Synopsis
		sfiwilbfgs < in.rsf > out.rsf source=source.rsf data=data.rsf wdso=wdso.rsf wstk=wstk.rsf precon=precon.rsf grad=grad.rsf verb=y deriv=y nh=0 load=n uts=0 npml=10 beta=0. grect1=5 grect2=5 gscale=-1. nhess=6 miter=10 mline=5 epsilon=1.e-7 lower=1.5 upper=7.5 order=

Parameters
		float beta=0. stacking power cost function   string data= auxiliary input file name   bool deriv=y [y/n] if y, apply derivative in z   float epsilon=1.e-7 L-BFGS termination epsilon   string grad= auxiliary output file name   int grect1=5 gradient smoothing radius on axis 1   int grect2=5 gradient smoothing radius on axis 2   float gscale=-1. gradient re-scale (enabled if (0,1))   bool load=n [y/n] load LU   float lower=1.5 lower bound of feasible set   int miter=10 L-BFGS maximum # of iterations   int mline=5 L-BFGS maximum # of line search   int nh=0 horizontal space-lag   int nhess=6 L-BFGS # of Hessian corrections   int npml=10 PML width   string order= discretization scheme (default optimal 9-point)   string precon= auxiliary input file name   string source= auxiliary input file name   float upper=7.5 upper bound of feasible set   int uts=0 number of OMP threads   bool verb=y [y/n] verbosity flag   string wdso= auxiliary input file name   string wstk= auxiliary input file name