ssfm_for_cnlse_withpmd.asv

是有分布傅立叶算法解薛定谔方程的一个程序（下面我就不乱说说了请认真阅读您的文件包然后写出其具体功能(至少要20个字)。尽量不要让站长把时间都花费在为您修正说明上。压缩包解压时不能有密码。系统会自动删除

function [u1x,u1y] = UPMPol(u0x,u0y,dt,dz,nz,alphax,alphay,betapx,betapy,gamma,correlationlength,angles,tetas,beta0s,beta1s);

nt=length(u0x); % nt is the number of point
w=2*pi*[(0:nt/2-1),(-nt/2:-1)]'/(dt*nt); %constructing used frequencies

linearoperatorx = -alphax/2; % in the lines above are constructed the linear operator in X and Y.
for ii = 0:length(betapx)-1; 
  linearoperatorx = linearoperatorx - j*betapx(ii+1)*(w).^ii/factorial(ii);
end

linearoperatory = -alphay/2;
for ii = 0:length(betapy)-1;
  linearoperatory = linearoperatory - j*betapy(ii+1)*(w).^ii/factorial(ii);
end

u1x = u0x;
ufftx = fft(u0x); 

u1y = u0y;
uffty = fft(u0y);
  
fiberlength = nz*dz;
propagedlength =0;
auxlength = 0;
jj=1; % it will be used to choise the correct input angles, tetas, beta0s and beta1s

%  dz=0.1; % this is to set the longitudinal step-size

while propagedlength < fiberlength,
    

  %------the next "if" and the variable "auxlenght" are used to check if
  %the correlation length was riched
  if (auxlength >= correlationlength | auxlength ==0),
      auxlength = 0;
      
      angle = angles(jj);  
      teta = tetas(jj);
      ufftxaux = ufftx * cos(teta) - uffty * sin(teta)*exp(-i*angle);
      ufftyaux = ufftx * sin(teta)*exp(i*angle) + uffty * cos(teta);
      ufftx = ufftxaux;
      uffty = ufftyaux;
      
      betapx(1) = -beta0s(jj); % observe that b0x and b0y are defined such that b0x = -b0y = -beta0s(1)
      betapy(1) = beta0s(jj);
      betapx(2) = -beta1s(jj); % tha same as above for beta1x and beta1y
      betapy(2) = beta1s(jj);
      
      jj=jj+1; % I increase the value of jj so in the next time the 
               %correlation length is reached, different pseudo-random variables are taken
      
      linearoperatorx = -alphax/2; % in the lines bellow I define the linear operators again using the new beta0s and beta1s
      for ii = 0:length(betapx)-1;
        linearoperatorx = linearoperatorx - j*betapx(ii+1)*(w).^ii/factorial(ii);
      end
      linearoperatory = -alphay/2;
      for ii = 0:length(betapy)-1;
        linearoperatory = linearoperatory - j*betapy(ii+1)*(w).^ii/factorial(ii);
      end  
  end % here ends the "if" part that is done just at each correlation length.
  
    %Performing the SSFM with the given dz
  %------linear propagation until dz/2------------
  halfstepx = exp(linearoperatorx*dz/2);
  uhalfx = ifft(halfstepx.*ufftx); 
  
  halfstepy = exp(linearoperatory*dz/2);
  uhalfy = ifft(halfstepy.*uffty); 
  %-----------transforming into a CP basis------------------------------
  uhalfxaux = uhalfx;
  uhalfx = (uhalfxaux + j*uhalfy)/sqrt(2); 
  uhalfy = (uhalfxaux - j*uhalfy)/sqrt(2); 

  %------non linear propagation in dz------------
  u1x = uhalfx .* exp(-j*gamma*(2/3)*( (abs(uhalfx).^2) + (2*abs(uhalfy).^2) )*dz); % non linear propagation in CP basis
  u1y = uhalfy .* exp(-j*gamma*(2/3)*( (abs(uhalfy).^2) + (2*abs(uhalfx).^2) )*dz); % non linear propagation in CP basis
  
  %-----------transforming into a LP basis---------------------------------
  u1xaux = u1x; 
  u1x = (u1xaux + u1y)/sqrt(2); % transforming into a LP basis
  u1y = -j*(u1xaux - u1y)/sqrt(2); % transforming into a LP basis
  
  %------linear propagation from dz/2 to dz------------
  ufftx = halfstepx.*fft(u1x); % 
  uffty = halfstepy.*fft(u1y);
  %----------------------------------------------------
  auxlength = auxlength + dz;
  propagedlength = propagedlength + dz;
end

u1x = ifft(ufftx);
u1y = ifft(uffty);