Matching with family

[wtam]

Hi Colleagues,

I have two questions:

First question:
There are two pieces of related comments in the TraceWin user's manual, quoted below. It seems to me that if I want to use match_fam_grad, there must be a command "set_twiss fn" in the code. But I have tried, it is not true. I can use match_fam_grad and set_size successfully without set_twiss. Do these comments in the manual mean something else?

p.79
Write “SET_TWISS fn” corresponds to write “SET_TWISS fn 0 0 0 0 0 0 1 1 1 1 1 1”. That disables the SET_TWISS command and allows different other optimisations by inserting another commands like for example “SET_SIZE”.

p.80
The command SET_TWISS is always obligatory to allow the matching of the gradrupole. And the option “Matching with family & Twiss commands” of page “Matching” has to be checked.

Second question:
Among all the set_xxx commands, only set_twiss supports the family number. On the other hand, all the match_fam_xxx commands support family number. How can I, for example, use only one collection of parameters (fn=1) to optimize for one set_size command and use another collection of parameters (fn=2) to optimize for another set_size command?

Best regards,
Wai-Ming

[Didier]

Dear Wai-Ming,

1- You're rigth, SET_TWISS is not mandory anymore, I'll fix manual about this point.

2- What you are asking for is not possible using this type of command because SET_SIZE commands do not have an 'fn' parameter and cannot be associated with a particular match_fam set. I strongly suggest that you stop using these commands, which are ultimately very limited, and switch to ADJUST with DIAG_XX, which is much more powerful.

Regards,

Didier

[wtam]

Dear Didier,

You gave me similar advice a long time ago when I inquired about the differences between set+match versus adjust+diag. Since then I focused on using adjust+diag until recently I revisited some examples on the manual and realized that set+match has unique capabilities such as set_size_max. It also supports weighted cost function. The command lattice+set_adv is also an effective tool to layout quadrupoles. These are examples that I found adjust+diag cannot do.

Best regards,
Wai-Ming

[Didier]

Dear Wai-Ming,

ADJUST supports also weighted cost function, hosted by DIAG_XXX elements (DIAG_XXX(weight value)....)
LATTICE and set_adv are still usuable with adjust command.
I recognize that the command set_size_max is not going to be possible with adjust, but I remain very sceptical about the real utility of such a command

In any case, I try my best to help you use the commands and procedures available, which have become more complex over time. One thing is sure, ADJUST + DIAG closer to the physical reality and more complete will continue to be developed while MACTCH_FAM_XX will only be maintained.

Regards,

Didier

[wtam]

Dear Didier,

Thank you for your comments. It is very useful.

I encourage you to teach a TraceWin based modeling course in USPAS or other accelerator schools. I am sure all of us will love it

Best regards,
Wai-Ming

[Didier]

Dear Wai-Ming,

There are TraceWin courses or rather pratical works given at the Paris-Saclay University and I have been teaching them for a few years. But honestly, development + hotline already take too much of my time. But I'm sure some training would be appreciated, I keep that in mind.

Regards,

Didier