@Mariusz Iwaniuk 

Thanks for your reply. But it gives no answer for

int(exp(I*k*x)*sech(x)^3,x=-infinity..infinity)

or for

int(exp(I*k*x)*tanh(x)*sech(x)^3,x=-infinity..infinity)

thanks again

k is a real parameter.

@Carl Love 
Thanks for your helps, but this is what I have received:

table(sparce,[1 = -phi(t, x, u[]) + diff(phi(t, x, u[]), t) + diff(phi(t, x, u[]), u[])*u[] - (diff(tau(t, x, u[]), u[])*u[] + diff(tau(t, x, u[]), t))*u[] - (-diff(tau(t, x, u[]), x)*u[] + diff(phi(t, x, u[]), x))*x + diff(diff(tau(t, x, u[]), x), x)*u[] - diff(diff(phi(t, x, u[]), x), x), u[2]*u[2, 2] = 2*diff(tau(t, x, u[]), u[])*x + 2*diff(diff(tau(t, x, u[]), x), u[]) + 2*diff(xi(t, x, u[]), u[]), u[2] = -xi(t, x, u[]) + diff(phi(t, x, u[]), u[])*x - (diff(tau(t, x, u[]), u[])*u[] + diff(tau(t, x, u[]), t))*x - diff(tau(t, x, u[]), u[])*x*u[] - diff(xi(t, x, u[]), t) - diff(xi(t, x, u[]), u[])*u[] - (-diff(tau(t, x, u[]), x)*x - diff(tau(t, x, u[]), u[])*u[] - diff(xi(t, x, u[]), x) + diff(phi(t, x, u[]), u[]))*x + 2*diff(diff(tau(t, x, u[]), x), u[])*u[] + diff(diff(tau(t, x, u[]), x), x)*x + diff(diff(xi(t, x, u[]), x), x) - 2*diff(diff(phi(t, x, u[]), x), u[]) + 4*diff(tau(t, x, u[]), x), u[2, 2] = 3*diff(tau(t, x, u[]), x)*x - diff(tau(t, x, u[]), u[])*u[] + diff(diff(tau(t, x, u[]), x), x) + 2*diff(xi(t, x, u[]), x) - diff(tau(t, x, u[]), t), u[2]^3 = diff(diff(tau(t, x, u[]), u[]), u[])*x + diff(diff(xi(t, x, u[]), u[]), u[]), u[2]*u[2, 2, 2] = 2*diff(tau(t, x, u[]), u[]), u[2, 2, 2] = 2*diff(tau(t, x, u[]), x), u[2]^2*u[2, 2] = diff(diff(tau(t, x, u[]), u[]), u[]), u[2]^2 = -diff(tau(t, x, u[]), u[])*x^2 - diff(xi(t, x, u[]), u[])*x - (-diff(tau(t, x, u[]), u[])*x - diff(xi(t, x, u[]), u[]))*x + diff(diff(tau(t, x, u[]), u[]), u[])*u[] + 2*diff(diff(tau(t, x, u[]), x), u[])*x + 2*diff(diff(xi(t, x, u[]), x), u[]) - diff(diff(phi(t, x, u[]), u[]), u[]) + 4*diff(tau(t, x, u[]), u[])]

How can I exploid the coefficients only in one vector or set?
thanks

@Rouben Rostamian

Hi again.

But it seems that the above precedure does not work for f given below:

restart;
f:=(diff(diff(tau(t, x, u[]), u[]), u[])*x + diff(diff(xi(t, x, u[]), u[]), u[]))*u[2]^3 + diff(diff(tau(t, x, u[]), u[]), u[])*u[2]^2*u[2, 2] + (-diff(tau(t, x, u[]), u[])*x^2 - diff(xi(t, x, u[]), u[])*x - (-diff(tau(t, x, u[]), u[])*x - diff(xi(t, x, u[]), u[]))*x + diff(diff(tau(t, x, u[]), u[]), u[])*u[] + 2*diff(diff(tau(t, x, u[]), x), u[])*x + 2*diff(diff(xi(t, x, u[]), x), u[]) - diff(diff(phi(t, x, u[]), u[]), u[]) + 4*diff(tau(t, x, u[]), u[]))*u[2]^2 + (2*diff(tau(t, x, u[]), u[])*x + 2*diff(diff(tau(t, x, u[]), x), u[]) + 2*diff(xi(t, x, u[]), u[]))*u[2]*u[2, 2] + (-xi(t, x, u[]) + diff(phi(t, x, u[]), u[])*x - (diff(tau(t, x, u[]), u[])*u[] + diff(tau(t, x, u[]), t))*x - diff(tau(t, x, u[]), u[])*x*u[] - diff(xi(t, x, u[]), t) - diff(xi(t, x, u[]), u[])*u[] - (-diff(tau(t, x, u[]), x)*x - diff(tau(t, x, u[]), u[])*u[] - diff(xi(t, x, u[]), x) + diff(phi(t, x, u[]), u[]))*x + 2*diff(diff(tau(t, x, u[]), x), u[])*u[] + diff(diff(tau(t, x, u[]), x), x)*x + diff(diff(xi(t, x, u[]), x), x) - 2*diff(diff(phi(t, x, u[]), x), u[]) + 4*diff(tau(t, x, u[]), x))*u[2] + 2*diff(tau(t, x, u[]), u[])*u[2, 2, 2]*u[2] + (3*diff(tau(t, x, u[]), x)*x - diff(tau(t, x, u[]), u[])*u[] + diff(diff(tau(t, x, u[]), x), x) + 2*diff(xi(t, x, u[]), x) - diff(tau(t, x, u[]), t))*u[2, 2] - phi(t, x, u[]) + diff(phi(t, x, u[]), t) + diff(phi(t, x, u[]), u[])*u[] - (diff(tau(t, x, u[]), u[])*u[] + diff(tau(t, x, u[]), t))*u[] - (-diff(tau(t, x, u[]), x)*u[] + diff(phi(t, x, u[]), x))*x + diff(diff(tau(t, x, u[]), x), x)*u[] + 2*diff(tau(t, x, u[]), x)*u[2, 2, 2] - diff(diff(phi(t, x, u[]), x), x):
collect(f,{u[2],u[2,2],u[2,2,2]},distributed):
op([1..-1], %):
L := eval([%], [u[2]=1,u[2,2]=1,u[2,2,2]=1]);

because the constant terms respect to u[2], u[2,2] and u[2,2,2] should be stored in one place not in different elements of L.

@Rouben Rostamian  Yes. This is exactly I needed. Thanks again
Sayed