3
$\begingroup$

I need to multiply 3 lists of matrices (b.a.b) as the following code

a = {{{-17.8227277373099`, -1.6565234964560602`}, 
{-1.6565234954649242`, 
 5.298073701591974`}}, {{-17.812203521929003`, 
-1.5013126607114478`}, {-1.5013126574896714`, 
 4.384050851253119`}}, {{-17.801677045750512`, 
 -1.4055541329078751`}, {-1.405554138172727`, 
 3.869511752542245`}}};

b = {{{0.8409518416651456`, 0}, {0, 
 0.1274293000222242`}}, {{0.8409815693580924`, 0}, {0, 
 0.14187218616724442`}}, {{0.841011296000238`, 0}, {0, 
 0.15290209433231844`}}};

I used the following:

  mat = b.a.b

But, I think this way is not correct because I didn't get the result. Also, I make a test for the first matrices as the following code, that what I wanted to get for the whole matrices multiplication.

 a1 = {{-17.8227277373099`, -1.6565234964560602`}, 
{-1.6565234954649242`, 5.298073701591974`}};

b1 = {{0.8409518416651456`, 0}, {0, 0.1274293000222242`}};

 mat1 = b1.a1.b1


 {{-12.6042, -0.177516}, {-0.177516, 0.0860313}}

Thanks.
$\endgroup$
2
  • 1
    $\begingroup$ Table[b[[i]].a[[i]].b[[i]], {i, 1, 3}] or #[[2]].#[[1]].#[[2]] & /@ Transpose[{a, b}]. $\endgroup$
    – corey979
    Mar 18, 2019 at 21:06
  • $\begingroup$ Thanks a lot corey979! $\endgroup$
    – Ghady
    Mar 19, 2019 at 0:30

2 Answers 2

8
$\begingroup$

Using MapThread and Dot:

MapThread[Dot, {b, a, b}]
$\endgroup$
1
  • $\begingroup$ Thank you very much, It is the easiest! $\endgroup$
    – Ghady
    Mar 19, 2019 at 0:33
6
$\begingroup$

If you need it really fast, then use Compile:

n = 1000000;
a = RandomReal[{-1, 1}, {n, 2, 2}];
b = RandomReal[{-1, 1}, {n, 2, 2}];

cf = Compile[{{a, _Real, 2}, {b, _Real, 2}},
   b.a.b,
   RuntimeAttributes -> Listable,
   Parallelization -> True
   ];

MapThread[Dot, {b, a, b}]; // AbsoluteTiming // First
cf[a, b]; // AbsoluteTiming // First

1.46343

0.075722

$\endgroup$
2
  • $\begingroup$ Thank you very much! $\endgroup$
    – Ghady
    Mar 19, 2019 at 0:31
  • $\begingroup$ You're welcome. $\endgroup$ Mar 19, 2019 at 8:00

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.