# How does broadcast! work ? Tips for memory efficiency while using 'kron'

**URL:** <https://discourse.julialang.org/t/how-does-broadcast-work-tips-for-memory-efficiency-while-using-kron/10957>\
**Category:** Performance\
**Tags:** question\
**Created:** [May 17, 2018, 2:33pm UTC](https://discourse.julialang.org/t/how-does-broadcast-work-tips-for-memory-efficiency-while-using-kron/10957 "2018-05-17T14:33:39Z")\
**Posts on this page:** 2\
**Page:** 1

<div class="post-metadata">

**Author:** ![amitjamadagni](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/amitjamadagni/32/3570_2.png) [@amitjamadagni](https://discourse.julialang.org/u/amitjamadagni)\
**Post date:** [May 17, 2018, 2:33pm UTC](https://discourse.julialang.org/t/how-does-broadcast-work-tips-for-memory-efficiency-while-using-kron/10957/1 "2018-05-17T14:33:39Z")

</div>

I was trying to use `broadcast!` in place of `broadcast` in the following code :

```julia
function _tensor_product_(st)
    ten = st[1]
    for s in st[2:end]
        ten = kron(ten, s)
    end
    return ten
end

@noinline function indi(arr, ind1, ind2, no_of_e)
    linit = [speye(2) for e=1:no_of_e]
    op = spzeros(2^no_of_e, 2^no_of_e)
    op1 = spzeros(2^no_of_e, 2^no_of_e)
    for elem in arr[ind1:ind2]
        println(" elem ", elem)
        for e in elem
            linit[e] = SparseMatrixCSC([0. 1.;1. 0.])
        end
        tp = _tensor_product_(linit)
        op += tp
        # op1 = broadcast(+, op1, tp)
        broadcast!(+, op1, op1, tp)
        # resetting the configuration for next iteration
        for i=1:length(linit)
            linit[i] = speye(2)
        end
        println(op1 == op)
        gc()
    end
    return op
end

r = [[1, 3, 6, 4], [2, 4, 7, 5], [6, 8, 1, 9], [7, 9, 2, 10]]
indi(r, 1, 4, 14)

```

The behavior of `broadcast` and `broadcast!` are not the same, while the former works as I expect but the latter does not replicate the former (was trying to do it in place).

Also, if I have the following the memory shoots up, one way to tackle is to use `mmap` and store it to disk as a tensor product is being performed. So how to use `mmap` in conjugation with sparse matrices and also are there any general methods to optimize the above routine for memory efficiency ?

```julia
# a = [[1, 3, 6, 4], [2, 4, 7, 5], [6, 8, 11, 9], [7, 9, 12, 10], [11, 13, 16, 14], [12, 14, 17, 15], [16, 18, 1, 19], [17, 19, 2, 20]]
# indi(a, 1, 4, 28)

```

Thanks !

---

<div class="post-metadata">

**Author:** ![kristoffer.carlsson](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/kristoffer.carlsson/32/22_2.png) [@kristoffer.carlsson](https://discourse.julialang.org/u/kristoffer.carlsson)\
**Post date:** [May 17, 2018, 2:55pm UTC](https://discourse.julialang.org/t/how-does-broadcast-work-tips-for-memory-efficiency-while-using-kron/10957/2 "2018-05-17T14:55:40Z")

</div>

Probably [https://github.com/JuliaLang/julia/issues/21693](https://github.com/JuliaLang/julia/issues/21693). Should be fixed on master by [https://github.com/JuliaLang/julia/pull/25890](https://github.com/JuliaLang/julia/pull/25890).