headers
Jens Nöckel | 5 Nov 2002 04:05
Favicon

Installation problem

Hi,

I just tried to compile mpb-1.4.1 on Mac OS X with Absoft's f77.
Everything looks OK, but the configure script stalls out over libctl.
I verified that I can run the example in libctl-2.2/examples
without getting an error message.
Still, mpb doesn't seem to agree that libsctl is installed properly.
I've tried recompiling libctl with various different options but
without success. Maybe there is a simple solution?
Here is what configure tells me:

checking for vendor f77 to be used instead of g77...
checking for f77... /Applications/Absoft/bin/f77
checking for Fortran 77 compiler default output... a.out
checking whether the Fortran 77 compiler works... yes
checking whether we are cross compiling... no
checking for suffix of executables...
checking for suffix of object files... o
checking whether we are using the GNU Fortran 77 compiler... yes
checking whether /Applications/Absoft/bin/f77 accepts -g... yes
checking for vendor cc to be used instead of gcc...
checking for cc... cc
checking whether we are using the GNU C compiler... yes
checking whether cc accepts -g... yes
checking for ranlib... ranlib
checking for a BSD-compatible install... /usr/bin/install -c
checking whether make sets ${MAKE}... yes
checking for sqrt in -lm... yes
checking how to get verbose linking output from 
/Applications/Absoft/bin/f77... -v
(Continue reading)

Permalink | Reply | 2 comments |
headers
Steven G. Johnson | 5 Nov 2002 04:16
Picon

Re: Installation problem

On Mon, 4 Nov 2002, Jens Nöckel wrote:
> I just tried to compile mpb-1.4.1 on Mac OS X with Absoft's f77.
> Everything looks OK, but the configure script stalls out over libctl.
> [...]
> checking for libctl dir... /usr/local/share/libctl
> checking for gen-ctl-io... yes
> checking for ctl_get_vector3 in -lctl... no
> configure: error: Couldn't find the required libctl library.

It looks like libctl is installed in /usr/local (the default).  It could
be that you need to configure with LDFLAGS=-L/usr/local/lib  but I thought
that Apple's cc (based on gcc) looked in that directory by default.

The best thing to do is to look at the config.log file and find the
compiler error message that caused the check for -lctl to fail.

Steven

_______________________________________________
mpb-discuss mailing list  -  mpb-discuss <at> ab-initio.mit.edu
http://ab-initio.mit.edu/mailman/listinfo/mpb-discuss
Permalink | Reply | 1 comment |
headers
Steven G. Johnson | 5 Nov 2002 04:23
Picon

Re: Slow convergence for linear defect

I have to admit that I've generally not used the targeted eigensolver for
line defects (waveguides), and just used the regular solver instead.  
(I'd like to try another algorithm, like the modified Jacobi-Davidson
algorithm, which is better conditioned for interior eigenvalues, i.e. the
"targeted" eigensolver.)

I imagine that things will be a lot worse if e.g. your target frequency is
not in the gap.  (Your frequency of 0.6 looks a bit high at first glance.)
It will also be worse if there aren't any guided modes at the particular
k_y you are looking at.

Steven

On Tue, 29 Oct 2002, Ben Cowan wrote:

> I'm trying to compute a guided mode of a linear air defect in a 2D array
> of air holes in silicon, using the mid-gap frequency as the target
> frequency (The .ctl file is attached below.) 
> ...
> The computation seems to converge extraordinarily slowly, as
> the "trace" output goes to 0.  (The frequency-targeted computation of a
> point defect of a 2D square lattice worked just fine, finishing after
> about 160 iterations.)  The computation behaves like this for k_x = 0 or
> 0.5.  Could another k_x solve this problem?  Could there be some sort of

_______________________________________________
mpb-discuss mailing list  -  mpb-discuss <at> ab-initio.mit.edu
http://ab-initio.mit.edu/mailman/listinfo/mpb-discuss
(Continue reading)

Permalink | Reply | 0 comments |
headers
Ahmed Sharkawy | 5 Nov 2002 18:21
Picon

kz Question

Hi Steven
       can i use MPB to analyze PC slab for non zero kz values?

 if yes how can the even and the odd modes then be separated?.

Best Regards
Ahmed Sharkawy

_______________________________________________
mpb-discuss mailing list  -  mpb-discuss <at> ab-initio.mit.edu
http://ab-initio.mit.edu/mailman/listinfo/mpb-discuss
Permalink | Reply | 1 comment |
headers
Steven G. Johnson | 5 Nov 2002 20:41
Picon

Re: kz Question

For a finite-height slab, kz is not a conserved quantum number.  (It also
breaks the even/odd symmetry, so you can't separate them any more.)

You can put a kz into the calculation for a slab, of course, but it only
makes sense in terms of the vertical supercell (it changes the relative
phase of the supercell's periodic slabs).

On Mon, 4 Nov 2002, Ahmed Sharkawy wrote:
>         can i use MPB to analyze PC slab for non zero kz values? if yes 
> how can the even and the odd modes then be separated?.

