/*
 * hsfft.c
 *
 *  Created on: Apr 14, 2013
 *      Author: Rafat Hussain
 */


#include "hsfft.h"

fft_object fft_init(int N, int sgn) {
	fft_object obj = NULL;
	// Change N/2 to N-1 for longvector case

	int twi_len,ct,out;
	out = dividebyN(N);

	if (out == 1) {
		obj = (fft_object) malloc (sizeof(struct fft_set) + sizeof(fft_data)* (N-1));
		obj->lf = factors(N,obj->factors);
		longvectorN(obj->twiddle,N,obj->factors,obj->lf);
		twi_len = N;
		obj->lt = 0;
	} else {
		int K,M;
		K = (int) pow(2.0,ceil(log10(N)/log10(2.0)));

		if (K < 2 * N - 2) {
			M = K * 2;
		} else {
			M = K;
		}
		obj = (fft_object) malloc (sizeof(struct fft_set) + sizeof(fft_data)* (M-1));
		obj->lf = factors(M,obj->factors);
		longvectorN(obj->twiddle,M,obj->factors,obj->lf);
		obj->lt = 1;
		twi_len = M;
	}


	obj->N = N;
	obj->sgn = sgn;

	if (sgn == -1) {
		for(ct = 0; ct < twi_len;ct++) {
			(obj->twiddle+ct)->im = - (obj->twiddle+ct)->im;

		}
	}

	return obj;
}



static void mixed_radix_dit_rec(fft_data *op,fft_data *ip,const fft_object obj, int sgn, int N,int l,int inc) {

	int radix,m,ll;
	if (N > 1) {
		radix = obj->factors[inc];
		//printf("%d \n",radix);
	}

	if (N == 1) {

		op[0].re = ip[0].re;
		op[0].im = ip[0].im;

	} else if (N == 2) {
		fft_type tau1r,tau1i;
		op[0].re = ip[0].re;
		op[0].im = ip[0].im;

		op[1].re = ip[l].re;
		op[1].im = ip[l].im;

		tau1r = op[0].re;
		tau1i = op[0].im;


		op[0].re = tau1r + op[1].re;
		op[0].im = tau1i + op[1].im;

		op[1].re = tau1r - op[1].re;
		op[1].im = tau1i - op[1].im;

	} else if (N == 3) {
		fft_type tau0r,tau0i,tau1r,tau1i,tau2r,tau2i;
		op[0].re = ip[0].re;
		op[0].im = ip[0].im;

		op[1].re = ip[l].re;
		op[1].im = ip[l].im;

		op[2].re = ip[2*l].re;
		op[2].im = ip[2*l].im;

		tau0r = op[1].re + op[2].re;
		tau0i = op[1].im + op[2].im;

		tau1r = sgn * 0.86602540378 * (op[1].re - op[2].re);
		tau1i = sgn * 0.86602540378 * (op[1].im - op[2].im);

		tau2r = op[0].re - tau0r * 0.5000000000;
		tau2i = op[0].im - tau0i * 0.5000000000;

		op[0].re = tau0r + op[0].re ;
		op[0].im = tau0i + op[0].im;

		op[1].re = tau2r + tau1i;
		op[1].im = tau2i - tau1r;

		op[2].re = tau2r - tau1i;
		op[2].im = tau2i + tau1r;

		return;


	} else if (N == 4) {
		fft_type tau0r,tau0i,tau1r,tau1i,tau2r,tau2i,tau3r,tau3i;
		op[0].re = ip[0].re;
		op[0].im = ip[0].im;

		op[1].re = ip[l].re;
		op[1].im = ip[l].im;

		op[2].re = ip[2*l].re;
		op[2].im = ip[2*l].im;

		op[3].re = ip[3*l].re;
		op[3].im = ip[3*l].im;

		tau0r = op[0].re + op[2].re;
		tau0i = op[0].im + op[2].im;

		tau1r = op[0].re - op[2].re;
		tau1i = op[0].im - op[2].im;

		tau2r = op[1].re + op[3].re;
		tau2i = op[1].im + op[3].im;

		tau3r = sgn* (op[1].re - op[3].re);
		tau3i = sgn* (op[1].im - op[3].im);

		op[0].re = tau0r + tau2r ;
		op[0].im = tau0i + tau2i;

		op[1].re = tau1r + tau3i;
		op[1].im = tau1i - tau3r;

		op[2].re = tau0r - tau2r;
		op[2].im = tau0i - tau2i;

		op[3].re = tau1r - tau3i;
		op[3].im = tau1i + tau3r;



	} else if (N == 5) {
		fft_type tau0r,tau0i,tau1r,tau1i,tau2r,tau2i,tau3r,tau3i,tau4r,tau4i,tau5r,tau5i,tau6r,tau6i;
		fft_type c1,c2,s1,s2;
		op[0].re = ip[0].re;
		op[0].im = ip[0].im;

		op[1].re = ip[l].re;
		op[1].im = ip[l].im;

		op[2].re = ip[2*l].re;
		op[2].im = ip[2*l].im;

		op[3].re = ip[3*l].re;
		op[3].im = ip[3*l].im;

		op[4].re = ip[4*l].re;
		op[4].im = ip[4*l].im;

		c1 = 0.30901699437;
		c2 = -0.80901699437;
		s1 = 0.95105651629;
		s2 = 0.58778525229;

		tau0r = op[1].re + op[4].re;
		tau2r = op[1].re - op[4].re;
		tau0i = op[1].im + op[4].im;
		tau2i = op[1].im - op[4].im;

		tau1r = op[2].re + op[3].re;
		tau3r = op[2].re - op[3].re;
		tau1i = op[2].im + op[3].im;
		tau3i = op[2].im - op[3].im;


		tau4r = c1 * tau0r + c2 * tau1r;
		tau4i = c1 * tau0i + c2 * tau1i;

		//tau5r = sgn * ( s1 * tau2r + s2 * tau3r);
		//tau5i = sgn * ( s1 * tau2i + s2 * tau3i);

		if (sgn == 1) {
			tau5r = s1 * tau2r + s2 * tau3r;
			tau5i = s1 * tau2i + s2 * tau3i;

		} else {
			tau5r = -s1 * tau2r - s2 * tau3r;
			tau5i = -s1 * tau2i - s2 * tau3i;

		}

		tau6r = op[0].re + tau4r;
		tau6i = op[0].im + tau4i;

		op[1].re = tau6r + tau5i;
		op[1].im = tau6i - tau5r;

		op[4].re = tau6r - tau5i;
		op[4].im = tau6i + tau5r;

		tau4r = c2 * tau0r + c1 * tau1r;
		tau4i = c2 * tau0i + c1 * tau1i;

		//tau5r = sgn * ( s2 * tau2r - s1 * tau3r);
		//tau5i = sgn * ( s2 * tau2i - s1 * tau3i);

		if (sgn == 1) {
			tau5r = s2 * tau2r - s1 * tau3r;
			tau5i = s2 * tau2i - s1 * tau3i;

		} else {
			tau5r = -s2 * tau2r + s1 * tau3r;
			tau5i = -s2 * tau2i + s1 * tau3i;

		}

		tau6r = op[0].re + tau4r;
		tau6i = op[0].im + tau4i;

		op[2].re = tau6r + tau5i;
		op[2].im = tau6i - tau5r;

		op[3].re = tau6r - tau5i;
		op[3].im = tau6i + tau5r;



		op[0].re += tau0r + tau1r;
		op[0].im += tau0i + tau1i;


