Skip to Content.
Sympa Menu

opal - Re: [Opal] V1.2 and V1.4

opal AT

Subject: The OPAL Discussion Forum

List archive

Re: [Opal] V1.2 and V1.4

Chronological Thread 
  • From: Nicole Neveu <nneveu AT>
  • To: "Adelmann Andreas (PSI)" <andreas.adelmann AT>
  • Cc: opal <opal AT>, "Russell, Steve" <srussell AT>
  • Subject: Re: [Opal] V1.2 and V1.4
  • Date: Wed, 18 Nov 2015 13:37:20 -0600
  • List-archive: <>
  • List-id: The OPAL Discussion Forum <>


​Regarding you cutoff question, here is some more information about the beam we are trying to create:

Transverse: Flattop with radius of 2mm
Longitudinal: Gaussian with FWHM 8ps (no flattop).

If you think there is a better way to define this distribution, it'd be great to hear your advice.
We didn't use the flattop distribution because we want a Gaussian with no flattop in the longitudinal direction.
Would it be more efficient to use a Flatttop distribution with TPULSEFWHM set to 0.0?
Would that leave only leave only the Gaussian portion?

I think TPULSEFWHM sets the traditional FWHM in the GUASSIAN distribution. 
In the FLATTOP distribution, I think TPULSEFWHM only sets the longitudinal flattop time,
and the Gaussian portion is defined by TRISE/TFALL (example on pg 119 of manual). 

Is that correct?


On Wed, Nov 18, 2015 at 12:20 PM, Nicole Neveu <nneveu AT> wrote:

You are definitely right. I should have changed the number of particles and turned on the FFT solver.
This is something I fixed in later versions of my file, sorry for the confusion. ​
So this file ran on several nodes on your cluster? I will try again with the above corrections.

As for the cutoffs, an initial beam radius of 2mm was requested, and that's what I used to set the cutoffs.
Did I set those incorrectly? This is a parameter that might be adjusted depending on the initial simulation results.



On Wed, Nov 18, 2015 at 10:58 AM, Adelmann Andreas (PSI) <andreas.adelmann AT> wrote:
Dear Nicole

In case you are running 1000 particles and no field
solver and using 16 cores does not make sense and 
is most probable the cause of your problem.

At the moment I am running 16 & 32 cores, on a 32^3 grid with 1E6 particles.
A node on your cluster has 16 cores. I copy YAGT.h5 and OTR.stat into you 
Dropbox for the 32P 1E6 simulation.

Also I note, you are using a Gauss distribution with a CUTOFF in X and Y which
is very tight. As a consequence, it takes a long time to generate the distribution. 
Is that necessary? 

