How to quickly do the calculation?

Dear All,

I have a code to calculate an array using several arrays. The code is as follows:

Ieq1 = Haltmr(mmm(jbb2),IndBus(solvedbuses))*v(IndBus(solvedbuses));

Where Haltmr is a complex 6000x3000 sparse matrix, v is a complex 3000 column vector, mmm (for example, 20 rows in Haltmr) is a selected rows from Haltmr and jbb2 ( 5 rows in mmm) is the part of mmm, and IndBus is the order of columns in Haltmr, and solvebuses ( 15 indecies) is a part of the columns.

I am wondering if there is a faster way to do the calculation.

Thanks.

Benson

[Steven Lord]

That's impossible to answer without seeing what Haltmr, mmm, jbb2, IndBus, solvedbuses, and v are. Which of those are variables (and what sizes and types are they?) and for those that are functions what are those functions do?

[dpb]

All depends on what the definitions of the quantities are -- are they variables or functions? It's indeterminate in isolation.

[dpb]

sparse saves memory; the cost is execution speed to decode.

If you have enough memory, you could try moving to full storage; H is "only" 138 MB on machines these days with GB memory.

I didn't try to see if could do something with the matrix algebra...

Answer 1

[Matt J]

This might be faster.

r=false(6000,1); r(mmm(jbb2))=1;
c=false(3000,1); c(IndBus(solvedbuses))=1;

Ieq1=Haltmr*sparse(v.*c);
Ieq1=Ieq1(r);