With the rules
#T(x_1,x_2,x_3) = (x_1-x_3,x_2-x_3,x_1)# we can define
#T = ((1,0,-1),(0,1,-1),(1,0,0))#
Assuming #S# as a linear operator such that
#S(x_1-x_3,x_2-x_3,x_1) = (x_1,x_2,x_3)# due to the linearity of #S# we have
#{(alpha_1(x_1-x_3)+alpha_2(x_2-x_3)+alpha_3 x_1 = x_1),(beta_1(x_1-x_3)+beta_2(x_2-x_3)+beta_3 x_1 = x_2),(gamma_1(x_1-x_3)+gamma_2(x_2-x_3)+gamma_3 x_1 = x_3):}#
or
#{((alpha_1+alpha_3)x_1+alpha_2 x_2 +(alpha_3-alpha_1-alpha_2)x_3 = x_1),
((beta_1+alpha_3)x_1+beta_2 x_2 +(beta_3-beta_1-beta_2)x_3 = x_2),((gamma_1+gamma_3)x_1+gamma_2 x_2 +(gamma_3-gamma_1-gamma_2)x_3 = x_3):}#
then solving the systems
#{(alpha_1+alpha_3=1),(alpha_2=0),(alpha_3-alpha_1-alpha_2=0):}#
#{(beta_1+beta_3=0),(beta_2=1),(alpha_3-alpha_1-alpha_2=0):}#
#{(gamma_1+gamma_3=0),(gamma_2=0),(gamma_3-gamma_1-gamma_2=1):}#
we obtain
#S =((alpha_1,alpha_2,alpha_3),(beta_1,beta_2,beta_3),(gamma_1,gamma_2,gamma_3))= ((0,0,1),(-1,1,1),(-1,0,1)) = T^-1#