	} else if (N == 7) {
		fft_type tau0r,tau0i,tau1r,tau1i,tau2r,tau2i,tau3r,tau3i,tau4r,tau4i,tau5r,tau5i,tau6r,tau6i,tau7r,tau7i;
		fft_type c1,c2,c3,s1,s2,s3;
		op[0].re = ip[0].re;
		op[0].im = ip[0].im;

		op[1].re = ip[l].re;
		op[1].im = ip[l].im;

		op[2].re = ip[2*l].re;
		op[2].im = ip[2*l].im;

		op[3].re = ip[3*l].re;
		op[3].im = ip[3*l].im;

		op[4].re = ip[4*l].re;
		op[4].im = ip[4*l].im;

		op[5].re = ip[5*l].re;
		op[5].im = ip[5*l].im;

		op[6].re = ip[6*l].re;
		op[6].im = ip[6*l].im;

		c1 = 0.62348980185;
		c2 = -0.22252093395;
		c3 = -0.9009688679;
		s1 = 0.78183148246;
		s2 = 0.97492791218;
		s3 = 0.43388373911;

		tau0r = op[1].re + op[6].re;
		tau3r = op[1].re - op[6].re;

		tau0i = op[1].im + op[6].im;
		tau3i = op[1].im - op[6].im;

		tau1r = op[2].re + op[5].re;
		tau4r = op[2].re - op[5].re;

		tau1i = op[2].im + op[5].im;
		tau4i = op[2].im - op[5].im;

		tau2r = op[3].re + op[4].re;
		tau5r = op[3].re - op[4].re;

		tau2i = op[3].im + op[4].im;
		tau5i = op[3].im - op[4].im;


		tau6r = op[0].re + c1 * tau0r + c2 * tau1r + c3 * tau2r;
		tau6i = op[0].im + c1 * tau0i + c2 * tau1i + c3 * tau2i;

		//tau7r = sgn * ( -s1 * tau3r - s2 * tau4r - s3 * tau5r);
		//tau7i = sgn * ( -s1 * tau3i - s2 * tau4i - s3 * tau5i);

		if (sgn == 1) {
			tau7r = -s1 * tau3r - s2 * tau4r - s3 * tau5r;
			tau7i = -s1 * tau3i - s2 * tau4i - s3 * tau5i;

		} else {
			tau7r = s1 * tau3r + s2 * tau4r + s3 * tau5r;
			tau7i = s1 * tau3i + s2 * tau4i + s3 * tau5i;

		}


		op[1].re = tau6r - tau7i;
		op[6].re = tau6r + tau7i;

		op[1].im = tau6i + tau7r;
		op[6].im = tau6i - tau7r;

		tau6r = op[0].re + c2 * tau0r + c3 * tau1r + c1 * tau2r;
		tau6i = op[0].im + c2 * tau0i + c3 * tau1i + c1 * tau2i;

		//tau7r = sgn * ( -s2 * tau3r + s3 * tau4r + s1 * tau5r);
		//tau7i = sgn * ( -s2 * tau3i + s3 * tau4i + s1 * tau5i);

		if (sgn == 1) {
			tau7r = -s2 * tau3r + s3 * tau4r + s1 * tau5r;
			tau7i = -s2 * tau3i + s3 * tau4i + s1 * tau5i;
		} else {
			tau7r = s2 * tau3r - s3 * tau4r - s1 * tau5r;
			tau7i = s2 * tau3i - s3 * tau4i - s1 * tau5i;

		}


		op[2].re = tau6r - tau7i;
		op[5].re = tau6r + tau7i;
		op[2].im = tau6i + tau7r;
		op[5].im = tau6i - tau7r;

		tau6r = op[0].re + c3 * tau0r + c1 * tau1r + c2 * tau2r;
		tau6i = op[0].im + c3 * tau0i + c1 * tau1i + c2 * tau2i;

		//tau7r = sgn * ( -s3 * tau3r + s1 * tau4r - s2 * tau5r);
		//tau7i = sgn * ( -s3 * tau3i + s1 * tau4i - s2 * tau5i);

		if (sgn == 1) {
			tau7r = -s3 * tau3r + s1 * tau4r - s2 * tau5r;
			tau7i = -s3 * tau3i + s1 * tau4i - s2 * tau5i;

		} else {
			tau7r = s3 * tau3r - s1 * tau4r + s2 * tau5r;
			tau7i = s3 * tau3i - s1 * tau4i + s2 * tau5i;

		}


		op[3].re = tau6r - tau7i;
		op[4].re = tau6r + tau7i;
		op[3].im = tau6i + tau7r;
		op[4].im = tau6i - tau7r;

		op[0].re += tau0r + tau1r + tau2r;
		op[0].im += tau0i + tau1i + tau2i;


	} else if (N == 8) {
		fft_type tau0r,tau0i,tau1r,tau1i,tau2r,tau2i,tau3r,tau3i,tau4r,tau4i,tau5r,tau5i,tau6r,tau6i,tau7r,tau7i,tau8r,tau8i,tau9r,tau9i;
		fft_type c1,s1,temp1r,temp1i,temp2r,temp2i;
		op[0].re = ip[0].re;
		op[0].im = ip[0].im;

		op[1].re = ip[l].re;
		op[1].im = ip[l].im;

		op[2].re = ip[2*l].re;
		op[2].im = ip[2*l].im;

		op[3].re = ip[3*l].re;
		op[3].im = ip[3*l].im;

		op[4].re = ip[4*l].re;
		op[4].im = ip[4*l].im;

		op[5].re = ip[5*l].re;
		op[5].im = ip[5*l].im;

		op[6].re = ip[6*l].re;
		op[6].im = ip[6*l].im;
		
		op[7].re = ip[7*l].re;
		op[7].im = ip[7*l].im;
		
		c1 = 0.70710678118654752440084436210485;
		s1 = 0.70710678118654752440084436210485;

		tau0r = op[0].re + op[4].re;
		tau4r = op[0].re - op[4].re;

		tau0i = op[0].im + op[4].im;
		tau4i = op[0].im - op[4].im;

		tau1r = op[1].re + op[7].re;
		tau5r = op[1].re - op[7].re;

		tau1i = op[1].im + op[7].im;
		tau5i = op[1].im - op[7].im;

		tau2r = op[3].re + op[5].re;
		tau6r = op[3].re - op[5].re;

		tau2i = op[3].im + op[5].im;
		tau6i = op[3].im - op[5].im;
		
		tau3r = op[2].re + op[6].re;
		tau7r = op[2].re - op[6].re;

		tau3i = op[2].im + op[6].im;
		tau7i = op[2].im - op[6].im;

		op[0].re = tau0r + tau1r + tau2r + tau3r;
		op[0].im = tau0i + tau1i + tau2i + tau3i;

		op[4].re = tau0r - tau1r - tau2r + tau3r;
		op[4].im = tau0i - tau1i - tau2i + tau3i;

		temp1r = tau1r - tau2r;
		temp1i = tau1i - tau2i;

		temp2r = tau5r + tau6r;
		temp2i = tau5i + tau6i;

		tau8r =  tau4r + c1 * temp1r;
		tau8i =  tau4i + c1 * temp1i;

		//tau9r = sgn * ( -s1 * temp2r - tau7r);
		//tau9i = sgn * ( -s1 * temp2i - tau7i);

		if (sgn == 1) {
			tau9r = -s1 * temp2r - tau7r;
			tau9i = -s1 * temp2i - tau7i;

		} else {
			tau9r = s1 * temp2r + tau7r;
			tau9i = s1 * temp2i + tau7i;

		}


		op[1].re = tau8r - tau9i;
		op[1].im = tau8i + tau9r;

		op[7].re = tau8r + tau9i;
		op[7].im = tau8i - tau9r;

		tau8r = tau0r - tau3r;
		tau8i = tau0i - tau3i;

		//tau9r = sgn * ( -tau5r + tau6r);
		//tau9i = sgn * ( -tau5i + tau6i);

		if (sgn == 1) {
			tau9r = -tau5r + tau6r;
			tau9i = -tau5i + tau6i;

		} else {
			tau9r = tau5r - tau6r;
			tau9i = tau5i - tau6i;

