The problem is: the program is taking too long to calculate a pseudosection, like four to five days, and the results are not understandable (see figure for an example). I've tried using the dataset tcds_55 (with the necessary changes in the therin input file) and the program works just fine (five to eight hours to generate a pseudosection) and the pseudosections are understandable.
WhatsApp Image 2020-08-03 at 09.08.24.jpeg (192.61 KiB) Viewed 2071 times
I suggest you have a try the recently updated version 20201103, which can be available from the URL below: https://titan.minpet.unibas.ch/minpet/t ... ruser.html. I have attained some good results using this version when constructing phase diagrams for a mafic eclogite, so i believe it may be helpful to you as well.
In addition, if you want to omit some compositions such as TI and MN, you can just delete them directly in your input datafile. Like this: SI(61.59)AL(17.63)FE(4.23)MG(2.93)CA(1.57)NA(4.47)K(4.21)H(3.37)O(?)O(0.01)