_______________________________________________
mpb-discuss mailing list  -  mpb-discuss <at> ab-initio.mit.edu
http://ab-initio.mit.edu/mailman/listinfo/mpb-discuss
Permalink | Reply | 0 comments |
headers
Jens Nöckel | 5 Nov 2002 23:35
Favicon

Re: Installation problem

Hi,

as was to be expected, the problem was extremely simple: All I had
to do was: ranlib libctl.a - no other problems occured after this.
Now I got through the installation and will start playing soon (I hope).

Regards,
		Jens

On Monday, November 4, 2002, at 07:16 PM, Steven G. Johnson wrote:

> On Mon, 4 Nov 2002, Jens Nöckel wrote:
>> I just tried to compile mpb-1.4.1 on Mac OS X with Absoft's f77.
>> Everything looks OK, but the configure script stalls out over libctl.
>> [...]
>> checking for libctl dir... /usr/local/share/libctl
>> checking for gen-ctl-io... yes
>> checking for ctl_get_vector3 in -lctl... no
>> configure: error: Couldn't find the required libctl library.
>
> It looks like libctl is installed in /usr/local (the default).  It 
> could
> be that you need to configure with LDFLAGS=-L/usr/local/lib  but I 
> thought
> that Apple's cc (based on gcc) looked in that directory by default.
>
> The best thing to do is to look at the config.log file and find the
> compiler error message that caused the check for -lctl to fail.
>
> Steven
(Continue reading)

Permalink | Reply | 0 comments |
headers
Dale Fried | 6 Nov 2002 20:13

feature request: MPB does not exit if dielectric function cannot be written to file

Hi,

I'm thinking of writing a patch to MPB so that it will not die if the
dielectric function, or any of the fields, cannot be written to a file.
This would be useful if, e.g., the disk fills up.

Is this fix already on anyone's to-do list?

Thanks for your comments.

-Dale

_______________________________________________
mpb-discuss mailing list  -  mpb-discuss <at> ab-initio.mit.edu
http://ab-initio.mit.edu/mailman/listinfo/mpb-discuss
Permalink | Reply | 1 comment |
headers
Steven G. Johnson | 7 Nov 2002 05:11
Picon

Re: feature request: MPB does not exit if dielectric function cannot be written to file

On Wed, 6 Nov 2002, Dale Fried wrote:
> I'm thinking of writing a patch to MPB so that it will not die if the
> dielectric function, or any of the fields, cannot be written to a file.
> This would be useful if, e.g., the disk fills up.

Hi Dale,

I hadn't really thought about this; usually if my disk fills up I have
bigger problems than worrying about whether MPB finishes its run.  Still,
it sounds like a reasonable improvement; feel free to send me a patch.

Steven

_______________________________________________
mpb-discuss mailing list  -  mpb-discuss <at> ab-initio.mit.edu
http://ab-initio.mit.edu/mailman/listinfo/mpb-discuss
Permalink | Reply | 0 comments |
headers
Tolliver, Todd (Research | 12 Nov 2002 23:46
Picon
Favicon

Normalized Frequency

Hello Everyone,

I have a stupid question to ask. I realize the frequency values that mpb generates are in units of
normalized frequency. However, during some simulations, like examples from the users manual, the
frequency values for some of the higher bands are greater than one. How can this be? I'm sure there
is something I'm doing wrong but I can't tell what it is. Listed below is an example ctl file that I
used and its output.

Thanks,
Todd

// begin ctl file

(set! num-bands 6)

(set! geometry-lattice (make lattice (size 1 1 no-size)
                         (basis1 (/ (sqrt 3) 2) 0.5)
                         (basis2 (/ (sqrt 3) 2) -0.5)))

(set! default-material (make dielectric (epsilon 13)))

(set! geometry (list (make cylinder
                       (center 0 0 0) (radius 0.48) (height infinity)
                       (material (make dielectric (epsilon 1))))))

(set! k-points (list (vector3 0 0 0)          ; Gamma
                     (vector3 0 0.5 0)        ; M
                     (vector3 (/ -3) (/ 3) 0) ; K
                     (vector3 0 0 0)))        ; Gamma
(set! k-points (interpolate 4 k-points))
(Continue reading)

Permalink | Reply | 1 comment |
headers
Steven G. Johnson | 13 Nov 2002 06:23
Picon

Re: Normalized Frequency

On Tue, 12 Nov 2002, Tolliver, Todd (Research) wrote:
> I have a stupid question to ask. I realize the frequency values that
> mpb generates are in units of normalized frequency. However, during
> some simulations, like examples from the users manual, the frequency
> values for some of the higher bands are greater than one. How can this
> be? I'm sure there is something I'm doing wrong but I can't tell what
> it is. Listed below is an example ctl file that I used and its output.

Todd,

There's nothing wrong---the frequencies are unbounded.  By "normalized" we
only mean that the frequency is given by a/lambda, or equivalently in
units of c/a, so that it is dimensionless.  lambda can be as small as you
want (for high bands); Mr. Maxwell imposes no restriction of lambda > a.

Steven

_______________________________________________
mpb-discuss mailing list  -  mpb-discuss <at> ab-initio.mit.edu
http://ab-initio.mit.edu/mailman/listinfo/mpb-discuss
Permalink | Reply | 0 comments |

Gmane