[petsc-users] Floating point exception when save complex vector into a hdf5 file

[Barry Smith]

  Great! Thanks for letting us know,

   Barry


> On Mar 25, 2025, at 1:54 AM, Ye Changqing <Ye_Changqing at outlook.com> wrote:
> 
> Thanks, Barry. It works!
> 
> Best,
> Changqing
> 
> ________________________________________
> 发件人: Barry Smith <bsmith at petsc.dev>
> 发送时间: 2025年3月25日 3:18
> 收件人: Ye Changqing
> 抄送: Mark Adams; petsc-users at mcs.anl.gov
> 主题: Re: [petsc-users] Floating point exception when save complex vector into a hdf5 file
> 
> 
>   Please switch to the main git branch of PETSc and try again. Some of the code has been updated to handle the very large arrays you are working with.
> 
>   Barry
> 
> 
> 
>> On Mar 24, 2025, at 7:57 AM, Ye Changqing <Ye_Changqing at outlook.com> wrote:
>> 
>> Dear Mark,
>> 
>> As suggested, I did a fresh configuration on PETSc. The problem is still there. The attachment is the configure.log for your reference.
>> 
>> Best,
>> Changqing
>> 
>> ________________________________________
>> 发件人: Mark Adams <mfadams at lbl.gov>
>> 发送时间: 2025年3月24日 19:14
>> 收件人: Ye Changqing
>> 抄送: petsc-users at mcs.anl.gov
>> 主题: Re: [petsc-users] 回复: Floating point exception when save complex vector into a hdf5 file
>> 
>> Just to check, you want to delete the linux-oneapi-complex-opt directory and do a fresh build when you get errors like this and you might send your configure log.
>> Mark
>> 
>> On Mon, Mar 24, 2025 at 5:22 AM Ye Changqing <Ye_Changqing at outlook.com<mailto:Ye_Changqing at outlook.com>> wrote:
>> Dear all,
>> 
>> I reconfigured petsc with "--with-debugging=1". It throws more messages which I attached below.
>> 
>> Best,
>> Changqing
>> 
>> ________________________________________
>> 发件人: Ye Changqing <Ye_Changqing at outlook.com<mailto:Ye_Changqing at outlook.com>>
>> 发送时间: 2025年3月24日 16:33
>> 收件人: petsc-users at mcs.anl.gov<mailto:petsc-users at mcs.anl.gov>
>> 主题: Floating point exception when save complex vector into a hdf5 file
>> 
>> Dear PETSc developers,
>> 
>> I encountered a strange problem when I tried to save a DMDA vector into an hdf5 file, a floating point error was thrown. I can repeat the problem on the cluster. However, the same codes run fine on my local computer.
>> 
>> Below the .cxx file is the minimal working example, the .txt is the runtime error obtained from SLURM, and the .py file should tell the configure options that I used to build the library.
>> 
>> Any suggestions?
>> 
>> Best,
>> Changqing
>> 
>> 
>> <configure.log>
>