		}


		op[2].re = tau8r - tau9i;
		op[2].im = tau8i + tau9r;

		op[6].re = tau8r + tau9i;
		op[6].im = tau8i - tau9r;

		tau8r = tau4r - c1 * temp1r;
		tau8i = tau4i - c1 * temp1i;

		//tau9r = sgn * ( -s1 * temp2r + tau7r);
		//tau9i = sgn * ( -s1 * temp2i + tau7i);

		if (sgn == 1) {
			tau9r = -s1 * temp2r + tau7r;
			tau9i = -s1 * temp2i + tau7i;

		} else {
			tau9r = s1 * temp2r - tau7r;
			tau9i = s1 * temp2i - tau7i;

		}


		op[3].re = tau8r - tau9i;
		op[3].im = tau8i + tau9r;

		op[5].re = tau8r + tau9i;
		op[5].im = tau8i - tau9r;

		


	} else if (radix == 2) {
		int k,tkm1,ind;
		fft_type wlr,wli;
		fft_type tau1r,tau1i,tau2r,tau2i;
		m = N/2;
		ll = 2*l;
		mixed_radix_dit_rec(op,ip,obj,sgn,m,ll,inc+1);
		mixed_radix_dit_rec(op+m,ip+l,obj,sgn,m,ll,inc+1);

		for (k = 0; k < m;k++) {
			ind = m-1+k;
			wlr = (obj->twiddle+ind)->re;
			wli = (obj->twiddle+ind)->im;

			tkm1 = k+m;

			tau1r = op[k].re;
			tau1i = op[k].im;

			tau2r = op[tkm1].re * wlr - op[tkm1].im * wli;
			tau2i = op[tkm1].im * wlr + op[tkm1].re * wli;

			op[k].re = tau1r + tau2r;
			op[k].im = tau1i + tau2i;

			op[tkm1].re = tau1r - tau2r;
			op[tkm1].im = tau1i - tau2i;



		}

	} else if (radix == 3) {
		int k,tkm1,tkm2,ind;
		fft_type wlr,wli,wl2r,wl2i;
		fft_type tau0r,tau0i,tau1r,tau1i,tau2r,tau2i;
		fft_type ar,ai,br,bi,cr,ci;
		m = N/3;
		ll = 3*l;
		mixed_radix_dit_rec(op,ip,obj,sgn,m,ll,inc+1);
		mixed_radix_dit_rec(op+m,ip+l,obj,sgn,m,ll,inc+1);
		mixed_radix_dit_rec(op+2*m,ip+2*l,obj,sgn,m,ll,inc+1);
		//printf("%d \n",inc);
		//mixed_radix3_dit_rec(op,ip,obj,sgn,ll,m);


		for (k = 0; k < m; ++k) {
			ind = m-1+2*k;
			wlr = (obj->twiddle+ind)->re;
			wli = (obj->twiddle+ind)->im;
			ind++;
			wl2r = (obj->twiddle+ind)->re;
			wl2i = (obj->twiddle+ind)->im;
			tkm1 = k + m;
			tkm2 = tkm1 + m;

			ar = op[k].re;
			ai = op[k].im;

			br = op[tkm1].re * wlr - op[tkm1].im * wli;
			bi = op[tkm1].im * wlr + op[tkm1].re * wli;

			cr = op[tkm2].re * wl2r - op[tkm2].im * wl2i;
			ci = op[tkm2].im * wl2r + op[tkm2].re * wl2i;

			tau0r = br + cr;
			tau0i = bi + ci;

			tau1r = sgn * 0.86602540378 * (br - cr);
			tau1i = sgn * 0.86602540378 * (bi - ci);

			tau2r = ar - tau0r * 0.5000000000;
			tau2i = ai - tau0i * 0.5000000000;


			op[k].re = ar + tau0r;
			op[k].im = ai + tau0i;

			op[tkm1].re = tau2r + tau1i;
			op[tkm1].im = tau2i - tau1r;

			op[tkm2].re = tau2r - tau1i;
			op[tkm2].im = tau2i + tau1r;

		}

	} else if (radix == 4) {
		int k,tkm1,tkm2,tkm3,ind;
		fft_type wlr,wli,wl2r,wl2i,wl3r,wl3i;
		fft_type tau0r,tau0i,tau1r,tau1i,tau2r,tau2i,tau3r,tau3i;
		fft_type ar,ai,br,bi,cr,ci,dr,di;
		m = N/4;
		ll = 4*l;
		mixed_radix_dit_rec(op,ip,obj,sgn,m,ll,inc+1);
		mixed_radix_dit_rec(op+m,ip+l,obj,sgn,m,ll,inc+1);
		mixed_radix_dit_rec(op+2*m,ip+2*l,obj,sgn,m,ll,inc+1);
		mixed_radix_dit_rec(op+3*m,ip+3*l,obj,sgn,m,ll,inc+1);

		//mixed_radix4_dit_rec(op,ip,obj,sgn,ll,m);

		tkm1 = m;
		tkm2 = tkm1 + m;
		tkm3 = tkm2 + m;

		ar = op[0].re;
		ai = op[0].im;

		br = op[tkm1].re;
		bi = op[tkm1].im;

		cr = op[tkm2].re;
		ci = op[tkm2].im;

		dr = op[tkm3].re;
		di = op[tkm3].im;

		tau0r = ar + cr;
		tau0i = ai + ci;

		tau1r = ar - cr;
		tau1i = ai - ci;

		tau2r = br + dr;
		tau2i = bi + di;

		tau3r = sgn* (br - dr);
		tau3i = sgn* (bi - di);

		op[0].re = tau0r + tau2r ;
		op[0].im = tau0i + tau2i;

		op[tkm1].re = tau1r + tau3i;
		op[tkm1].im = tau1i - tau3r;

		op[tkm2].re = tau0r - tau2r;
		op[tkm2].im = tau0i - tau2i;

		op[tkm3].re = tau1r - tau3i;
		op[tkm3].im = tau1i + tau3r;


		for (k = 1; k < m; k++) {
			ind = m-1+3*k;
			wlr = (obj->twiddle+ind)->re;
			wli = (obj->twiddle+ind)->im;
			ind++;
			wl2r = (obj->twiddle+ind)->re;
			wl2i = (obj->twiddle+ind)->im;
			ind++;
			wl3r = (obj->twiddle+ind)->re;
			wl3i = (obj->twiddle+ind)->im;

			tkm1 = k+m;
			tkm2 = tkm1 + m;
			tkm3 = tkm2 + m;

			ar = op[k].re;
			ai = op[k].im;

			br = op[tkm1].re * wlr - op[tkm1].im * wli;
			bi = op[tkm1].im * wlr + op[tkm1].re * wli;

			cr = op[tkm2].re * wl2r - op[tkm2].im * wl2i;
			ci = op[tkm2].im * wl2r + op[tkm2].re * wl2i;

			dr = op[tkm3].re * wl3r - op[tkm3].im * wl3i;
			di = op[tkm3].im * wl3r + op[tkm3].re * wl3i;

			tau0r = ar + cr;
			tau0i = ai + ci;

			tau1r = ar - cr;
			tau1i = ai - ci;

			tau2r = br + dr;
			tau2i = bi + di;

			tau3r = sgn* (br - dr);
			tau3i = sgn* (bi - di);

			op[k].re = tau0r + tau2r ;
			op[k].im = tau0i + tau2i;

			op[tkm1].re = tau1r + tau3i;
			op[tkm1].im = tau1i - tau3r;

			op[tkm2].re = tau0r - tau2r;
			op[tkm2].im = tau0i - tau2i;

			op[tkm3].re = tau1r - tau3i;
			op[tkm3].im = tau1i + tau3r;

