# Type inference in constructors
**URL:** https://discourse.julialang.org/t/type-inference-in-constructors/11592
**Category:** New to Julia
**Created:** [June 11, 2018, 5:44pm UTC](https://discourse.julialang.org/t/type-inference-in-constructors/11592 "2018-06-11T17:44:02Z")
**Posts on this page:** 6
**Page:** 1
### Author:  [@Pavel\_Kalouguine](https://discourse.julialang.org/u/Pavel_Kalouguine)
#### Post date: [June 11, 2018, 5:44pm UTC](https://discourse.julialang.org/t/type-inference-in-constructors/11592/1 "2018-06-11T17:44:02Z")
I am trying to create a type representing crystallographic space groups in arbitrary dimensions. The group operations are stored as static matrices of integers (plus eventually a static rational vector for non-symmorphic groups, but this part is not done yet). Here’s the code:
```julia
struct SpaceGroup{N, N2}
e2i::Dict{SMatrix{N,N,Int,N2},Int}
i2e::Vector{SMatrix{N,N,Int,N2}}
function SpaceGroup{N,N2}(g::Vector{SMatrix{N,N,Int,N2}}) where {N,N2}
sg=new(Dict{SMatrix{N,N,Int,N2},Int}(), Vector{SMatrix{N,N,Int,N2}}())
# More initialization here...
return sg
end
end
```
The inner constructor takes a vector of static matrices (group generators), with explicitely supplied type parameters. The question is: is it possible to infer these parameters from the type of the constructor argument, instead of passing them explicitely as in the snippet below?:
```julia
g=[@SMatrix [0 1 0 0; 0 0 1 0; 0 0 0 1; -1 0 0 0]]
sg=SpaceGroup{4,16}(g) # The type of g fixes the values of N and N2, does it?
```
Thanks.
---
### Author:  [@Elrod](https://discourse.julialang.org/u/Elrod)
#### Post date: [June 11, 2018, 6:05pm UTC](https://discourse.julialang.org/t/type-inference-in-constructors/11592/2 "2018-06-11T18:05:10Z")
```julia
julia> SpaceGroup(g::Vector{SMatrix{N,N,Int,N2}}) where {N,N2} = SpaceGroup{N,N2}(g)
SpaceGroup
julia> SpaceGroup(g)
SpaceGroup{4,16}(Dict{StaticArrays.SArray{Tuple{4,4},Int64,2,16},Int64}(), StaticArrays.SArray{Tuple{4,4},Int64,2,16}[])
```
---
### Author:  [@Pavel\_Kalouguine](https://discourse.julialang.org/u/Pavel_Kalouguine)
#### Post date: [June 11, 2018, 6:28pm UTC](https://discourse.julialang.org/t/type-inference-in-constructors/11592/3 "2018-06-11T18:28:57Z")
Wow, that was quick! Thank you, this is exactly what I wanted.
---
### Author:  [@Tamas\_Papp](https://discourse.julialang.org/u/Tamas_Papp)
#### Post date: [June 12, 2018, 6:52am UTC](https://discourse.julialang.org/t/type-inference-in-constructors/11592/4 "2018-06-12T06:52:52Z")
Also note [the section on parametric constructors in the manual](https://docs.julialang.org/en/latest/manual/constructors/#Parametric-Constructors-1), especially the example following
> This automatic provision of constructors is equivalent to the following explicit declaration:
---
### Author:  [@chobbes](https://discourse.julialang.org/u/chobbes)
#### Post date: [June 12, 2018, 9:39am UTC](https://discourse.julialang.org/t/type-inference-in-constructors/11592/5 "2018-06-12T09:39:22Z")
> [@Tamas\_Papp](#):
>
> Also note [the section on parametric constructors in the manual](https://docs.julialang.org/en/latest/manual/constructors/#Parametric-Constructors-1), especially the example following
>
> > This automatic provision of constructors is equivalent to the following explicit declaration:
To be honest, the section of parametric constructor could have been largely augmented to provide more elaboration and examples. I found the information in that section insufficient every time when I looked for something more than the very basic.
---
### Author:  [@Tamas\_Papp](https://discourse.julialang.org/u/Tamas_Papp)
#### Post date: [June 12, 2018, 11:59am UTC](https://discourse.julialang.org/t/type-inference-in-constructors/11592/6 "2018-06-12T11:59:16Z")
> [@chobbes](#):
>
> section of parametric constructor could have been largely augmented to provide more elaboration and examples
I agree, the best way to resolve that is an issue or a PR.