[FLASH-USERS] Bug in Ionize module?

[Suoqing Ji]

Hi Klaus,

Thanks for your detailed response!

In addition, I was able to solve Eqi by directly calling Ionize_equil.F90 (as what’s done in Simulation_initBlock.F90 in NeiTest problem).

Best wishes,
--
Suoqing JI
Ph.D Candidate
Department of Physics
University of California, Santa Barbara
http://web.physics.ucsb.edu/~suoqing

> On Oct 18, 2016, at 2:47 PM, Klaus Weide <klaus at flash.uchicago.edu> wrote:
> 
> On Mon, 17 Oct 2016, Suoqing Ji wrote:
> 
>> Hi,
>> 
>> I’m looking at the Ionize module and suspect there might be a bug in IonizeMain/Eqi/nei.F90 (perhaps this file is out of date). I think the line
>> 
>> xout(jj2) = xout2(i)/selnel
>> 
>> should be
>> 
>> xout(jj2) = xout2(i) * selnel(jj2)
>> 
>> Even from the view of code grammar, selnel is defined as a single real number, but it refers to an array ion_xfrac when nei.F90 is called from Ioinze.F90. 
> 
> There are two versions of the file nei.F90 in the Ionize unit:
> 
>  IonizeMain/Eqi/nei.F90
>  IonizeMain/Nei/nei.F90
> 
> Each contains an implementation of
> 
>  subroutine neimn
> 
> but their interfaces are incompatible with each other. In particular, 
> compare declarations for the last dummy argument:
> 
>  IonizeMain/Nei/nei.F90:  real, intent(in) :: xfrac(*)     ! an array
>  IonizeMain/Eqi/nei.F90:  real,    intent(in) ::  selnel   ! a scalar
> 
> The declaration in IonizeMain/Nei matches the call in Ionize.F90, but the 
> one in IonizeMain/Eqi doesn't (as you have found). 
> 
> The combination of IonizeMain/Ionize.F90 and IonizeMain/Eqi/nei.F90 does 
> not result in a valid Fortran program; but compilers will not usually 
> detect this because the code does not use an explicit interface for neimn.
> 
> We are running regular regression tests using the NeiTest simulation, but 
> those only use the Nei variant.
> 
> It appears that IonizeMain/Eqi/nei.F90 is severely out of date;
> it also appears that two incompatible versions of nei.F90 were already 
> part of FLASH version 2.  So it seems the directory IonizeMain/Eqi is 
> mostly of historical value, and the Users Guide is wrong as far as it 
> gives the impression that Eqi is a viable alternative to Eqi.
> 
> I am not going to discuss under which physical conditions the approach in 
> the Ionize unit is applicable / valid / useful; I hope somebody else can 
> clarify if needed, and answer your question:
> 
>> And I have an additional question: if I understood this correctly, the 
>> electron population fraction is set to constant in the code (by electron 
>> to proton ratio). But even if we can ignore the contribution of metals, 
>> helium might not be fully ionized and may make a significant change to 
>> electron density. Then why is it valid that electron density remain 
>> unchanged throughout ionization?
> 
> 
> Klaus