		}

	} else if (radix == 5) {
		int k,tkm1,tkm2,tkm3,tkm4,ind;
		fft_type wlr,wli,wl2r,wl2i,wl3r,wl3i,wl4r,wl4i;
		fft_type tau0r,tau0i,tau1r,tau1i,tau2r,tau2i,tau3r,tau3i;
		fft_type ar,ai,br,bi,cr,ci,dr,di,er,ei;
		fft_type tau4r,tau4i,tau5r,tau5i,tau6r,tau6i;
		fft_type c1,c2,s1,s2;
		m = N/5;
		ll = 5*l;
		mixed_radix_dit_rec(op,ip,obj,sgn,m,ll,inc+1);
		mixed_radix_dit_rec(op+m,ip+l,obj,sgn,m,ll,inc+1);
		mixed_radix_dit_rec(op+2*m,ip+2*l,obj,sgn,m,ll,inc+1);
		mixed_radix_dit_rec(op+3*m,ip+3*l,obj,sgn,m,ll,inc+1);
		mixed_radix_dit_rec(op+4*m,ip+4*l,obj,sgn,m,ll,inc+1);
		//printf("%d \n",inc);
		//mixed_radix3_dit_rec(op,ip,obj,sgn,ll,m);

		c1 = 0.30901699437;
		c2 = -0.80901699437;
		s1 = 0.95105651629;
		s2 = 0.58778525229;

		tkm1 = m;
		tkm2 = tkm1 + m;
		tkm3 = tkm2 + m;
		tkm4 = tkm3 + m;

		ar = op[0].re;
		ai = op[0].im;

		br = op[tkm1].re;
		bi = op[tkm1].im;

		cr = op[tkm2].re;
		ci = op[tkm2].im;

		dr = op[tkm3].re;
		di = op[tkm3].im;

		er = op[tkm4].re;
		ei = op[tkm4].im;

		tau0r = br + er;
		tau0i = bi + ei;

		tau1r = cr + dr;
		tau1i = ci + di;

		tau2r = br - er;
		tau2i = bi - ei;

		tau3r = cr - dr;
		tau3i = ci - di;

		op[0].re = ar + tau0r + tau1r;
		op[0].im = ai + tau0i + tau1i;

		tau4r = c1 * tau0r + c2 * tau1r;
		tau4i = c1 * tau0i + c2 * tau1i;

		tau5r = sgn * ( s1 * tau2r + s2 * tau3r);
		tau5i = sgn * ( s1 * tau2i + s2 * tau3i);

		tau6r = ar + tau4r;
		tau6i = ai + tau4i;

		op[tkm1].re = tau6r + tau5i;
		op[tkm1].im = tau6i - tau5r;

		op[tkm4].re = tau6r - tau5i;
		op[tkm4].im = tau6i + tau5r;

		tau4r = c2 * tau0r + c1 * tau1r;
		tau4i = c2 * tau0i + c1 * tau1i;

		tau5r = sgn * ( s2 * tau2r - s1 * tau3r);
		tau5i = sgn * ( s2 * tau2i - s1 * tau3i);

		tau6r = ar + tau4r;
		tau6i = ai + tau4i;

		op[tkm2].re = tau6r + tau5i;
		op[tkm2].im = tau6i - tau5r;

		op[tkm3].re = tau6r - tau5i;
		op[tkm3].im = tau6i + tau5r;

		for (k = 1; k < m; k++) {
			ind = m-1+4*k;
			wlr = (obj->twiddle+ind)->re;
			wli = (obj->twiddle+ind)->im;
			ind++;
			wl2r = (obj->twiddle+ind)->re;
			wl2i = (obj->twiddle+ind)->im;
			ind++;
			wl3r = (obj->twiddle+ind)->re;
			wl3i = (obj->twiddle+ind)->im;
			ind++;
			wl4r = (obj->twiddle+ind)->re;
			wl4i = (obj->twiddle+ind)->im;

			tkm1 = k + m;
			tkm2 = tkm1 + m;
			tkm3 = tkm2 + m;
			tkm4 = tkm3 + m;

			ar = op[k].re;
			ai = op[k].im;

			br = op[tkm1].re * wlr - op[tkm1].im * wli;
			bi = op[tkm1].im * wlr + op[tkm1].re * wli;

			cr = op[tkm2].re * wl2r - op[tkm2].im * wl2i;
			ci = op[tkm2].im * wl2r + op[tkm2].re * wl2i;

			dr = op[tkm3].re * wl3r - op[tkm3].im * wl3i;
			di = op[tkm3].im * wl3r + op[tkm3].re * wl3i;

			er = op[tkm4].re * wl4r - op[tkm4].im * wl4i;
			ei = op[tkm4].im * wl4r + op[tkm4].re * wl4i;

			tau0r = br + er;
			tau0i = bi + ei;

			tau1r = cr + dr;
			tau1i = ci + di;

			tau2r = br - er;
			tau2i = bi - ei;

			tau3r = cr - dr;
			tau3i = ci - di;

			op[k].re = ar + tau0r + tau1r;
			op[k].im = ai + tau0i + tau1i;

			tau4r = c1 * tau0r + c2 * tau1r;
			tau4i = c1 * tau0i + c2 * tau1i;

			//tau5r = sgn * ( s1 * tau2r + s2 * tau3r);
			//tau5i = sgn * ( s1 * tau2i + s2 * tau3i);

			if (sgn == 1) {
				tau5r = s1 * tau2r + s2 * tau3r;
				tau5i = s1 * tau2i + s2 * tau3i;

			} else {
				tau5r = -s1 * tau2r - s2 * tau3r;
				tau5i = -s1 * tau2i - s2 * tau3i;

			}

			tau6r = ar + tau4r;
			tau6i = ai + tau4i;

			op[tkm1].re = tau6r + tau5i;
			op[tkm1].im = tau6i - tau5r;

			op[tkm4].re = tau6r - tau5i;
			op[tkm4].im = tau6i + tau5r;

			tau4r = c2 * tau0r + c1 * tau1r;
			tau4i = c2 * tau0i + c1 * tau1i;

			//tau5r = sgn * ( s2 * tau2r - s1 * tau3r);
			//tau5i = sgn * ( s2 * tau2i - s1 * tau3i);

			if (sgn == 1) {
				tau5r = s2 * tau2r - s1 * tau3r;
				tau5i = s2 * tau2i - s1 * tau3i;

			} else {
				tau5r = -s2 * tau2r + s1 * tau3r;
				tau5i = -s2 * tau2i + s1 * tau3i;

			}

			tau6r = ar + tau4r;
		    tau6i = ai + tau4i;

			op[tkm2].re = tau6r + tau5i;
			op[tkm2].im = tau6i - tau5r;

			op[tkm3].re = tau6r - tau5i;
			op[tkm3].im = tau6i + tau5r;

		}

	} else if (radix == 7) {
		int k,tkm1,tkm2,tkm3,tkm4,tkm5,tkm6,ind;
		fft_type wlr,wli,wl2r,wl2i,wl3r,wl3i,wl4r,wl4i,wl5r,wl5i,wl6r,wl6i;
		fft_type tau0r,tau0i,tau1r,tau1i,tau2r,tau2i,tau3r,tau3i;
		fft_type ar,ai,br,bi,cr,ci,dr,di,er,ei,fr,fi,gr,gi;
		fft_type tau4r,tau4i,tau5r,tau5i,tau6r,tau6i,tau7r,tau7i;
		fft_type c1,c2,c3,s1,s2,s3;
		m = N/7;
		ll = 7*l;
		mixed_radix_dit_rec(op,ip,obj,sgn,m,ll,inc+1);
		mixed_radix_dit_rec(op+m,ip+l,obj,sgn,m,ll,inc+1);
		mixed_radix_dit_rec(op+2*m,ip+2*l,obj,sgn,m,ll,inc+1);
		mixed_radix_dit_rec(op+3*m,ip+3*l,obj,sgn,m,ll,inc+1);
		mixed_radix_dit_rec(op+4*m,ip+4*l,obj,sgn,m,ll,inc+1);
		mixed_radix_dit_rec(op+5*m,ip+5*l,obj,sgn,m,ll,inc+1);
		mixed_radix_dit_rec(op+6*m,ip+6*l,obj,sgn,m,ll,inc+1);
		//printf("%d \n",inc);
		//mixed_radix3_dit_rec(op,ip,obj,sgn,ll,m);

