Float * dual doesnt work

[huand]

// 2.0 * Dual::from_re(0.0); //doesnt work
        Dual::from_re(0.0) * 2.0; //works

is there a reason for this?
doesnt this makes some operation like matrix of scalar multiplication on the left side impossible with a vec of duals?

[prehner]

Hi, thank you for the question. You are right, at the moment, we implement Mul<f64> for Dual but not Mul<Dual> for f64. On the level of individual dual numbers, we should actually be able to rectify that straightforwardly by implementing the necessary traits.

I guess, the reason we haven't added those implementations so far, is that we often use dual numbers in a generic way, e.g., with the DualNum trait and I am not sure, whether we could set "right-hand side" operations as super traits of DualNum. I'll have a look into that.

You are of course that this would limit some matrix operations with dual numbers, but at the same time, the issue can always be rectified by converting the f64 data structures to dual numbers first. If this is not within the innermost hot loop, the performance penalty should be fine, but for certainty, you would need to benchmark it.