The CASINO forum

Dear CASINO users,

I get the following error when I perform vmc_opt run:

==================================================
PERFORMING VARIANCE MINIMIZATION CALCULATION No. 1
==================================================


ERROR : COPY_CONFIG_FILE
Problem reading FIDET for config 1.

CASINO internal traceback:
Problem detected at COPY_CONFIG_FILE
Called from SETUP
Called from MAIN

It works fine when I make single VMC run, but I couldn't figure out why this occurs when I make optimization run. Here I uploaded input and output files, but I can also give other input files if necessary.

Thanks!

Here is a correction that I should make to myself. I realized that I forgot to subtract the electrons that are in the pseudo potential, that is why I get the gwfdet_setup warning. However, the final error is still the same. Just to be sure, I also increased vmc_nstep and vmc_nconfig_write, but it haven't changed the error. Here are my new input and output files, thanks!

Here is the answer to my own question with current beta, v2.13.415, it works, but I would appreciate if anyone is willing to answer why such an occurred in the first place, to make sure that I am doing everything correct.

Hi Kayahan,

Sorry for the slow response - as you may have seen, the email notifications have not been working (I think they are since about 5 minutes ago - but let's see).

Are you using the Intel version 13.x Fortran compiler? If so, that contains a hideous bug which occurs if (deep breath) you have an 'exit' or 'cycle' statement in a 'select case' block which is inside a do loop. That would be perfectly legitimate Fortran, of course - so this is just a stupid compiler bug. However, that particular construct did appear in the 'copy_config_file' routine in CASINO, hence the problem.

By extensive trial and error which nearly bored me to death last year I discovered the error could be bypassed - for no readily apparent reason - by using the deprecated 'goto' statement to escape from the do loop, instead of the more modern 'exit'. That fix was published in CASINO 2.13.38 - shortly after the 12.1 version that you were using.

If you weren't using the Intel compiler, then I don't know what the problem is, but as you have seen it can be fixed by moving to the current beta. We officially recommend moving from 2.12 to the current beta anyway, as it is more or less what will be in the forthcoming 2.14 public release - the delay in publication is only because I haven't been able to stop people asking me to do things recently (sorry).

Cheers,
Mike

Hi Mike,

I am using 13.3 intel fortran compiler, indeed. Good to hear that it is not something that can affect the results. Thanks for the feedback. I should have switched to current beta already.

Best,
Kayahan