[gmx-users] Steepest Descents converged to machine precision

[Mark Abraham]

nitu sharma wrote:
> Dear Mark
> 
> I explain you what I  want to ask . I give you a detail of my processing -
> * step for packing lipid around protein-
> 1. concatanation of protein and lipid bilayer in proper orientation- I did.
> 2.Inflate the bilayer by using inflategro script with scaling factor 4 
> and cutoff 14- I did .
>  there is 8 lipid removed in this process and a/c to that   I have 
> update my topology file .
> 3. Energy minimisation by applying strong position restrain on protein 
> the force constant=100000
> 
> problem occur here is-
> 
> Step=  344, Dmax= 1.8e-05 nm, Epot=  4.29315e+06 Fmax= 1.02525e+05, 
> atom= 4615
> Step=  349, Dmax= 1.4e-06 nm, Epot=  4.29315e+06 Fmax= 9.86524e+04, 
> atom= 4615
> Stepsize too small, or no change in energy.
> Converged to machine precision,
> but not to the requested precision Fmax < 1000
> 
> Double precision normally gives you higher accuracy.
> 
> writing lowest energy coordinates.
> 
> Steepest Descents converged to machine precision in 350 steps,
> but did not reach the requested Fmax < 1000.
> Potential Energy  =  4.2931510e+06
> Maximum force     =  1.0252509e+05 on atom 4615
> Norm of force     =  1.3668822e+03

These are so large or positive that your starting structure is grossly 
broken. You need to find out why. Look at the structures at each point 
of your procedure.

Mark