Re: DCD Error

From: Maxim Belkin (mbelkin_at_ks.uiuc.edu)
Date: Wed Mar 18 2015 - 10:06:34 CDT

> On Mar 18, 2015, at 9:36 AM, Josh Vermaas <vermaas2_at_illinois.edu> wrote:
>
> Hi Abhi,
>
> Did you check if the NAMD job terminated normally? Oftentimes during an abnormal termination (such as when you hit the wallclock limit), the dcd generated is too short for what the header claims. Since the header is written at the beginning of the simulation,

I don't believe this is true. I always set my simulations to run for some large number of steps and dcd files that I get have correct headers.

I think such kinda errors happen when there are troubles writing to the dcd file. Writer reports that it finished (when, in fact, it did not) and updates dcd header (space for dcd header had already been allocated so there are no problems updating it). So, this error could indicate some issues with the computer (hard drive?) that simulation ran on.

Maxim

> the dcd header will claim a framecount based on the number of steps you are planning to run. What are the filesizes for the dcds in question? Is the sum of the sizes for Production.dcd and Production_1.dcd equal to that of Production_100.dcd? Catdcd is telling you no, so if I were to bet, I'd assume that the last simulation terminated abnormally (check the logfile!).
>
> -Josh
>
> On 3/18/15 8:29 AM, Abhishek Tyagi wrote:
>> Thankyou for your reply,
>>
>> I have again checked it, I am a bit confused, could you suggest what else could be done in this
>>
>> Thanks
>>
>> ABhi
>>
>>> On 18 Mar, 2015, at 10:27 pm, btreece_at_andrew.cmu.edu wrote:
>>>
>>> Hi Abhi,
>>>
>>> After reading your output, it appears there is an issue with
>>> Production_100.dcd. The other two wrote 50001 frames and the header for
>>> Production_100.dcd claimed to have the other 50000, but only 23701 were
>>> able to be read. I know you said you checked it, but I would inspect that
>>> file more closely. Perhaps you missed something?
>>>
>>> Bradley
>>>
>>>> Dear Experts,
>>>>
>>>> I am using NAMD 2.10 on gpu cluster, initially I had performed MD for
>>>> 50ns, when I restarted from 50 to 100ns, the MD finishes successfully.
>>>> Finally, I was trying to use catdcd to merge these files in to one big dcd
>>>> file as follows:
>>>> This is first time I am experiencing this error.
>>>>
>>>>
>>>> catdcd -o full-100.dcd Production.dcd Prodution_1.dcd Production_100.dcd
>>>> CatDCD 4.0
>>>> dcdplugin) detected standard 32-bit DCD file of native endianness
>>>> dcdplugin) CHARMM format DCD file (also NAMD 2.1 and later)
>>>> Opening file 'full-100.dcd' for writing.
>>>> dcdplugin) detected standard 32-bit DCD file of native endianness
>>>> dcdplugin) CHARMM format DCD file (also NAMD 2.1 and later)
>>>> Opened file 'Production.dcd' for reading.
>>>> Read 13157 frames from file Production.dcd, wrote 13157.
>>>> dcdplugin) detected standard 32-bit DCD file of native endianness
>>>> dcdplugin) CHARMM format DCD file (also NAMD 2.1 and later)
>>>> Opened file 'Prodution_1.dcd' for reading.
>>>> Read 36844 frames from file Prodution_1.dcd, wrote 36844.
>>>> dcdplugin) detected standard 32-bit DCD file of native endianness
>>>> dcdplugin) CHARMM format DCD file (also NAMD 2.1 and later)
>>>> dcdplugin) Warning: DCD header claims 50000 frames, file size indicates
>>>> there are actually 23071 frames
>>>> Opened file 'Production_100.dcd' for reading.
>>>> Read 23071 frames from file Production_100.dcd, wrote 23071.
>>>> Total frames: 73072
>>>> Frames written: 73072
>>>> CatDCD exited normally.
>>>>
>>>>
>>>> ?What is doubtful here is the number of frames should be 100000 but these
>>>> are 73000, I checked these trajectories in vmd and i cant see any problem
>>>> their.
>>>>
>>>> Could you please tell me what is happened to this MD.
>>>>
>>>>
>>>> Thanks in advance
>>>>
>>>>
>>>> Abhi
>>>>
>>

This archive was generated by hypermail 2.1.6 : Thu Dec 31 2015 - 23:21:44 CST