# MTK & units

**URL:** <https://discourse.julialang.org/t/mtk-units/133795>\
**Category:** Modelling & Simulations\
**Created:** [November 11, 2025, 1:55pm UTC](https://discourse.julialang.org/t/mtk-units/133795 "2025-11-11T13:55:45Z")\
**Posts on this page:** 2\
**Page:** 1

<div class="post-metadata">

**Author:** ![BLI](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/bli/32/37206_2.png) [@BLI](https://discourse.julialang.org/u/BLI)\
**Post date:** [November 11, 2025, 1:55pm UTC](https://discourse.julialang.org/t/mtk-units/133795/1 "2025-11-11T13:55:45Z")

</div>

In the past, I have more or less converted units manually, but have now started to use `Unitful.jl` – which also ties into `Plots.jl` and gives convenient typesetting of units along the axes.

To me, there are at least 3 reasons why units are useful:

1. Correct handling of units and unit conversion.
2. Checking a model for unit consistency.
3. Simple inclusions of units in plots.

General allowance for units in quantities carries some (considerate) computational overhead in `ModelingToolkit.jl`. Because of that, units may be specified via the `unit` construct:

```julia
params = @parameters begin
        # Constant
        mb = 14, [unit=u"u", description = "Nitrogen atom mass"]
        ... 
        #
    end

```

My understanding is that this support is limited to SI units (base SI units only?), and that units may be used for checking unit consistency - which is great (reason 12). Perhaps units are also included along `Plots.jl` axes via the recipes (reason 3)?

So the `unit` construct in MTK partially solves my problem by at least addressing reason 2. However, if only base SI units are allowed, this will probably make the model stiff? Say I am modeling atomic interaction where the natural time unit is `ps`. If I am forced to use unit `s`, that leads to clumsy numbers along the axes.

* * *

Next, consider the case that I am mainly (at the moment) interested in reasons 1 and 3 above.

**Question 1** :  
To handle conversion of units, etc., I assume that I can do as follows:

```julia
params = @parameters begin
        # Constant
        mb = ustrip(uconvert(u"g", 14u"u")), [description = "Nitrogen atom mass"]
        ... 
        #
    end

```

and that this will not have a detrimental effect on the solution time - even if I use `Unitful.jl`?

**Question 2** :  
How can I handle automatic addition of units along the axes? Suppose I have instantiated an MTK model with name `tp_h` and solved this model, with solution `sol_h` where one of the unknowns is `x_1`

Doing

```julia
plot(sol_h, idxs=(tp_h.t, tp_h.x_1))

```

plots this result without units.

However, I don’t see how I can add units here. After all, `tp_h.t` is an index, so I can not write `tp_h.t*u"ps"` - as an example.

I can, of course, create a time vector manually, and add units:

```julia
tm_vec = collect(range(0,0.1,length=100))
x_1 = sol_h.(tm_vec, idxs=tp_h.x_1)

plot(tm_vec*u"ps", x_1*u"nm")

```

but this way I don’t utilize the elegant recipe of MTK.

Any way to combine MTK recipe with adding units?

---

<div class="post-metadata">

**Author:** ![ChrisRackauckas](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/chrisrackauckas/32/77_2.png) [@ChrisRackauckas](https://discourse.julialang.org/u/ChrisRackauckas)\
**Post date:** [November 11, 2025, 6:14pm UTC](https://discourse.julialang.org/t/mtk-units/133795/2 "2025-11-11T18:14:21Z")

</div>

> [@BLI](#):
>
> In the past, I have more or less converted units manually, but have now started to use `Unitful.jl` – which also ties into `Plots.jl` and gives convenient typesetting of units along the axes.

We will be removing Unitful in the very near future. Unitful just gives so many compilation problems that it cannot be well-supported. We’ll keep a certain amount of DynamicQuantities.jl support, but for the most part the unit support is still very basic.

> [@BLI](#):
>
> My understanding is that this support is limited to SI units (base SI units only?), and that units may be used for checking unit consistency - which is great (reason 12). Perhaps units are also included along `Plots.jl` axes via the recipes (reason 3)?

Yes, codegen cannot at this time add conversion factors.

> [@BLI](#):
>
> How can I handle automatic addition of units along the axes? Suppose I have instantiated an MTK model with name `tp_h` and solved this model, with solution `sol_h` where one of the unknowns is `x_1`

At this time no.