An interesting problem in a tune footprint of FMA

[gwei]

Hi michael,

A result of the tune footprint of FMA is shown in fma2216.png. The tune of the lattice is (.22, .16), But the top in the tune footprint plot is not.
the top in the tune footprint is like (.204, .163).

Another result plot is made by LEGO．　

I changed several number parameters in the ele file, but there is not change of the strange top value of (.204, .163) in the tune footprint.

[gwei]

And I checked the .twi file also.
the nux and nuy are .22 and .16.

[michael_borland]

Can you provide your .lte and .ele files? Otherwise, it is impossible to suggest a solution.

--Michael

[gwei]

HI Michael

ele file was attached already as the first attached file. 'da_fma.ele'
I modified my question topics

[michael_borland]

The problem is more likely to be in the .lte file. You can email it to me (borland at aps.anl.gov) if you don't want to post it.

Things to check are a sufficient number of kicks (N_KICKS parameter) on KQUAD and CSBEND elements.

--michael

[gwei]

The problem is more likely to be in the .lte file. You can email it to me (borland at aps.anl.gov) if you don't want to post it.

Things to check are a sufficient number of kicks (N_KICKS parameter) on KQUAD and CSBEND elements.

--michael

Thank you. I already email to you.

[michael_borland]

Guohui,

Thanks for sending the lattice. One problem is too few kicks for the quadrupoles. Some of your quadrupoles are quite long and not weak, so the lattice functions can change considerably inside them. This means you need more kicks to get a good model.

I got tunes of 0.219 and 0.160 by increasing the number of kicks by a factor of 2 on all quadrupoles. You may also try increasing this only for quads with larger K1*L.

You should also vary the value of N_KICKS on the CSBEND elements to ensure things have converged sufficiently.

--Michael

[gwei]

Guohui,

Thanks for sending the lattice. One problem is too few kicks for the quadrupoles. Some of your quadrupoles are quite long and not weak, so the lattice functions can change considerably inside them. This means you need more kicks to get a good model.

I got tunes of 0.219 and 0.160 by increasing the number of kicks by a factor of 2 on all quadrupoles. You may also try increasing this only for quads with larger K1*L.

You should also vary the value of N_KICKS on the CSBEND elements to ensure things have converged sufficiently.

--Michael

Thank you very much.