ECollimator works (in 1.4) and yes you can overlap 
collimators with other elements.
Here the modified distribution:
OPAL{0}> * ************* D I S T R I B U T I O N ********************************************
OPAL{0}> * 
OPAL{0}> * Number of particles: 1000000
OPAL{0}> * 
OPAL{0}> * Distribution type: GAUSS
OPAL{0}> * 
OPAL{0}> * Sigma Time Rise               = 3.40265e-12 [sec]
OPAL{0}> * TPULSEFWHM                    = 8e-12 [sec]
OPAL{0}> * Sigma Time Fall               = 3.40265e-12 [sec]
OPAL{0}> * Longitudinal cutoff           = 4 [units of Sigma Time]
OPAL{0}> * Flat top modulation amplitude = 0 [Percent of distribution amplitude]
OPAL{0}> * Flat top modulation periods   = 0
OPAL{0}> * SIGMAX                        = 0.002 [m]
OPAL{0}> * SIGMAY                        = 0.002 [m]
OPAL{0}> * SIGMAPX                       = 0 [Beta Gamma]
OPAL{0}> * SIGMAPY                       = 0 [Beta Gamma]
OPAL{0}> * CORRX                         = 0
OPAL{0}> * CORRY                         = 0
OPAL{0}> * CUTOFFX                       = 1 [units of SIGMAX]
OPAL{0}> * CUTOFFY                       = 1 [units of SIGMAY]
OPAL{0}> * CUTOFFPX                      = 0 [units of SIGMAPX]
OPAL{0}> * CUTOFFPY                      = 0 [units of SIGMAPY]
OPAL{0}> * 
OPAL{0}> * Number of energy bins    = 9
OPAL{0}> * Distribution is emitted. 
OPAL{0}> * Emission time            = 2.72085e-11 [sec]
OPAL{0}> * Time per bin             = 3.02317e-12 [sec]
OPAL{0}> * Delta t during emission  = 5.49668e-14 [sec]
OPAL{0}> * 
OPAL{0}> * ------------- THERMAL EMITTANCE MODEL --------------------------------------------
OPAL{0}> *  Cathode work function     = 3.2 [eV]  
OPAL{0}> *  Cathode Fermi energy      = 3.2 [eV]  
OPAL{0}> *  Cathode temperature       = 321.95 [K]  
OPAL{0}> *  Photocathode laser energy = 5 [eV]  
OPAL{0}> * ----------------------------------------------------------------------------------
OPAL{0}> * 
OPAL{0}> * 
OPAL{0}> * **********************************************************************************
OPAL{0}> * ************* B E A M ************************************************************ 
OPAL{0}> * BEAM        BEAM1
OPAL{0}> * CURRENT     1.3 A
OPAL{0}> * FREQUENCY   1.3e+09 Hz
OPAL{0}> * CHARGE      -e * 1 
OPAL{0}> * REST MASS   0.000510999 GeV
OPAL{0}> * MOMENTUM    1.01094e-06
OPAL{0}> * NPART       1e+06
OPAL{0}> * ********************************************************************************** 
OPAL{0}> * ************* F I E L D S O L V E R ********************************************** 
OPAL{0}> * TYPE         FFT
OPAL{0}> * MX           32
OPAL{0}> * MY           32
OPAL{0}> * MT           32
OPAL{0}> * BBOXINCR     1
OPAL{0}> * XDIM         parallel  
OPAL{0}> * YDIM         parallel  
OPAL{0}> * Z(T)DIM      serial  
OPAL{0}> * ********************************************************************************** 
OPAL{0}> Track start at: 15:54:06, t= 0; zstop at: 3 [m]
ParallelTTracker {0}> Change dT from 1e-13 [s] to 5.49668e-14 [s] during emission 
OPAL{0}> Executing ParallelTTracker, initial DT 1e-13 [s];
OPAL{0}> max integration steps 190000, next step= 0
OPAL{0}> Using default (Boris-Buneman) integrator

Please let me know if that makes sense.

With the best regards 

Dr. sc. math. Andreas (Andy) Adelmann
Staff Scientist
Paul Scherrer Institut WBBA/219 CH-5232 Villigen PSI
Phone Office: xx41 56 310 42 33 Fax: xx41 56 310 31 91
Phone Home: xx41 62 891 91 44
Friday: ETH CAB H 85.1  +41 44 632 75 22
The more exotic, the more abstract the knowledge, 
the more profound will be its consequences.
Leon Lederman 

On 17 Nov 2015, at 09:01, Andreas Adelmann <andreas.adelmann AT> wrote:

Hello Nicole 

Hello OPAL experts,
I've been using OPAL for the past few months and things are going well. Thanks again for your help with my thermionic gun simulations when I first started. I am currently running V1.2, but plan to switch to V1.4 when it is released. I have a few questions about the differences, ect.

Yes 1.4 is a bit late :( but we are close :-)
1. Does OPAL scale magnetic field maps to the max B value in the file, or max Bz on axis? Pg. 183 of the manual says: "the input field map will be normalized so the peak field magnitude value is equal to ...1 T in the case of magnetic field maps."  I think I see this behavior in my simulations, but appendix C (pg. 203) says scaling is done using max Bz on axis values. Is this a feature that is changed from V1.2 to V1.4?

The behaviour is as described on P. 183 and the description on P 203 is corrected in the new version of the manual.

@Steve: can you check please 

2. Should I use the R or E collimator? When we were emailing about the thermionic gun, it was suggested that I use the R collimator. The manual says E is supported in OPAL-T (pg.98). Is this difference an artifact of using V1.2? 

In 1.4 you can use both, indeed there was a problem in 1.2.

3. I have been running on a cluster at ANL, and I've been able to run on multiple processors, but not on multiple nodes. I get segmentation faults (11). Is this an artifact of using V1.2 that will be fixed in V1.4, or is it something I am doing wrong?  The cluster uses Torque PBS manager and Maui PBS scheduler, and my run commands look like the following:

mpirun -np 16 opal --info 0 | tee OTR.out

That should work. Are you using the pre-build binaries for the closure runs? 
If you send me the input file I can have a look and check on our cluster.
Also tell me where do you want to have the E-collimator so I can try. 
Do we have a Dropbox?

Cheers Andreas 

Archive powered by MHonArc 2.6.19.

Top of Page