Probably the easiest workaround is to use the field-integration code, which
works fine in MPI (unlike get-field-pt). Integrating over a volume of size 0
is equivalent to just evaluating the function at a point (since zero-size
dimensions in the integration volume are just interpolated, not integrated).
In fact, you could define a drop-in-replacement for get-field-pt as follows:
(define (my-get-field-pt c pt)
(integrate-field-function (list c) (lambda (p f) f) (volume
(center pt))))
The only caveat is that c must be a component (Ez, Hx, etcetera), and not a
derived-component (Sx, EnergyDensity, etcetera); if you want the latter, you
must compute them manually from the field components.
You could also output the fields at a point to an HDF5 file (using in-volume
with a size of 0).
(The whole get-field-pt code in Meep is basically a legacy of the early
implementation; I should probably rip it out and replace it with something that
calls the equivalent of the integration code automatically.)
On Jul 4, 2012, at 4:21 PM, Eric CASSAN wrote:
> Dear Steven,
> thanks one more time for the MEEP software.
>
> I would like using the "get-field-point" tool in the MPI version of MEEP.
>
> I then obtain the following message:
> CHECK failure on line 583 of ../../../mpb-ctl/fields.c: get-*-point not yet
> implemented for MPI!
> [0] MPI Abort by user Aborting program !
> [0] Aborting program!
> p0_22008: p4_error: : 1
> p0_22008: (24562.839989) net_send: could not write to fd=4, errno = 32
>
> Is there any solution to circumvent the fact that get-field-point seems not
> to be implemented in MPI-MEEP?
> My study indeed needs simulation in 3D and I heavily use "get-field-point",
> which slows down a lot the program.
>
> Thank you for any help!
>
> With best regards,
> Eric Cassan
>
>
>
> ===================================================
> Eric CASSAN
> http://silicon-photonics.ief.u-psud.fr/
> Institut d'Electronique Fondamentale (UMR CNRS 8622)
> Bâtiment 220 de l'U.F.R. Sciences
> Université PARIS-SUD (PSud)
> 91405, Orsay, France
> Phone : + 33 1 69 15 78 52
> ===================================================
>
_______________________________________________
meep-discuss mailing list
[email protected]
http://ab-initio.mit.edu/cgi-bin/mailman/listinfo/meep-discuss
