Follow-up Comment #10, bug #22021 (project relax): Andrew Baldwin has sent these following lines of code, which shows that I have implemented the function wrong. ###### optA='y' if(optA=='y'): sqrt_zeta2_Psi2 = numpy.sqrt(zeta**2 + Psi**2) g3 = g_fact * numpy.sqrt( Psi + sqrt_zeta2_Psi2) g4 = g_fact * numpy.sqrt(-Psi + sqrt_zeta2_Psi2) else: quad_zeta2_Psi2 = (zeta**2 + Psi**2)**0.25 g3=cos(0.5*atan2(-zeta,Psi))*quad_zeta2_Psi2 #trig faster than square roots g4=sin(0.5*atan2(-zeta,Psi))*quad_zeta2_Psi2 #trig faster than square roots ####### with the following comments: #### #R2eff = R2eff - 1/Trelax * numpy.log(Tog.real) #NOT CORRECT # Fastest calculation. R2eff = (R2g + R2e + kex) / 2.0 - 1/(Trelax) * ( ncyc * numpy.arccosh(v1c.real) + numpy.log(Tog.real) ) #NOT CORRECT - error equal to previous #### I base from these comments, that Andrew explicit suggest the trigometric implementations. This will be corrected, after testing the _______________________________________________________ Reply to this item at: <http://gna.org/bugs/?22021> _______________________________________________ Message sent via/by Gna! http://gna.org/