{concave} - {concave}

1

Why I can’t use the formula? please help me to solve it!

Error:
g1

f1

Disciplined convex programming error: Illegal operation: {concave} - {concave}

error

Code:
[M1(I1),g1(I1)]=calcMaxVal_group1(K,N,L,WN,p1(:,I1),elta1(o),H,re11,v,c1,c2,alpha);
while 1
[g1_diff]=g1_firstorderdiff(K,N,L,WN,p1(:,I1),alpha,H,re11);
cvx_begin
variable p_opt1(2*U)
expression tempRate1 %求f
expression tempRate_g1 %求g
expression f1
expression E1
expression M1

          % 计算f1和g1
          for l=1:L % 2
            for k=1:K  %12 组
              for wn=1:WN %2
                 for n=1:N  % 2 
                    pkn1=sum(p_opt1((l-1)*K*N+(k-1)*N+1:(l-1)*K*N+(k-1)*N+n)); 
                    if n+1<=N  
                        for nv=n+1:N  
                           pkn_gp1=sum(p_opt1((l-1)*K*N+(k-1)*N+(n+1):(l-1)*K*N+(k-1)*N+nv)); 
                        end
                    end
                 end 
                 tempRate1=tempRate1+log(alpha*(H^2+re11((l-1)*K*WN+(k-1)*WN+wn))+pkn1)/log(2); 
                 tempRate_g1=tempRate_g1+log(alpha*(H^2+re11((l-1)*K*WN+(k-1)*WN+wn))+pkn_gp1)/log(2); 
              end
            end  
            E1=E1+c1*pow_pos(norm(v(l)),3)+c2*inv_pos(v(l,1)); 
          end
          f1=tempRate1-elta1(o)*E1;
          g1=tempRate_g1+g1_diff*(p_opt1-p1(:,I1));
          M1=f1-g1;
2

f1 and g1 are both concave. Hence the error message for {concave} - {concave}, which is not allowed by CVX"s DCP rules.

I don’t know what you intend to do with Mq, so perhaps there is a way of writing the code which does not require forming M1.

Your first step is to probe your optimization problem is convex.