I see two reasons why the number of q-point needed by ph.x could differ from the original one: 1) the symmetry of your crystal (the one relevant for the phonon calc.) is lower than the symmetry of your bravais lattice (the one generated by kpoints.x) ... use your q-points list as the k_points list in a scf calculation and see how many points pw.x generates... that is the list you should use in the ph.x calculation. How many symmetry operations are found by pw.x in the opriginal scf calculation ? Are you sure you provide enough digits in the atomic coordinates in order to have all expected symmetries ? 2) the test condition used to check whether two points are equivalent is tighter than the error due to truncation in the number of digits printed in the q-point list. This however would INCREASE the number of points not decrese its. In any case, to check this possibility you could modify kpoint.x so as to print a few more digits and see if this fix the problem. let me know stefano
wangjunjie1981_0 wrote: > Dear stefano > Thank you for your answer. you say the case I have noticed . As a > matter of fact the number of qpoints generating by kpoints.x is 64 > .but I don't know why the number of qpoints calculating in q2r.out > becomes 58 .Can you tell me what the cause is . Thank you ! > my kpoints.x file is > > 12 > 1 0.0000000 0.0000000 0.0000000 1.00 > 2 0.2500000 0.1443376 0.0000000 6.00 > 3 0.5000000 0.2886751 0.0000000 3.00 > 4 0.2500000 0.4330127 0.0000000 6.00 > 5 0.0000000 0.0000000 0.2340828 2.00 > 6 0.2500000 0.1443376 0.2340828 12.00 > 7 0.5000000 0.2886751 0.2340828 6.00 > 8 0.2500000 0.4330127 0.2340828 12.00 > 9 0.0000000 0.0000000 0.4681655 1.00 > 10 0.2500000 0.1443376 0.4681655 6.00 > 11 0.5000000 0.2886751 0.4681655 3.00 > 12 0.2500000 04330127 0.4681655 6.00 > > > ------------------------------------------------------------------------ > ??????????1.9?????????????????????? www.126.com <http://www.126.com/>
