Mathematical notation has been developed over the centuries, focused on a 2-dimensional presentation media (clay, blackboard, paper) with the philosophy that no name should be more than one glyph. This is convenient if you have to write long derivations by hand, but has led to a search for all kind of strange symbols as more names where needed: Latin characters, Greek characters, Hebrew characters, caligraphic characters and so forth. In addition to this insistment of "every character by itself", there is also the use of "icons" such as (pd)/(pdx) which is an operator "partial derivative in the 1st space coordinate" and SUMi=efg which is a folding operator declaring a local context where i is an integer variable being bound in g, e and f are integer expressions and g is some numerical expression. Mathematical notation is well developed for the parts of mathematics with the longest history, adapted to the kind of computational technology one had at the time of its development, i.e., hand calculation. It is highly ambiguous (you have to know what is meant to be able to parse it), lacks a tradition for declarations, overloads notation heavily, but is rooted in a semantical understanding of the underlying concepts.
In contrast programming language notation was forced to use a limited character set (such as those on IBM punched cards) and had to have a fairly linear presentation (linebreaks and punched column positions as in old ForTran-IV is still linear). An early understanding of the block nature of program structure motivated context free grammars as the preferred syntactic apparatus. Semantic issues of programming languages have to a large extent been left in the dark.