bug in reading B3?

[pieterd]

Hi Rich,

when I try:

$:> gyre_ad < gyre_nad.in

(yes, I use gyre_ad but call my input file gyre_nad, sorry about that)

I get the following output:

gyre_ad [hg]
OpenMP Threads : 4
MPI Processors : 1

Initialization
==============

At line 164 of file gyre_b3_file.f90
Fortran runtime error: Array bound mismatch for dimension 1 of array 'delta' (0/857)



Could it be that the 'delta' array is simply not initialized? I don't see it being read from the data, and it is also not computed before use. In "gyre_mesa_file.f90", I see that the `delta` array is first computed from the input, and then used to compute nable_ad.

Attached the model and input file I used. I still see on the website that the B3 format will be superseded/removed. Should I simply switch to GSM? In the file format description, I see that delta is included.

Setup:
- Kubuntu 12.10
- Latest MADSDK with gfortran 4.9
- latest GYRE version
- test examples passed

Cheers,
Pieter

[rhtownsend]

pieterd wrote:
> Hi Rich,
>
> when I try:
>
> $:> gyre_ad < gyre_nad.in
>
> (yes, I use gyre_ad but call my input file gyre_nad, sorry about that)
>
> I get the following output:
>
> gyre_ad [hg]
> OpenMP Threads : 4
> MPI Processors : 1
>
> Initialization
> ==============
>
> At line 164 of file gyre_b3_file.f90
> Fortran runtime error: Array bound mismatch for dimension 1 of array 'delta' (0/857)
>
>
>
> Could it be that the 'delta' array is simply not initialized? I don't see it being
> read from the data, and it is also not computed before use. In "gyre_mesa_file.f90",
> I see that the `delta` array is first computed from the input, and then used to
> compute nable_ad.

Yep, this is a bug. Try reversing the lines in gyre_b3_file.fpp where nabla_ad and delta are calculated. This will of course get fixed in the next release.

>
> Attached the model and input file I used. I still see on the website that the B3
> format will be superseded/removed. Should I simply switch to GSM? In the file format
> description, I see that delta is included.
>

Long-term, that would be the better option. As you can see, GSM format is very similar to B3 (in fact, I designed GSM with you in mind!).

Thanks for the report!

cheers,

Rich

[rhtownsend]

I've just release 0.9.2, which fixes this problem.

cheers,

Rich

[pieterd]

Thanks! And thanks for the GSM format, to which I'm switching now.

Concering this GSM: I see in the documentation on

https://bitbucket.org/rhdtownsend/gyre/ ... format.pdf

1. that "nabla" is defined as dlnp/dlnT. Shouldn't this read dlnT/dlnp?
2. that in gyre_gsm_file.f90, "c_p" is no longer required but instead "nabla_ad" is required

Cheers,
Pieter

[rhtownsend]

pieterd wrote:
> Thanks! And thanks for the GSM format, to which I'm switching now.
>
> Concering this GSM: I see in the documentation on
>
> https://bitbucket.org/rhdtownsend/gyre/ ... format.pdf
>
> that "nabla" is defined as dlnp/dlnT. Shouldn't this read dlnT/dlnp?
>

Yep -- silly mistake, will fix!

cheers,

Rich

[rhtownsend]

pieterd wrote:
> Thanks! And thanks for the GSM format, to which I'm switching now.
>
> Concering this GSM: I see in the documentation on
>
> https://bitbucket.org/rhdtownsend/gyre/ ... format.pdf
>
> 1. that "nabla" is defined as dlnp/dlnT. Shouldn't this read dlnT/dlnp?
> 2. that in gyre_gsm_file.f90, "c_p" is no longer required but instead "nabla_ad"
> is required
>

Yep also, the documentation didn't get updated, apparently. Will fix also

cheers,

Rich