"I have some variables .mat named new_x_edges_50 , new_y_edges_50, new_z_edges_50, new_x_edges_60, new_y_edges_60, new_z_edges_60, new_x_edges_70, new_y_edges_70, new_z_edges_70 and so on... "
Ah, so you forced meta-data into your variable names. Well, that will make processing your data much harder, if you then try to access those variable names dynamically...
"In this code I would like that the user has the possibility to say that If z_start =70 (instead of z_start = 50 like in this code), then in dist_s I have to use new_x_edges_70, new_y_edges_70 and new_z_edges_70 (instead of new_x_edges_50, new_y_edges_50, new_z_edges_50), then xts = new_x_edges_70(ids) yts = new_y_edges_70(ids) zts = new_z_edges_70(ids)."
... like that.
Meta-data is data, and data belongs in variables, not in variable names. Once you do that, then the solution to your question will be simple (e.g. indexing, fieldnames, etc). But as posed, your data design forces you into writing slow, complex, inefficient, insecure, obfuscated code which is buggy and difficult to debug:
So far you have not told us the most important information: how did you get all of those variables into the workspace? That would be correct place to fix your code, e.g. by placing all of those arrays into one cell array. After that, you can use a simple numeric vector and some basic indexing to achieve your task (and much neater, simpler, and more efficient than you current approach using numbered variable names).
"How to use the command window in order to make a script faster..."
By using arrays and indexing, not by accessing variable names dynamically.