		c1 = 0.62348980185;
		c2 = -0.22252093395;
		c3 = -0.9009688679;
		s1 = 0.78183148246;
		s2 = 0.97492791218;
		s3 = 0.43388373911;

		tkm1 = m;
		tkm2 = tkm1 + m;
		tkm3 = tkm2 + m;
		tkm4 = tkm3 + m;
		tkm5 = tkm4 + m;
		tkm6 = tkm5 + m;

		ar = op[0].re;
		ai = op[0].im;

		br = op[tkm1].re;
		bi = op[tkm1].im;

		cr = op[tkm2].re;
		ci = op[tkm2].im;

		dr = op[tkm3].re;
		di = op[tkm3].im;

		er = op[tkm4].re;
		ei = op[tkm4].im;

		fr = op[tkm5].re;
		fi = op[tkm5].im;

		gr = op[tkm6].re;
		gi = op[tkm6].im;

		tau0r = br + gr;
		tau3r = br - gr;
		tau0i = bi + gi;
		tau3i = bi - gi;

		tau1r = cr + fr;
		tau4r = cr - fr;
		tau1i = ci + fi;
		tau4i = ci - fi;

		tau2r = dr + er;
		tau5r = dr - er;
		tau2i = di + ei;
		tau5i = di - ei;

		op[0].re = ar + tau0r + tau1r + tau2r;
		op[0].im = ai + tau0i + tau1i + tau2i;

		tau6r = ar + c1 * tau0r + c2 * tau1r + c3 * tau2r;
		tau6i = ai + c1 * tau0i + c2 * tau1i + c3 * tau2i;

		//tau7r = sgn * ( -s1 * tau3r - s2 * tau4r - s3 * tau5r);
		//tau7i = sgn * ( -s1 * tau3i - s2 * tau4i - s3 * tau5i);

		if (sgn == 1) {
			tau7r = -s1 * tau3r - s2 * tau4r - s3 * tau5r;
			tau7i = -s1 * tau3i - s2 * tau4i - s3 * tau5i;

		} else {
			tau7r = s1 * tau3r + s2 * tau4r + s3 * tau5r;
			tau7i = s1 * tau3i + s2 * tau4i + s3 * tau5i;

		}


		op[tkm1].re = tau6r - tau7i;
		op[tkm1].im = tau6i + tau7r;

		op[tkm6].re = tau6r + tau7i;
		op[tkm6].im = tau6i - tau7r;

		tau6r = ar + c2 * tau0r + c3 * tau1r + c1 * tau2r;
		tau6i = ai + c2 * tau0i + c3 * tau1i + c1 * tau2i;

		//tau7r = sgn * ( -s2 * tau3r + s3 * tau4r + s1 * tau5r);
		//tau7i = sgn * ( -s2 * tau3i + s3 * tau4i + s1 * tau5i);

		if (sgn == 1) {
			tau7r = -s2 * tau3r + s3 * tau4r + s1 * tau5r;
			tau7i = -s2 * tau3i + s3 * tau4i + s1 * tau5i;

		} else {
			tau7r = s2 * tau3r - s3 * tau4r - s1 * tau5r;
			tau7i = s2 * tau3i - s3 * tau4i - s1 * tau5i;

		}


		op[tkm2].re = tau6r - tau7i;
		op[tkm2].im = tau6i + tau7r;

		op[tkm5].re = tau6r + tau7i;
		op[tkm5].im = tau6i - tau7r;

		tau6r = ar + c3 * tau0r + c1 * tau1r + c2 * tau2r;
		tau6i = ai + c3 * tau0i + c1 * tau1i + c2 * tau2i;

		//tau7r = sgn * ( -s3 * tau3r + s1 * tau4r - s2 * tau5r);
		//tau7i = sgn * ( -s3 * tau3i + s1 * tau4i - s2 * tau5i);

		if (sgn == 1) {
			tau7r = -s3 * tau3r + s1 * tau4r - s2 * tau5r;
			tau7i = -s3 * tau3i + s1 * tau4i - s2 * tau5i;

		} else {
			tau7r = s3 * tau3r - s1 * tau4r + s2 * tau5r;
			tau7i = s3 * tau3i - s1 * tau4i + s2 * tau5i;

		}


		op[tkm3].re = tau6r - tau7i;
		op[tkm3].im = tau6i + tau7r;

		op[tkm4].re = tau6r + tau7i;
		op[tkm4].im = tau6i - tau7r;


		for (k = 1; k < m; k++) {
			ind = m-1+6*k;
			wlr = (obj->twiddle+ind)->re;
			wli = (obj->twiddle+ind)->im;
			ind++;
			wl2r = (obj->twiddle+ind)->re;
			wl2i = (obj->twiddle+ind)->im;
			ind++;
			wl3r = (obj->twiddle+ind)->re;
			wl3i = (obj->twiddle+ind)->im;
			ind++;
			wl4r = (obj->twiddle+ind)->re;
			wl4i = (obj->twiddle+ind)->im;
			ind++;
			wl5r = (obj->twiddle+ind)->re;
			wl5i = (obj->twiddle+ind)->im;
			ind++;
			wl6r = (obj->twiddle+ind)->re;
			wl6i = (obj->twiddle+ind)->im;

			tkm1 = k + m;
			tkm2 = tkm1 + m;
			tkm3 = tkm2 + m;
			tkm4 = tkm3 + m;
			tkm5 = tkm4 + m;
			tkm6 = tkm5 + m;

			ar = op[k].re;
			ai = op[k].im;

			br = op[tkm1].re * wlr - op[tkm1].im * wli;
			bi = op[tkm1].im * wlr + op[tkm1].re * wli;

			cr = op[tkm2].re * wl2r - op[tkm2].im * wl2i;
			ci = op[tkm2].im * wl2r + op[tkm2].re * wl2i;

			dr = op[tkm3].re * wl3r - op[tkm3].im * wl3i;
			di = op[tkm3].im * wl3r + op[tkm3].re * wl3i;

			er = op[tkm4].re * wl4r - op[tkm4].im * wl4i;
			ei = op[tkm4].im * wl4r + op[tkm4].re * wl4i;

			fr = op[tkm5].re * wl5r - op[tkm5].im * wl5i;
			fi = op[tkm5].im * wl5r + op[tkm5].re * wl5i;

			gr = op[tkm6].re * wl6r - op[tkm6].im * wl6i;
			gi = op[tkm6].im * wl6r + op[tkm6].re * wl6i;

			tau0r = br + gr;
			tau3r = br - gr;
			tau0i = bi + gi;
			tau3i = bi - gi;

			tau1r = cr + fr;
			tau4r = cr - fr;
			tau1i = ci + fi;
			tau4i = ci - fi;

			tau2r = dr + er;
			tau5r = dr - er;
			tau2i = di + ei;
			tau5i = di - ei;

			op[k].re = ar + tau0r + tau1r + tau2r;
			op[k].im = ai + tau0i + tau1i + tau2i;

			tau6r = ar + c1 * tau0r + c2 * tau1r + c3 * tau2r;
			tau6i = ai + c1 * tau0i + c2 * tau1i + c3 * tau2i;

			//tau7r = sgn * ( -s1 * tau3r - s2 * tau4r - s3 * tau5r);
			//tau7i = sgn * ( -s1 * tau3i - s2 * tau4i - s3 * tau5i);

			if (sgn == 1) {
				tau7r = -s1 * tau3r - s2 * tau4r - s3 * tau5r;
				tau7i = -s1 * tau3i - s2 * tau4i - s3 * tau5i;

			} else {
				tau7r = s1 * tau3r + s2 * tau4r + s3 * tau5r;
				tau7i = s1 * tau3i + s2 * tau4i + s3 * tau5i;

