profile
viewpoint
If you are wondering where the data of this site comes from, please visit https://api.github.com/users/nschloe/events. GitMemory does not store any data, but only uses NGINX to cache data for a period of time. The idea behind GitMemory is simply to give users a better reading experience.
Nico Schlömer nschloe Berlin, Germany Mathematics, numerical analysis, scientific computing, Python. Always interested in new problems. Contact me!

nschloe/awesome-scientific-computing 520

:sunglasses: Curated list of awesome software for numerical analysis and scientific computing

nschloe/betterbib 392

Update BibTeX files with info from online resources.

nschloe/colorio 118

Tools for color models

nschloe/dmsh 118

Simple mesh generator inspired by distmesh.

nschloe/accupy 67

Accurate sums and dot products for Python.

nschloe/blacktex 56

Cleans up your LaTeX files.

nschloe/betterspy 31

Visualization for SciPy sparse matrices.

nschloe/dufte 27

Clean matplotlib line plots

nschloe/color-data 26

Collected experimental color data.

nschloe/cplot 20

Color maps for complex-valued functions

startednschloe/termplotlib

started time in 29 minutes

startednschloe/perfplot

started time in 3 hours

fork vn-os/meshzoo

Some examples for mesh generation in Python

fork in 6 hours

startednschloe/meshzoo

started time in 6 hours

startednschloe/krylov

started time in 7 hours

PR opened nschloe/meshio

Nastran long support

I've added a new test with a valid Nastran grid in long format that currently fails. The issue is that chunks2[1:3]] in the code below will only be length 1 since the second line in the test grid card is not long enough (not more than 16 characters long). This leads leads to ValueError: not enough values to unpack (expected 2, got 1).

            chunks2 = _chunk_line(next_line)
            next_line = f.readline()
            points.append(
                [
                    _nastran_string_to_float(i + j)
                    for i, j in [chunks[5:7], chunks[7:9], chunks2[1:3]]
                ]
            )

The solution I've added here is to simply use 72 for the line length instead of len(line), which will result in the fixed format line "chunks" always having 9 fields. It fills the empty ones with empty strings, ''.

+17 -1

0 comment

2 changed files

pr created time in 8 hours

fork Spectre5/meshio

input/output for many mesh formats

fork in 8 hours

startednschloe/meshio

started time in 9 hours

startednschloe/dufte

started time in 10 hours

startednschloe/pytest-codeblocks

started time in 11 hours

startednschloe/quadpy

started time in 11 hours

startednschloe/optimesh

started time in 12 hours

startednschloe/meshio

started time in 13 hours

startednschloe/stressberry

started time in 13 hours

startednschloe/pytest-codeblocks

started time in 21 hours

startednschloe/pytest-codeblocks

started time in a day

PR opened nschloe/tikzplotlib

Add test case for issue #453

I've starting debugging #453 and found the place where the bug happens, but I don't know how to fix it.

https://github.com/nschloe/tikzplotlib/blob/dc919b3c538544ecf4b3d86deaf2cb9b80f0af80/tikzplotlib/_util.py#L9

 def get_legend_text(obj): 
    """Check if line is in legend."""
    leg = obj.axes.get_legend()
    if leg is None:
        return None

    # leg.legendHandles = [<matplotlib.collections.LineCollection object at 0x7fc13a90d278>, <matplotlib.collections.LineCollection object at 0x7fc13a90d780>]
    keys = [h.get_label() for h in leg.legendHandles if h is not None] # = ['_nolegend_', '_nolegend_']
    values = [t.get_text() for t in leg.texts] # = ['y1', 'y2']

    # obj = <matplotlib.lines.Line2D object at 0x7fc13c957dd8>
    label = obj.get_label() # = '_nolegend_'
    d = dict(zip(keys, values)) # = {'_nolegend_': 'y2'}
    if label in d:
        return d[label] # Both return the same label

    return None

Could you give guidance hints on how to proceed? I think I need to either get from the Line2D object to the LineCollection or vice versa to check if the Line2D is included in the one of the LineCollections.

+98 -0

0 comment

2 changed files

pr created time in a day

fork mb/tikzplotlib

Convert matplotlib figures to TikZ/PGFplots for smooth integration into LaTeX.

https://pypi.org/project/tikzplotlib

fork in a day

pull request commentnschloe/awesome-scientific-computing

Adding Kratos Multiphysics

Thanks for the feedback!

loumalouomega

comment created time in a day

startednschloe/pynosh

started time in a day

startednschloe/optimesh

started time in a day

startednschloe/quadpy

started time in a day

startednschloe/perfplot

started time in a day

startednschloe/awesome-scientific-computing

started time in a day

startednschloe/awesome-scientific-computing

started time in a day

pull request commentnschloe/awesome-scientific-computing

Adding Kratos Multiphysics

I just checked out Kratos. What's awesome:

* It's open source.

* It's on GitHub.

* It's got a Python interface.

What's not awesome:

Installation:

* No native packages (Debian, Red Hat,...)

We are working in the pypy packages...

* 1.73 GiB git check out (???)

Many developments, we need to perform some clean up.

* Had to abort compilation after 10+ mins at only 17% progress.

Yes, but many large projects take long to compile.

Use:

* There's a Python interface, but it mostly looks like C++ code, e.g.,
  ```python
  gid_output.ExecuteInitializeSolutionStep()
  ## The actual solution goes here
  if gid_output.IsOutputStep():
     # Call ExecuteBeforeOutputStep on any auxiliar processes you have here
     gid_output.PrintOutput()
     # Call AfterBeforeOutputStep on any auxiliar processes you have here
  gid_output.ExecuteFinalizeSolutionStep()
  gid_output.ExcecuteFinalize()
  ```
  
  
  (From [here](https://github.com/KratosMultiphysics/Kratos/wiki/Python-Script-Tutorial:-Writing-Output-File)).

That's because we focus to make one consistent interface between C++ and python, like Qt does (being less pythonic)

More:

* 100+ contributors, but <600 gh stars?

Mostly researchers, not very well known yet

* bundling of third-party libraries (admittedly other packages do that, too, but it's not awesome)

Do you prefer something like PETSc, where third party are automatically downloaded?

I'm inclined not to merge it.

Completely fair, I just want it to add my two cents.

loumalouomega

comment created time in a day

startednschloe/meshplex

started time in a day

fork loumalouomega/awesome-scientific-computing

:sunglasses: Curated list of awesome software for numerical analysis and scientific computing

fork in a day

startednschloe/awesome-scientific-computing

started time in a day