difference between save_mat and save_lattice

[simoneliuzzo]

Is there any difference between the two functions save_mat and save_lattice?

Also, the trivial user (as I am) would like to use scipy.io.savemat to save matlab format files, but the Lattice object does not seem to be a savable variable. Is there any way to warn the user, that savemat is not to be used, and instead save_mat or save_lattice are provided?

[simoneliuzzo]

Here an example of the outputs I get for the lattice "rerr"

print(rerr)
Lattice(<4434 elements>, name='sr', energy=6000000000.0, particle=Particle('relativistic'), periodicity=1, harmonic_number=992, beam_current=0.0, nbunch=1)

at.save_lattice

>>> at.save_lattice(rerr, lat_err_name, mat_key='rerr')
Traceback (most recent call last):
  File "/mnt/multipath-shares/mcssware/devel_tools/packages/debian9/pycharm/pycharm-community-2019.2.4/helpers/pydev/_pydevd_bundle/pydevd_exec2.py", line 3, in Exec
    exec(exp, global_vars, local_vars)
  File "<input>", line 1, in <module>
  File "/machfs/liuzzo/fcc_py38/lib/python3.8/site-packages/at/load/allfiles.py", line 80, in save_lattice
    return save_func(ring, filepath, **kwargs)
  File "/machfs/liuzzo/fcc_py38/lib/python3.8/site-packages/at/load/matfile.py", line 275, in save_mat
    lring = tuple((element_to_dict(elem),) for elem in matlab_ring(ring))
  File "/machfs/liuzzo/fcc_py38/lib/python3.8/site-packages/at/load/matfile.py", line 275, in <genexpr>
    lring = tuple((element_to_dict(elem),) for elem in matlab_ring(ring))
  File "/machfs/liuzzo/fcc_py38/lib/python3.8/site-packages/at/load/utils.py", line 347, in element_to_dict
    dct = dict((k, _mattype_map.get(type(v), lambda attr: attr)(v))
  File "/machfs/liuzzo/fcc_py38/lib/python3.8/site-packages/at/load/utils.py", line 347, in <genexpr>
    dct = dict((k, _mattype_map.get(type(v), lambda attr: attr)(v))
  File "/machfs/liuzzo/fcc_py38/lib/python3.8/site-packages/at/lattice/elements.py", line 146, in items
    yield k, getattr(self, k)
  File "/machfs/liuzzo/fcc_py38/lib/python3.8/site-packages/at/lattice/elements.py", line 426, in K
    return self.PolynomB[1]
IndexError: index 1 is out of bounds for axis 0 with size 1


>>> at.save_lattice(rerr, lat_err_name)
Traceback (most recent call last):
  File "/mnt/multipath-shares/mcssware/devel_tools/packages/debian9/pycharm/pycharm-community-2019.2.4/helpers/pydev/_pydevd_bundle/pydevd_exec2.py", line 3, in Exec
    exec(exp, global_vars, local_vars)
  File "<input>", line 1, in <module>
  File "/machfs/liuzzo/fcc_py38/lib/python3.8/site-packages/at/load/allfiles.py", line 80, in save_lattice
    return save_func(ring, filepath, **kwargs)
  File "/machfs/liuzzo/fcc_py38/lib/python3.8/site-packages/at/load/matfile.py", line 275, in save_mat
    lring = tuple((element_to_dict(elem),) for elem in matlab_ring(ring))
  File "/machfs/liuzzo/fcc_py38/lib/python3.8/site-packages/at/load/matfile.py", line 275, in <genexpr>
    lring = tuple((element_to_dict(elem),) for elem in matlab_ring(ring))
  File "/machfs/liuzzo/fcc_py38/lib/python3.8/site-packages/at/load/utils.py", line 347, in element_to_dict
    dct = dict((k, _mattype_map.get(type(v), lambda attr: attr)(v))
  File "/machfs/liuzzo/fcc_py38/lib/python3.8/site-packages/at/load/utils.py", line 347, in <genexpr>
    dct = dict((k, _mattype_map.get(type(v), lambda attr: attr)(v))
  File "/machfs/liuzzo/fcc_py38/lib/python3.8/site-packages/at/lattice/elements.py", line 146, in items
    yield k, getattr(self, k)
  File "/machfs/liuzzo/fcc_py38/lib/python3.8/site-packages/at/lattice/elements.py", line 426, in K
    return self.PolynomB[1]
IndexError: index 1 is out of bounds for axis 0 with size 1

at.save_mat(rerr, lat_err_name)

