StrMPoleSymplectic4RadPass',
            sextupole_pass='StrMPoleSymplectic4RadPass')

before saving the .mat file

ring.save('z_ballistic.mat', mat_key='ring')

[lfarv]

@elafmusa
This answers two questions:

1. xconverters automatically tunes the rf frequency to the "nominal" value, load_madx does not. Should it do it ? Up to now, creating a lattice from any source takes the data as it is, with no modification. I would prefer load_madx to stick to that rule, unless there is a general agreement to change.
2. Radiation in the sextupoles is your choice, so nothing to do…

In the mean time, I found another problem converting entrance/exit angles of bending magnets with a negative bending angle. I'm working on that…

[lfarv]

Fixed in #851