matmodlab / matmodlab2 Goto Github PK

Looking at core/matmodlab.py, it seems that J0 is initially evaluated during startup but never updated (with a numerical jacobian or with the returned jacobian). Also curious, the returned DDSDDE seems to always be discarded (whether it is None or a full matrix).

Material name should also be changed to PolynomialHyperelastic

When you prescribe the deformation gradient, it's immediately recast in terms of strain (defined by the current kappa) and then that strain is linearly interpolated over the step. At each point along the interpolation, the strain is converted to deformation gradient.

This gives a non-linear behavior when you look at the deformation gradient components in the output. When you would expect a nice straight line over the step, you get a curve.

See the test in /tests/swan/test_defgrad_basic.py.

While running HardeningJ2Plasticity.ipynb it failed when trying to run plastic3.py because of a call to symmetric_dyad() in core/tensor.py. I believe that the problem is that symmetric_dyad() expects two length-3 arrays but it's used in plastic3.py as a way to construct a 4th-order tensor from two 2nd order tensors.

As such, HardeningJ2Plasticity.ipynb cannot be run.

I am getting this error when I tried to follow the installation instruction and could not run it. Can you please take a look into it?

udayan@MEB-3560-02:~/Downloads/matmodlab-3.0.6/matmodlab$ mml test
Traceback (most recent call last):
File "/home/udayan/anaconda3/bin/mml", line 6, in
exec(compile(open(file).read(), file, 'exec'))
File "/home/udayan/Downloads/matmodlab-3.0.6/matmodlab/bin/mml", line 7, in
from matmodlab.mmd.startup import main
File "/home/udayan/Downloads/matmodlab-3.0.6/matmodlab/init.py", line 8, in
from StringIO import StringIO
ImportError: No module named 'StringIO'

Thanks

In Matmodlab, as suggested by @sswan, steps are run as they are added to the Simulator object. In Matmodlab2, steps are not run until the MaterialPointSimulator.run method is called. This was done for a couple reasons:

• If steps are run as they are added, the material must first be assigned to the simulator
• If all steps are run with the MaterialPointSimulator.run method, the output database can be setup with the correct dimensions.

If steps are to be run as the are added:

• The material must be assigned before the first step
• The size output database won't be known so it will either need to be allocated with some large size (that can be changed if space runs out), or allowed to grow.
• If you want simulation results to be written to a file, the call to MaterialPointSimulator.dump will have to be explicitly requested.

@sswan, do you still believe it is advantageous to run steps as they are added? I don't feel strongly either way.

Hello,

I'm still new to this tool, so my question is very simple.

I downloaded matmodlab, and (prerequisites: python, numpy, and scipy), but I don't know how to install matmodlab itself, I cannot understand how it works. could you help please?

Thanks a lot.

Many functions in core.linalg have repeated code