			}


			op[tkm1].re = tau6r - tau7i;
			op[tkm1].im = tau6i + tau7r;

			op[tkm6].re = tau6r + tau7i;
			op[tkm6].im = tau6i - tau7r;

			tau6r = ar + c2 * tau0r + c3 * tau1r + c1 * tau2r;
			tau6i = ai + c2 * tau0i + c3 * tau1i + c1 * tau2i;

			//tau7r = sgn * ( -s2 * tau3r + s3 * tau4r + s1 * tau5r);
			//tau7i = sgn * ( -s2 * tau3i + s3 * tau4i + s1 * tau5i);

			if (sgn == 1) {
				tau7r = -s2 * tau3r + s3 * tau4r + s1 * tau5r;
				tau7i = -s2 * tau3i + s3 * tau4i + s1 * tau5i;

			} else {
				tau7r = s2 * tau3r - s3 * tau4r - s1 * tau5r;
				tau7i = s2 * tau3i - s3 * tau4i - s1 * tau5i;

			}


			op[tkm2].re = tau6r - tau7i;
			op[tkm2].im = tau6i + tau7r;

			op[tkm5].re = tau6r + tau7i;
			op[tkm5].im = tau6i - tau7r;

			tau6r = ar + c3 * tau0r + c1 * tau1r + c2 * tau2r;
			tau6i = ai + c3 * tau0i + c1 * tau1i + c2 * tau2i;

			//tau7r = sgn * ( -s3 * tau3r + s1 * tau4r - s2 * tau5r);
			//tau7i = sgn * ( -s3 * tau3i + s1 * tau4i - s2 * tau5i);

			if (sgn == 1) {
				tau7r = -s3 * tau3r + s1 * tau4r - s2 * tau5r;
				tau7i = -s3 * tau3i + s1 * tau4i - s2 * tau5i;

			} else {
				tau7r = s3 * tau3r - s1 * tau4r + s2 * tau5r;
				tau7i = s3 * tau3i - s1 * tau4i + s2 * tau5i;

			}


			op[tkm3].re = tau6r - tau7i;
			op[tkm3].im = tau6i + tau7r;

			op[tkm4].re = tau6r + tau7i;
			op[tkm4].im = tau6i - tau7r;

		}

	} else if (radix == 8) {
		int k,tkm1,tkm2,tkm3,tkm4,tkm5,tkm6,tkm7,ind;
		fft_type wlr,wli,wl2r,wl2i,wl3r,wl3i,wl4r,wl4i,wl5r,wl5i,wl6r,wl6i,wl7r,wl7i;
		fft_type tau0r,tau0i,tau1r,tau1i,tau2r,tau2i,tau3r,tau3i;
		fft_type ar,ai,br,bi,cr,ci,dr,di,er,ei,fr,fi,gr,gi,hr,hi;
		fft_type tau4r,tau4i,tau5r,tau5i,tau6r,tau6i,tau7r,tau7i,tau8r,tau8i,tau9r,tau9i;
		fft_type c1,s1,temp1r,temp1i,temp2r,temp2i;
		m = N/8;
		ll = 8*l;
		mixed_radix_dit_rec(op,ip,obj,sgn,m,ll,inc+1);
		mixed_radix_dit_rec(op+m,ip+l,obj,sgn,m,ll,inc+1);
		mixed_radix_dit_rec(op+2*m,ip+2*l,obj,sgn,m,ll,inc+1);
		mixed_radix_dit_rec(op+3*m,ip+3*l,obj,sgn,m,ll,inc+1);
		mixed_radix_dit_rec(op+4*m,ip+4*l,obj,sgn,m,ll,inc+1);
		mixed_radix_dit_rec(op+5*m,ip+5*l,obj,sgn,m,ll,inc+1);
		mixed_radix_dit_rec(op+6*m,ip+6*l,obj,sgn,m,ll,inc+1);
		mixed_radix_dit_rec(op+7*m,ip+7*l,obj,sgn,m,ll,inc+1);
		//printf("%d \n",inc);
		//mixed_radix3_dit_rec(op,ip,obj,sgn,ll,m);

		c1 = 0.70710678118654752440084436210485;
		s1 = 0.70710678118654752440084436210485;


		for (k = 0; k < m; k++) {
			ind = m-1+7*k;
			wlr = (obj->twiddle+ind)->re;
			wli = (obj->twiddle+ind)->im;
			ind++;
			wl2r = (obj->twiddle+ind)->re;
			wl2i = (obj->twiddle+ind)->im;
			ind++;
			wl3r = (obj->twiddle+ind)->re;
			wl3i = (obj->twiddle+ind)->im;
			ind++;
			wl4r = (obj->twiddle+ind)->re;
			wl4i = (obj->twiddle+ind)->im;
			ind++;
			wl5r = (obj->twiddle+ind)->re;
			wl5i = (obj->twiddle+ind)->im;
			ind++;
			wl6r = (obj->twiddle+ind)->re;
			wl6i = (obj->twiddle+ind)->im;
			ind++;
			wl7r = (obj->twiddle+ind)->re;
			wl7i = (obj->twiddle+ind)->im;

			tkm1 = k + m;
			tkm2 = tkm1 + m;
			tkm3 = tkm2 + m;
			tkm4 = tkm3 + m;
			tkm5 = tkm4 + m;
			tkm6 = tkm5 + m;
			tkm7 = tkm6 + m;

			ar = op[k].re;
			ai = op[k].im;

			br = op[tkm1].re * wlr - op[tkm1].im * wli;
			bi = op[tkm1].im * wlr + op[tkm1].re * wli;

			cr = op[tkm2].re * wl2r - op[tkm2].im * wl2i;
			ci = op[tkm2].im * wl2r + op[tkm2].re * wl2i;

			dr = op[tkm3].re * wl3r - op[tkm3].im * wl3i;
			di = op[tkm3].im * wl3r + op[tkm3].re * wl3i;

			er = op[tkm4].re * wl4r - op[tkm4].im * wl4i;
			ei = op[tkm4].im * wl4r + op[tkm4].re * wl4i;

			fr = op[tkm5].re * wl5r - op[tkm5].im * wl5i;
			fi = op[tkm5].im * wl5r + op[tkm5].re * wl5i;

			gr = op[tkm6].re * wl6r - op[tkm6].im * wl6i;
			gi = op[tkm6].im * wl6r + op[tkm6].re * wl6i;

			hr = op[tkm7].re * wl7r - op[tkm7].im * wl7i;
			hi = op[tkm7].im * wl7r + op[tkm7].re * wl7i;

			tau0r = ar + er;
			tau4r = ar - er;
			tau0i = ai + ei;
			tau4i = ai - ei;

			tau1r = br + hr;
			tau5r = br - hr;
			tau1i = bi + hi;
			tau5i = bi - hi;

			tau2r = dr + fr;
			tau6r = dr - fr;
			tau6i = di - fi;
			tau2i = di + fi;

			tau3r = cr + gr;
			tau7r = cr - gr;
			tau7i = ci - gi;
			tau3i = ci + gi;

			op[k].re = tau0r + tau1r + tau2r + tau3r;
			op[k].im = tau0i + tau1i + tau2i + tau3i;

			op[tkm4].re = tau0r - tau1r - tau2r + tau3r;
			op[tkm4].im = tau0i - tau1i - tau2i + tau3i;

			temp1r = tau1r - tau2r;
			temp1i = tau1i - tau2i;

			temp2r = tau5r + tau6r;
			temp2i = tau5i + tau6i;

			tau8r =  tau4r + c1 * temp1r;
			tau8i =  tau4i + c1 * temp1i;

			//tau9r = sgn * ( -s1 * temp2r - tau7r);
			//tau9i = sgn * ( -s1 * temp2i - tau7i);

			if (sgn == 1) {
				tau9r = -s1 * temp2r - tau7r;
				tau9i = -s1 * temp2i - tau7i;

