python heat_tmax.py --optimizer newton fails

[slitvinov]

I do

$ python3.11 -m ensurepip
$ python3.11 -m pip install numpy matplotlib tensorflow
$ python3.11 -m pip install git+https://github.com/cselab/odil.git
$ cd examples/heat_tmax
$ python3.11 heat_tmax.py --optimizer newton

and it fails with the following error message

WARNING: All log messages before absl::InitializeLog() is called are written to STDERR
E0000 00:00:1731580699.079602 3103184 cuda_dnn.cc:8310] Unable to register cuDNN factory: Attempting to register factory for plugin cuDNN when one has already been registered
E0000 00:00:1731580699.083125 3103184 cuda_blas.cc:1418] Unable to register cuBLAS factory: Attempting to register factory for plugin cuBLAS when one has already been registered
heat_tmax.py --optimizer newton
Running Newton optimizer

epoch=00000
residual: eqn:5.6127, imp:0.011106
tmax=1
memory: 544 MiB
walltime: 2.017 s
walltime+callback: 2.019 s
u_00000.png
Traceback (most recent call last):
  File "/n/home13/slitvinov/odil/examples/heat_tmax/heat_tmax.py", line 189, in <module>
    main()
  File "/n/home13/slitvinov/odil/examples/heat_tmax/heat_tmax.py", line 185, in main
    odil.optimize(args, args.optimizer, problem, state, callback)
  File "/n/home13/slitvinov/.local/lib/python3.11/site-packages/odil/util.py", line 316, in optimize
    return optimize_newton(args, problem, state, callback, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/n/home13/slitvinov/.local/lib/python3.11/site-packages/odil/util.py", line 244, in optimize_newton
    vector, matrix = problem.linearize(state)
                     ^^^^^^^^^^^^^^^^^^^^^^^^
  File "/n/home13/slitvinov/.local/lib/python3.11/site-packages/odil/core.py", line 1321, in linearize
    m = modsp.csr_array(garray)
        ^^^^^^^^^^^^^^^^^^^^^^^
  File "/n/home13/slitvinov/.local/lib/python3.11/site-packages/scipy/sparse/_compressed.py", line 92, in __init__
    coo = self._coo_container(arg1, dtype=dtype)
          ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/n/home13/slitvinov/.local/lib/python3.11/site-packages/scipy/sparse/_coo.py", line 84, in __init__
n    self._shape = check_shape(M.shape, allow_1d=is_array)
                  ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/n/home13/slitvinov/.local/lib/python3.11/site-packages/scipy/sparse/_sputils.py", line 332, in check_shape
    raise ValueError('shape must be a 1- or 2-tuple of positive '
ValueError: shape must be a 1- or 2-tuple of positive integers

It works as expected with default optimizer

$ python3.11 heat_tmax.py
heat_tmax.py
Running L-BFGS-B optimizer

epoch=00000
residual: eqn:5.6127, imp:0.011106
tmax=1
memory: 547 MiB
walltime: 2.092 s
walltime+callback: 2.093 s
u_00000.png

epoch=01000
residual: eqn:0.0074659, imp:0.002916
tmax=3.6493
memory: 588 MiB
walltime: 5.631 s
walltime+callback: 5.860 s
walltime/epoch: 5.631 ms
throughput: 0.727M cells/s
u_00001.png
...

I have

$ python3.11 --version
Python 3.11.9
$ python3.11 -c 'import tensorflow as tf; print(tf.version.VERSION)'
2.18.0
$ python3.11 -c 'import numpy; print(numpy.version.version)'
2.0.2
$ cat /etc/*rel*
Rocky Linux release 8.9 (Green Obsidian)
NAME="Rocky Linux"
VERSION="8.9 (Green Obsidian)"
ID="rocky"
ID_LIKE="rhel centos fedora"
VERSION_ID="8.9"
PLATFORM_ID="platform:el8"
PRETTY_NAME="Rocky Linux 8.9 (Green Obsidian)"
ANSI_COLOR="0;32"
LOGO="fedora-logo-icon"
CPE_NAME="cpe:/o:rocky:rocky:8:GA"
HOME_URL="https://rockylinux.org/"
BUG_REPORT_URL="https://bugs.rockylinux.org/"
SUPPORT_END="2029-05-31"
ROCKY_SUPPORT_PRODUCT="Rocky-Linux-8"
ROCKY_SUPPORT_PRODUCT_VERSION="8.9"
REDHAT_SUPPORT_PRODUCT="Rocky Linux"
REDHAT_SUPPORT_PRODUCT_VERSION="8.9"
Rocky Linux release 8.9 (Green Obsidian)
Rocky Linux release 8.9 (Green Obsidian)
Derived from Red Hat Enterprise Linux 8.9
Rocky Linux release 8.9 (Green Obsidian)
cpe:/o:rocky:rocky:8:GA

[kangbeicai]

Thanks for your reply. I found that the inverse problem cannot be solved by Newton, which has been mentioned in the paper. When I remove the imp term, "shape must be a 1- or 2-tuple of positive integers" no longer appears. However, res after optimization becomes nan.