permanent of square matrix

[John Keith]

(01-25-2024 03:49 PM)Albert Chan Wrote:  I guess it depends on programming language used.

...

Note: 2*d[j] is just a number, not a list

For LuaJIT, the cost is smaller still (OP 9×9 matrix, speedup under 5%)
Array access cost a lot more than multiply. Bit operations cost even more.
(this is just relatively speaking ... LuaJIT is FAST!)

2*d[j] is just a number, but it is multiplied by row j of the matrix, requiring n multiplies in every iteration for an n X n matrix.

Of course, the amount of improvement depends on the language and the hardware that the program is run on. Considering that RPL is an interpreted language running on an 80's era CPU (real or emulated) one would expect a greater degree of improvement than a compiled language running on a modern CPU. I have not tried the program on the Prime, but I expect that the degree of improvement would be somewhere in between.