			} else {
				tau9r = s1 * temp2r + tau7r;
				tau9i = s1 * temp2i + tau7i;

			}


			op[tkm1].re = tau8r - tau9i;
			op[tkm1].im = tau8i + tau9r;

			op[tkm7].re = tau8r + tau9i;
			op[tkm7].im = tau8i - tau9r;

			tau8r = tau0r - tau3r;
			tau8i = tau0i - tau3i;

			//tau9r = sgn * ( -tau5r + tau6r);
			//tau9i = sgn * ( -tau5i + tau6i);

			if (sgn == 1) {
				tau9r = -tau5r + tau6r;
				tau9i = -tau5i + tau6i;

			} else {
				tau9r = tau5r - tau6r;
				tau9i = tau5i - tau6i;

			}


			op[tkm2].re = tau8r - tau9i;
			op[tkm2].im = tau8i + tau9r;

			op[tkm6].re = tau8r + tau9i;
			op[tkm6].im = tau8i - tau9r;

			tau8r = tau4r - c1 * temp1r;
			tau8i = tau4i - c1 * temp1i;

			//tau9r = sgn * ( -s1 * temp2r + tau7r);
			//tau9i = sgn * ( -s1 * temp2i + tau7i);

			if (sgn == 1) {
				tau9r = -s1 * temp2r + tau7r;
				tau9i = -s1 * temp2i + tau7i;

			} else {
				tau9r = s1 * temp2r - tau7r;
				tau9i = s1 * temp2i - tau7i;

			}


			op[tkm3].re = tau8r - tau9i;
			op[tkm3].im = tau8i + tau9r;

			op[tkm5].re = tau8r + tau9i;
			op[tkm5].im = tau8i - tau9r;

		}

	} else {
		int k,i,ind;
		int M,tkm,u,v,t,tt;
		fft_type temp1r,temp1i,temp2r,temp2i;
		fft_type* wlr = (fft_type*) malloc (sizeof(fft_type) * (radix-1));
		fft_type* wli = (fft_type*) malloc (sizeof(fft_type) * (radix-1));
		fft_type* taur = (fft_type*) malloc (sizeof(fft_type) * (radix-1));
		fft_type* taui = (fft_type*) malloc (sizeof(fft_type) * (radix-1));
		fft_type* c1 = (fft_type*) malloc (sizeof(fft_type) * (radix-1));
		fft_type* s1 = (fft_type*) malloc (sizeof(fft_type) * (radix-1));
		fft_type* yr = (fft_type*) malloc (sizeof(fft_type) * (radix));
		fft_type* yi = (fft_type*) malloc (sizeof(fft_type) * (radix));
		
		m = N /radix;
		ll = radix * l;

		for (i = 0; i < radix;++i) {
			mixed_radix_dit_rec(op+i*m,ip+i*l,obj,sgn,m,ll,inc+1);
		}

		M = (radix - 1 )/2;

		for (i = 1; i < M+1;++i) {
			c1[i-1] = cos(i*PI2/radix);
			s1[i-1] = sin(i*PI2/radix);
		}

		for (i = 0; i < M;++i) {
			s1[i+M] = -s1[M-1-i];
			c1[i+M] =  c1[M-1-i];
		}

		for (k = 0; k < m;++k) {
			ind = m-1+(radix-1)*k;
			yr[0] = op[k].re;
			yi[0] = op[k].im;
			for(i =0; i < radix -1;++i) {
				wlr[i] = (obj->twiddle+ind)->re;
				wli[i] = (obj->twiddle+ind)->im;
				tkm = k + (i+1)*m;
				yr[i+1] = op[tkm].re * wlr[i] - op[tkm].im * wli[i];
				yi[i+1] = op[tkm].im * wlr[i] + op[tkm].re * wli[i];
				ind++;
			}

			for (i = 0; i < M; ++i) {
				taur[i] = yr[i+1] + yr[radix-1-i];
				taui[i+M] = yi[i+1] - yi[radix-1-i];
				taui[i] = yi[i+1] + yi[radix-1-i];
				taur[i+M] = yr[i+1] - yr[radix-1-i];
			}

			temp1r = yr[0];
			temp1i = yi[0];

			for (i = 0; i < M; ++i) {
				temp1r+= taur[i];
				temp1i+= taui[i];
			}

			op[k].re = temp1r;
			op[k].im = temp1i;

			for (u = 0; u < M; u++) {
				temp1r = yr[0];
				temp1i = yi[0];
				temp2r = 0.0;
				temp2i = 0.0;
				for (v = 0; v < M; v++) {
					//int ind2 = (u+v)%M;
					t = (u+1)*(v+1);
					while(t >= radix)
						t-=radix;
					tt = t-1;

					temp1r+= c1[tt]*taur[v];
					temp1i+= c1[tt]*taui[v];
					temp2r-= s1[tt]*taur[v+M];
					temp2i-= s1[tt]*taui[v+M];
				}
				temp2r = sgn * temp2r;
				temp2i = sgn * temp2i;


				op[k + (u+1)*m].re = temp1r - temp2i;
				op[k + (u+1)*m].im = temp1i + temp2r;

				op[k + (radix-u-1)*m].re = temp1r + temp2i;
				op[k + (radix-u-1)*m].im = temp1i - temp2r;
			}


		}
		free(wlr);
		free(wli);
		free(taur);
		free(taui);
		free(c1);
		free(s1);
		free(yr);
		free(yi);

	}


}

static void bluestein_exp(fft_data *hl, fft_data *hlt, int len, int M) {
	fft_type PI,theta,angle;
	int l2,len2,i;
	PI = 3.1415926535897932384626433832795;
	theta = PI / len;
	l2 = 0;
	len2 = 2 * len;

	for (i = 0 ; i < len; ++i) {
		angle = theta * l2;
		hlt[i].re = cos(angle);
		hlt[i].im = sin(angle);
		hl[i].re = hlt[i].re;
		hl[i].im = hlt[i].im;
		l2+=2*i+1;
		while (l2 > len2) {
			l2-=len2;
		}

	}

	for (i = len; i < M-len+1; i++) {
		hl[i].re = 0.0;
		hl[i].im = 0.0;
	}

	for (i = M - len + 1; i < M; i++) {
		hl[i].re = hlt[M-i].re;
		hl[i].im = hlt[M-i].im;
	}

}

static void bluestein_fft(fft_data *data, fft_data *oup,fft_object obj,int sgn, int N) {

	int K,M,ii,i;
	int def_lt,def_N,def_sgn;
	fft_type scale,temp;
	fft_data* yn;
	fft_data* hk;
	fft_data* tempop;
	fft_data* yno;
	fft_data* hlt;
	obj->lt = 0;
	K = (int) pow(2.0,ceil((double) log10((double) N)/log10((double) 2.0)));
	def_lt = 1;
	def_sgn = obj->sgn;
	def_N = obj->N;

	if (K < 2 * N - 2) {
		M = K * 2;
	} else {
		M = K;
	}
	obj->N = M;

	yn = (fft_data*) malloc (sizeof(fft_data) * M);
	hk = (fft_data*) malloc (sizeof(fft_data) * M);
	tempop = (fft_data*) malloc (sizeof(fft_data) * M);
	yno = (fft_data*) malloc (sizeof(fft_data) * M);
    hlt = (fft_data*) malloc (sizeof(fft_data) * N);
    //fft_data* twi = (fft_data*) malloc (sizeof(fft_data) * M);

    bluestein_exp(tempop,hlt,N,M);
    scale = 1.0/M;

    for (ii = 0; ii < M; ++ii) {
		tempop[ii].im *= scale;
		tempop[ii].re *= scale;
	}

