[mat64] Quick way to divide by certain row

[Bert Chang]

Hi,

I'm wondering when using mat64.Dense, how can I quickly divide the whole matrix by certain row?

For example, I have a matrix

[10, 6]

[20, 18]

[35, 60]

[5, 3]

and I want to divide the first column by 5, and the second column by 3.

I've checked MulElem, which seems to require same shape of matrix ..

Or I can only use for loop to do it myself?

Thanks.

[Dan Kortschak]

I'd iterate taking a column view of each column and scale that view by your column factor.

untested:

v := []float64{.......}
_, c := m.Dims()
if len(v) != c {
panic(mat64.ErrShape)
}
for j, f := range v {
vec := m.ColView(j).RawVector()
for i := 0; i < len(vec.Data); i+= vec.Inc {
vec.Data[i] *= f
}
}

If Vector had a Scale method you could use that. It doesn't (it would essentially be the inner for loop). Feel free to file an issue for that and we will consider adding it.

[Dan Kortschak]

Alternatively, if you import (also untested)

v := []float64{.......}
r, c := m.Dims()

if len(v) != c {
panic(mat64.ErrShape)
}
for j, f := range v {
vec := m.ColView(j).RawVector()

blas64.Axpy(r, f, vec, vec)

[Dan Kortschak]

Sorry - this is clearly incorrect, and making it correct is not worth
the effort. Use the previous.

[Dan Kortschak]

... or not...

v := []float64{.......}
r, c := m.Dims()
if len(v) != c {
panic(mat64.ErrShape)
}
for j, f := range v {

blas64.Scal(r, f, m.ColView(j).RawVector())

[Chih-Wei Chang]

Got it!

Thanks, this one is very clear and elegant. Is there any chance we can make it an mat64's API?

Thanks.

-- Chih-Wei (Bert)

於 2015年2月21日 於 上午4:13:12, Dan Kortschak (dan.ko...@adelaide.edu.au) 寫:

[Dan Kortschak]

On Sat, 2015-02-21 at 09:55 +0800, Chih-Wei Chang wrote:
> Thanks, this one is very clear and elegant. Is there any chance we can
> make it an mat64's API?

Please file an issue on gonum/matrix.

[Jonathan Lawlor]

You can also use Mul as follows:

package main

import (

"fmt"

"github.com/gonum/matrix/mat64"

)

func main() {

A := mat64.NewDense(4, 2, []float64{

10, 6,

20, 18,

35, 60,

5, 3,

})

b := []float64{5, 3}

B := mat64.NewDense(2, 2, nil)

for i, v := range b {

B.Set(i, i, 1/v)

}

A.Mul(A, B)

fmt.Println(A)

}

I don't know which way would be faster.