>>> at.save_mat(rerr, lat_err_name)
Traceback (most recent call last):
  File "/mnt/multipath-shares/mcssware/devel_tools/packages/debian9/pycharm/pycharm-community-2019.2.4/helpers/pydev/_pydevd_bundle/pydevd_exec2.py", line 3, in Exec
    exec(exp, global_vars, local_vars)
  File "<input>", line 1, in <module>
  File "/machfs/liuzzo/fcc_py38/lib/python3.8/site-packages/at/load/matfile.py", line 275, in save_mat
    lring = tuple((element_to_dict(elem),) for elem in matlab_ring(ring))
  File "/machfs/liuzzo/fcc_py38/lib/python3.8/site-packages/at/load/matfile.py", line 275, in <genexpr>
    lring = tuple((element_to_dict(elem),) for elem in matlab_ring(ring))
  File "/machfs/liuzzo/fcc_py38/lib/python3.8/site-packages/at/load/utils.py", line 347, in element_to_dict
    dct = dict((k, _mattype_map.get(type(v), lambda attr: attr)(v))
  File "/machfs/liuzzo/fcc_py38/lib/python3.8/site-packages/at/load/utils.py", line 347, in <genexpr>
    dct = dict((k, _mattype_map.get(type(v), lambda attr: attr)(v))
  File "/machfs/liuzzo/fcc_py38/lib/python3.8/site-packages/at/lattice/elements.py", line 146, in items
    yield k, getattr(self, k)
  File "/machfs/liuzzo/fcc_py38/lib/python3.8/site-packages/at/lattice/elements.py", line 426, in K
    return self.PolynomB[1]
IndexError: index 1 is out of bounds for axis 0 with size 1

then finally a simple scipy.io.savemat, that does not give errors, but does not save the lattice variable either (it contains the first element only, it seems)

savemat(lat_err_name, mdict={'lat':rerr})

[lfarv]

Hello `save_lattice`, equivalent to the `Lattice.saveˋ method is the general saving function, where the format is defined by the file name extension. See help(save_lattice) for info on the available file formats. ˋsave_mat` Is the underlying function specialized for .mat files. Scipy.io.savemat is a low-level scipy function. It can only save some python types, but not class instances. See help(savemat) for info on what you can do with it. Finally, apparently your rerr lattice is broken. It seems that one element has a wrong PolynomB (not the right shape, or incompatible with MaxOrder).

[simoneliuzzo]

The rerr lattice is plotting beta without issues, computes optics etc. please see below a screenshot:

the lattice itself is not "printing", the plot_beta and linopt6 commands instead are apparently working.

I may not attach here the file, as I am unable to save it.

After a bit of search, I have an element defined as follows:

There is a K field (K=0), but this field seems "unremovable":

So, the questions now are: why is it forbidden to have an element with len(PolynomB)=1, MaxOrder=0 and K=0 ? How to successfully remove the field K?

In order to find these functions for lattice saving, I simply searched the online AT help and google. As in several other cases, it is somehow difficult to chose the correct function to use. May be the functions that are not supposed to serve directly the use, such as save_mat, may be renamed to _save_mat? May be there is an official python way to do this.

[simoneliuzzo]

I answer in advance a foreseen return question: Why is K there?

it is NOT there in the original matlab AT file.

[simoneliuzzo]

Here below two picture of the same commands. The K attribute appears when loading the lattice in pyAT

MATLAB (index 2021):

PYTHON (index 2019, due to ringparam element missing and python starting to count at 0):

[lfarv]

@simoneliuzzo: you are reporting 2 different points, so I start with the 'K' problem:

In PyAT, K is not an attribute, it's a property of Quadrupoles and Dipoles. A property is computed upon request, not stored. That's why you cannot remove it.

The unexpected presence of this K property in Multipoles (other than Quadrupoles and Dipoles) was reported in #454 and corrected in #455. Can you try with a recent PyAT version ? It should solve your problem. If not, let me know.

[lfarv]

@simoneliuzzo: 2nd question (functions for saving lattices)

PyAT exposes both the generic saving function save_lattice, and all the specific functions corresponding the the various available formats: save_mat, save_m, save_repr. I agree that the access to the format-specific functions is not crucial, but it's there from the beginning.
The best way to answer your question is to add a missing "See also:" item in the functions' help, so that the user may compare the different possibilities. I'll do that.

[simoneliuzzo]

Dear Laurent,

I was indeed on an "old" branch.

In master there are no errors when using the lattice and load_lattice and save_lattice seem to work.
When printing the lattice to see its content something happens that I may not explain:

as seen in the above screenshots print() is ok.
simply pressing enter instead prints only a part of the lattice and it does not seem to start from zero.

[lfarv]

There are 2 python functions to get a text representation of an object:

• str gives a "human-friendly" representation. For a lattice, it just shows the number of elements to avoid a too long display,
• repr gives a string which should enable to rebuild the object from the string, when possible. For a lattice, it means building a list of the repr string of all its elements, and appending its attributes (be careful: it's a very long output).

By default, print displays the str representation, while simply entering the name of a variable displays its repr string.