Petar Mlinarić

Here are the environment variables currently listed in `environment.rst`: ```bash $ sed -n -e 's/^.*\(PYMOR_[A-Z_]\+\).*$/\1/p' docs/source/environment.rst | sort | uniq PYMOR_CACHE_DISABLE PYMOR_COLORS_DISABLE PYMOR_CONFIG_DISABLE PYMOR_DEFAULTS PYMOR_FIXTURES_DISABLE_BUILTIN PYMOR_HYPOTHESIS_PROFILE PYMOR_MPI_FINALIZE PYMOR_WITH_SPHINX ``` and...

Regarding the used environment variables: - `PYMOR_CODECOV_TOKEN` is only used in `.ci/gitlab/coverage_submit.bash` - `PYMOR_COMMIT` is only used by `.ci/gitlab/docs_deploy.bash` to edit `docker/Dockerfile.binder.tocopy` - `PYMOR_SUDO` is only mentioned in developer docs...

Here is again the status (now excluding `environment.rst` in the second one): ```bash $ sed -n -e 's/^.*\(PYMOR_[A-Z_]\+\).*$/\1/p' docs/source/environment.rst | sort | uniq PYMOR_CACHE_DISABLE PYMOR_COLORS_DISABLE PYMOR_CONFIG_DISABLE PYMOR_DEFAULTS PYMOR_FIXTURES_DISABLE_BUILTIN PYMOR_HYPOTHESIS_PROFILE PYMOR_MPI_FINALIZE...

Thanks @sdrave

I tried now also using `lu_factor` and `lu_solve`, and it seems that makes the difference.  Here is the code (which was run with wheels and `OMP_NUM_THREADS=1`). ```python import numpy...

Here are also timings relative to NumPy.  Updated plotting code: ```python plot_opts = dict(logx=True) out.save('numpy_vs_scipy_abs.png', **plot_opts) out.save('numpy_vs_scipy_rel.png', relative_to=0, **plot_opts) out.show(**plot_opts) ```

@artpelling Any idea what is the "right" definition of the impulse response for discrete-time systems? In the Antoulas book, the sampling time is ignored (as far as I understand), but...

@artpelling What do you think time-horizon `T` should represent for discrete-time systems? Should it be the number of steps (i.e., `solve` should return $x_0, x_1, \ldots, x_T$) or time in...

> > @artpelling Any idea what is the "right" definition of the impulse response for discrete-time systems? In the Antoulas book, the sampling time is ignored (as far as I...

> > @artpelling What do you think time-horizon `T` should represent for discrete-time systems? Should it be the number of steps (i.e., `solve` should return x0,x1,…,xT) or time in seconds...