	//fft_object obj = initialize_fft2(M,1);
	fft_exec(obj,tempop,hk);

    if (sgn == 1) {
		for (i = 0; i < N; i++) {
			tempop[i].re = data[i].re * hlt[i].re + data[i].im * hlt[i].im;
			tempop[i].im = -data[i].re * hlt[i].im + data[i].im * hlt[i].re;
		}
	} else {
		for (i = 0; i < N; i++) {
			tempop[i].re = data[i].re * hlt[i].re - data[i].im * hlt[i].im;
			tempop[i].im = data[i].re * hlt[i].im + data[i].im * hlt[i].re;
		}
	}

	for (i = N;i < M;i++) {
		tempop[i].re = 0.0;
		tempop[i].im = 0.0;
	}

	fft_exec(obj,tempop,yn);

	if (sgn == 1) {
		for (i = 0; i < M; i++) {
			temp = yn[i].re * hk[i].re - yn[i].im * hk[i].im;
			yn[i].im = yn[i].re * hk[i].im + yn[i].im * hk[i].re;
			yn[i].re = temp;
		}
	} else {
		for (i = 0; i < M; i++) {
			temp = yn[i].re * hk[i].re + yn[i].im * hk[i].im;
			yn[i].im = -yn[i].re * hk[i].im + yn[i].im * hk[i].re;
			yn[i].re = temp;
		}

	}

    //IFFT

    for (ii = 0; ii < M; ++ii) {
		(obj->twiddle+ii)->im = -(obj->twiddle+ii)->im;
	}

	obj->sgn = -1*sgn;

	fft_exec(obj,yn,yno);

	if (sgn == 1) {
		for (i = 0; i < N; i++) {
			oup[i].re = yno[i].re * hlt[i].re + yno[i].im * hlt[i].im;
			oup[i].im = -yno[i].re * hlt[i].im + yno[i].im * hlt[i].re;
		}
	} else {
		for (i = 0; i < N; i++) {
			oup[i].re = yno[i].re * hlt[i].re - yno[i].im * hlt[i].im;
			oup[i].im = yno[i].re * hlt[i].im + yno[i].im * hlt[i].re;
		}

	}

	obj->sgn = def_sgn;
	obj->N = def_N;
	obj->lt = def_lt;
	for (ii = 0; ii < M; ++ii) {
		(obj->twiddle+ii)->im = -(obj->twiddle+ii)->im;
	}

    free(yn);
    free(yno);
    free(tempop);
    free(hk);
    free(hlt);

}



void fft_exec(fft_object obj,fft_data *inp,fft_data *oup) {
	if (obj->lt == 0) {
		//fftct_radix3_dit_rec(inp,oup,obj, obj->sgn, obj->N);
		//fftct_mixed_rec(inp,oup,obj, obj->sgn, obj->N);
		//printf("%f \n", 1.785);
		int l,inc;
		int nn,sgn1;
		nn = obj->N;
		sgn1 = obj->sgn;
		l = 1;
		inc = 0;
		//radix3_dit_rec(oup,inp,obj,sgn1,nn,l);
		mixed_radix_dit_rec(oup,inp,obj,sgn1,nn,l,inc);
	} else if (obj->lt == 1){
		//printf("%f \n", 1.785);
		int nn,sgn1;
		nn = obj->N;
		sgn1 = obj->sgn;
		bluestein_fft(inp,oup,obj,sgn1,nn);

	}

}

int divideby(int M,int d) {
	while (M%d == 0) {
		M = M/d;
	}
	if (M == 1) {
		return 1;
	}
	 return 0;
}

int dividebyN(int N) {
	while (N%53 == 0) {
		N = N/53;
	}
	while (N%47 == 0) {
		N = N/47;
	}
	while (N%43 == 0) {
		N = N/43;
	}
	while (N%41 == 0) {
		N = N/41;
	}
	while (N%37 == 0) {
		N = N/37;
	}
	while (N%31 == 0) {
		N = N/31;
	}
	while (N%29 == 0) {
		N = N/29;
	}
	while (N%23 == 0) {
		N = N/23;
	}
	while (N%17 == 0) {
		N = N/17;
	}
	while (N%13 == 0) {
		N = N/13;
	}
	while (N%11 == 0) {
		N = N/11;
	}
	while (N%8 == 0) {
		N = N/8;
	}
	while (N%7 == 0) {
		N = N/7;
	}
	while (N%5 == 0) {
		N = N/5;
	}
	while (N%4 == 0) {
		N = N/4;
	}
	while (N%3 == 0) {
		N = N/3;
	}
	while (N%2 == 0) {
		N = N/2;
	}
	if (N == 1) {
		return 1;
	}
	 return 0;

}

int factors(int M, int* arr) {
	int i,N,num,mult,m1,m2;
	i = 0;
	N = M;
	while (N%53 == 0) {
		N = N/53;
		arr[i] = 53;
		i++;
	}
	while (N%47 == 0) {
		N = N/47;
		arr[i] = 47;
		i++;
	}
	while (N%43 == 0) {
		N = N/43;
		arr[i] = 43;
		i++;
	}
	while (N%41 == 0) {
		N = N/41;
		arr[i] = 41;
		i++;
	}
	while (N%37 == 0) {
		N = N/37;
		arr[i] = 37;
		i++;
	}
	while (N%31 == 0) {
		N = N/31;
		arr[i] = 31;
		i++;
	}
	while (N%29 == 0) {
		N = N/29;
		arr[i] = 29;
		i++;
	}
	while (N%23 == 0) {
		N = N/23;
		arr[i] = 23;
		i++;
	}
	while (N%19 == 0) {
		N = N/19;
		arr[i] = 19;
		i++;
	}
	while (N%17 == 0) {
		N = N/17;
		arr[i] = 17;
		i++;
	}
	while (N%13 == 0) {
		N = N/13;
		arr[i] = 13;
		i++;
	}
	while (N%11 == 0) {
		N = N/11;
		arr[i] = 11;
		i++;
	}
	while (N%8 == 0) {
		N = N/8;
		arr[i] = 8;
		i++;
	}
	while (N%7 == 0) {
		N = N/7;
		arr[i] = 7;
		i++;
	}
	while (N%5 == 0) {
		N = N/5;
		arr[i] = 5;
		i++;
	}
	while (N%4 == 0) {
		N = N/4;
		arr[i] = 4;
		i++;
	}
	while (N%3 == 0) {
		N = N/3;
		arr[i] = 3;
		i++;
	}
	while (N%2 == 0) {
		N = N/2;
		arr[i] = 2;
		i++;
	}
	if (N > 31) {
		num = 2;

		while (N > 1) {
			mult = num*6;
			m1 = mult-1;
			m2 = mult+1;
			while (N%m1 == 0 ) {
				arr[i] = m1;
				i++;
				N = N / m1;
			}
			while (N%m2 == 0 ) {
				arr[i] = m2;
				i++;
				N = N / m2;
			}
			num+=1;

		}
	}
	return i;

}


void twiddle(fft_data *vec,int N, int radix) {
	int K,KL;
	fft_type theta,theta2;
	theta = PI2/N;
	KL = N/radix;
	vec[0].re = 1.0;
	vec[0].im = 0.0;

	for (K = 1; K < KL;K++) {
		theta2 = theta * K;
		vec[K].re = cos(theta2);
		vec[K].im = -sin(theta2);
	}

}

void longvectorN(fft_data *sig,int N, int *array, int tx) {
	int L,i,Ls,ct,j,k;
	fft_type theta;
	L = 1;
	ct = 0;
	for (i = 0; i < tx; i++) {
		L = L * array[tx-1-i];
		Ls = L / array[tx-1-i];
		theta = -1.0 * PI2/L;
		for (j = 0; j < Ls;j++) {
			for (k = 0; k < array[tx-1-i] -1 ;k++) {
				sig[ct].re = cos((k+1)*j*theta);
				sig[ct].im = sin((k+1)*j*theta);
				ct++;
			}
		}

	}

}




void free_fft(fft_object object) {
	free(object);
}