out of bounds for array 'phimap3'

Please post your Suggestion or report bugs here !

Moderators: zhej, lli5

Post Reply
ymei
Posts: 3
Joined: Thu Apr 06, 2017 10:06 am

out of bounds for array 'phimap3'

Post by ymei » Thu Apr 06, 2017 10:22 am

I am using DelPhi_Linux_DP_F95 to study a large protein with PDB ID 3EOJ. The external medium is water.
This whole system contains 20238 atoms and a charge of -3 au.
I got wrong surface charges after the calculation by checking the sum of array schrg.
The sum of schrg is around -15.8 au.
Then I recompiled the code with an option "-fbounds-check", and the program died with the following error message:
At line 173 of file inc_setbcmod_nitit.f95
Fortran runtime error: Array reference out of bounds for array 'phimap3', upper bound of dimension 1 exceeded

To make a dirty correction, I changed
ngp=igrid*igrid*igrid+1
to
ngp=10*igrid*igrid*igrid+1
in "qdiff4v.F95".
And the program could terminate without error messages.
However, the sum of schrg is 0.98 au, which was still not reasonable.

So I think the bug is more than the size of phimap3. Can anyone tell me how to deal with this bug?

Ye

ymei
Posts: 3
Joined: Thu Apr 06, 2017 10:06 am

Re: out of bounds for array 'phimap3'

Post by ymei » Thu Apr 06, 2017 10:33 am

In"qdiff4v.F95",
ngp=igrid*igrid*igrid+1
allocate(phimap1(nhgp),phimap2(nhgp),phimap3(ngp))
In inc_setbcmod_nitit.f95
do ix=2,icgrid+1,2
iy=ix/2 ; phimap1(iy)=phimap3(ix); phimap2(iy)=phimap3(ix+1)
end do

That is why I got this out-of-bound error.

Ye

zhej
Posts: 7
Joined: Tue Feb 07, 2017 11:18 am

Re: out of bounds for array 'phimap3'

Post by zhej » Thu Apr 06, 2017 1:01 pm

What is the grid size in parameter file? I recommend using "perfil" rather than "grid size". If the grid size is smaller than the size of the protein, it may have errors likes this.

ymei
Posts: 3
Joined: Thu Apr 06, 2017 10:06 am

Re: out of bounds for array 'phimap3'

Post by ymei » Thu Apr 06, 2017 8:31 pm

I used perfil instead of grid size. I tried perfil=75 and 80. Both failed.
I noticed there is a warning in the log file, which reads
initial cube size too small,
in assigning accessible points to a grid
therefore rescaling...
Is this warning critical?

Here is the input file
perfil= 80.0
scale= 3.000
in(pdb,file="peptide.out.pdb")
in(siz,file="delphi.siz")
in(crg,file="peptide.crg")
indi= 1.000
exdi= 80.000
prbrad= 1.400
salt= 0.100
bndcon= 2
maxc= 0.00010
linit= 1500
nonit= 0
energy(s,c,g)

zhej
Posts: 7
Joined: Tue Feb 07, 2017 11:18 am

Re: out of bounds for array 'phimap3'

Post by zhej » Fri Apr 07, 2017 9:04 am

Your parameter file looks fine. Please send all your input files to delphi@clemson.edu, so that I can have a better look at the errors.

lli5
Posts: 38
Joined: Thu Sep 22, 2011 8:44 am

Re: out of bounds for array 'phimap3'

Post by lli5 » Sat Apr 15, 2017 7:41 am

Hi Ye,

The array boundary issue is solved.
In"qdiff4v.F95",
allocate(phimap1(nhgp),phimap2(nhgp),phimap3(ngp))
This allocation of arrays is correct.

But there is an issue in
In inc_setbcmod_nitit.f95

do ix=2,icgrid+1,2
iy=ix/2 ; phimap1(iy)=phimap3(ix); phimap2(iy)=phimap3(ix+1)
end do

Here the line of "do ix=2,icgrid+1,2" need to be modified as:
do ix=2,icgrid-1,2
Then the issue is fixed.

However, the issue of induced charge is not related to this array boundary exceed problem. I checked the run of the trimer case, in the nonlinear case the relaxation parameter is calculated as a very small number. So the nonlinear case didn't really converge.

In this case we need to fix the relaxation parameter, by using the command:
relpar=1.0
and I see even after 8000 steps it doesn't converge well.
We will still work on the convergence issue.

Currently the best way for you to solve this problem is just use the linear PB. Because the net charge of your system is a very small number, the linear PB will gives you very reliable results.

Lin

indianfashion5
Posts: 1
Joined: Wed Aug 30, 2017 2:05 am
Contact:

Re: out of bounds for array 'phimap3'

Post by indianfashion5 » Wed Aug 30, 2017 2:08 am

Thanks for providing useful information . I found some really interesting stuff in your blog especially this discussion. Keep up the good work.

charlotte_mason
Posts: 6
Joined: Mon Sep 04, 2017 9:56 am
Location: Orlando

Re: out of bounds for array 'phimap3'

Post by charlotte_mason » Mon Sep 04, 2017 11:15 am

Hmm, I hope you will deal with it
My dream is to inspire and help others through my custom papers writing and one day to hopefully write and be successful with it enough to do it for a living rather than just a hobby.

Post Reply

Who is online

Users browsing this forum: No registered users and 1 guest