Petar Mlinarić

> > It shouldn't be a problem really, just lets me think about whether its cleaner to manually do the seeding when necessesary and then only pass random_state to the...

Option 1 sounds good to me (maybe 'square' instead of 'pow(2)').

This seems to be working now after rebasing on main with #1469 merged. @sdrave The log message from `output` looks a bit strange: ``` LTIModel: Solving LTIModel for {input: {x...

Ok, thanks, I'll take a look at #1305. > The intention is to separate the time steps taken by the time stepper and the length of the solution trajectory returned...

> > This is still a bit unclear to me. Wouldn't you always apply POD to the snapshots and pick a fixed number of modes from there? > > Not...

I rebased on `main` and fixed the tutorial example. I forgot how `input` works for `Models`, so I want to look again through the code. Also, I still didn't implement...

Hi, thank you for opening an issue. Do the methods in [`pymor.algorithms.svd_va`](https://github.com/pymor/pymor/blob/f357630387d6fe3765f2911830283cafe65acd54/src/pymor/algorithms/svd_va.py) work for you?

Yes, that is by design, and maybe the expectation for RB methods. @sdrave is more familiar with this.

How about adding a helper function for constructing Loewner matrices in `pymor.reductors.interpolation` instead of an `Operator`? We discussed a bit some fast approximate matrix-vector methods, but it didn't sound like...

@lbalicki I now also moved Gramian computation for FDBT to `LTIModel` (with a bunch of other changes). Does it look ok to you?