Problem with the "dilatmx" in the structure optimization

[lxbyf]

Dear all,

My structure optimizations always stopped in the same step, and the warning was:

---SELF-CONSISTENT-FIELD CONVERGENCE--------------------------------------------

chkdilatmx: ERROR -
The new primitive vectors rprimd (an evolving quantity)
are too large with respect to the old rprimd and the accompanying dilatmx :
this large change of unit cell parameters is not allowed by the present value of dilatmx.
You need at least dilatmx= 5.194561E+00
Action : increase the input variable dilatmx.

I think the suggestion value "5.194561E+00" is too large, isn't it? So,is anyone can give me some suggestion? Thank you!

The structure is 2*2*1 AlGaN with 16 atoms, and the calculation will not start unless I set the "chkprim=0", is it right?
and the sturcture is :

acell 3*3.189 angstrom

rprim 0.0000000000000000 -1.0000000000000000 0.0000000000000000
0.8660254037844386 0.5000000000000000 0.0000000000000000
0.0000000000000000 0.0000000000000000 1.6250000000000000

xred 0.666666667 0.333333333 0
0.333333333 0.666666667 0.5
0.166666667 0.333333333 0
0.166666667 0.833333333 0
0.666666667 0.833333333 0
0.333333333 0.166666667 0.5
0.833333333 0.166666667 0.5
0.833333333 0.666666667 0.5
0.166666667 0.333333333 0.375
0.666666667 0.333333333 0.375
0.166666667 0.833333333 0.375
0.666666667 0.833333333 0.375
0.333333333 0.166666667 0.875
0.833333333 0.166666667 0.875
0.333333333 0.666666667 0.875
0.833333333 0.666666667 0.875

[david.waroquiers]

Hello

I dont know what your structure is but
1) why do you want a non primitive supercell (2*2*1) ?
2) if you really need a 2*2*1, you acell and rprim look strange to me, so if you could send your input file with the primitive cell so that we can compare...

David

[lxbyf]

I am sorry , it is my fault.
I forgot to change the primitive vectors, when I used the 2*2 supercell.
The primitive vectors was still used for a 1*1 supercell.

Thanks a lot!