Differences in SOLVE output between MATLAB version 2010b and 2012a

Dear Users

The title says it all really. I am using SOLVE to solve a system of 12 equations with 12 unknown variables. My unknowns are declared as;

z11=sym('z11','real')
z12=sym('z12','real')
z21=sym('z21','real')
z22=sym('z22','real')
sig1=sym('sig1','positive')
sig2=sym('sig2','positive')
sigu1=sym('sigu1','positive')
sigu2=sym('sigu2','positive')
sig1_h=sym('sig1_h','positive')
sig2_h=sym('sig2_h','positive')
sigu1_h=sym('sigu1_h','positive')
sigu2_h=sym('sigu2_h','positive')

I am using SOLVE as follows

S = solve('sigu1*(1+z11*z11) + sig1     = sigu1_h*(1+z11*z11) + sig1_h',...
          'sigu1*(1+z12*z12) + sig1     = sigu1_h*(1+z12*z12) + sig1_h',...
          'sigu1*(1+z12*z11)            = sigu1_h*(1+z12*z11)',...
          'sigu2*(1+z11*z11) + sig2     = sigu2_h*(1+z11*z11) + sig2_h',...
          'sigu2*(1+z12*z12) + sig2     = sigu2_h*(1+z12*z12) + sig2_h',...
          'sigu2*(1+z12*z11)            = sigu2_h*(1+z12*z11)',...          
          'sigu1*(1+z21*z21) + sig1     = sigu1_h*(1+z21*z21) + sig1_h',...
          'sigu1*(1+z22*z22) + sig1     = sigu1_h*(1+z22*z22) + sig1_h',...
          'sigu1*(1+z22*z21)            = sigu1_h*(1+z22*z21)',...
          'sigu2*(1+z21*z21) + sig2     = sigu2_h*(1+z21*z21) + sig2_h',...
          'sigu2*(1+z22*z22) + sig2     = sigu2_h*(1+z22*z22) + sig2_h',...
          'sigu2*(1+z22*z21)            = sigu2_h*(1+z22*z21)')

This system always has a solution when sig1=sig1_h, sig2=sig2_h, sigu1=sigu1_h, sigu2=sigu2_h, regardless of what z11, z12, z21 and z22 are set to (although I am not really interested in this particular solution).

In MATLAb version 2010b there are 17 solutions (including the one described above), but in Matlab 2012a I get the warning that no explicit solutions can be found. I noticed in the 2012a documentation that sometimes the "==" symbol is used instead of "=". However changing my code to use this makes no difference. I cannot see any other reasons why I need to change my code for 2012a.

Does anybody have any thoughts as to why 2012a is not giving me the same results, and/or how to get it to give me the correct results if I am using it incorrectly?

Any help would be greatly appreciated.

For information I am running both MATLAB versions concurrently on a 32-bit Windows Vista machine.

[Daniel]

I have been informed this is a bug in version 2012. I will keep everybody posted on progress regarding fixing it, or perhaps a workaround.

[Salvador C. Cardona Navarrete]

Hello Daniel:

I have detected some similar bug between Matlab 2010b and Matlab 2013b. Do you know if Mathworks solved the bug you detected?.

Salva