cp2k geometry optimization energy not decreasing

[Raghav]

Dear cp2k users,

                        I am doing geometry optimization of a configuration but the ENERGY is stuck at a value and is not decreasing and the calculation is running like this and not getting finished. Am I doing something wrong?? Need help.

 *******************************************************************************
 ***                 BRENT   - NUMBER OF ENERGY EVALUATIONS :       1        ***
 *******************************************************************************

 --------  Informations at step =   253 ------------
  Optimization Method        =                   SD
  Total Energy               =      -361.6413293878
  Real energy change         =         0.0000000001
  Decrease in energy         =                   NO
  Used time                  =              185.503

  Convergence check :
  Max. step size             =         0.0000000000
  Conv. limit for step size  =         0.0030000000
  Convergence in step size   =                  YES
  RMS step size              =         0.0000000000
  Conv. limit for RMS step   =         0.0015000000
  Convergence in RMS step    =                  YES
  Max. gradient              =         0.0006486901
  Conv. limit for gradients  =         0.0004500000
  Conv. for gradients        =                   NO
  RMS gradient               =         0.0001373171
  Conv. limit for RMS grad.  =         0.0003000000
  Conv. in RMS gradients     =                  YES
 ---------------------------------------------------

With regards

Raghav

[Matt W]

Nothing is badly wrong at a quick look. But there are a some things in your input that could be improved. 

I think the most important thing is to improve your forces  - use a larger CUTOFF value, this might be helpful. 

Use a better minimizer, especially if you restart from near the minima already found - I'd suggest removing all the GEO_OPT options to use the default BFGS optimizer.

Just for efficiency, you don't need such a large box, I don't think. 5 angstrom away from any atom should be enough.

Matt 

[Raghav]

Thanks your suggestions worked