Report generation for LGST results object about pygsti HOT 6 CLOSED

pygsti.do_lgst(...) returns a GateSet object, not a Results object (see its docstring). The AttributeError shown above is due to you passing a GateSet to create_standard_report where is expects a Results object - so the error is not a bug. You raise a good point, however, which is "how do I just run LGST and generate a report?". There isn't a good way to do this currently, but I think this would be a great feature/functionality to include. I'll try (but not promise :) to create a nice way of doing this by the next release, which should only be about 2 weeks away.

The closest thing you can do currently (in version 0.9.4.x) is run do_stdpractice_gst with a single maximum-length of 0 and with a large tolerance, so the slow iterative optimization exits quickly. You should be able to do this using something like (where I've also added 'onBadFit': [], so it doesn't do anything extra if the fit is bad, and modes="TP" so it only does a single optimization) :

maxLengths= [0]
global_opts = {'tolerance':1000.0, 'onBadFit': []}
result = pygsti.do_stdpractice_gst(ds, gs_target, prep_fiducials,
                                   meas_fiducials, germs, maxLenghts,
                                   advancedOptions={'all': global_opts},
                                   modes="TP")

Let me know how long this takes. It will still do more than it needs to, so it won't be 2.5s, but it will hopefully take significantly less than an hour.

from pygsti.

@enielse

The closest thing you can do currently (in version 0.9.4.x) is run do_stdpractice_gst with ....

Thanks for pointing that out, I'll try that today. Looking forward to the inclusion of this new feature.

from pygsti.

@enielse I tried the "quick" GST. It seems to work as expected. However, when specifying maxLenghts=[0], an Attribute Error is raised in the report generation. When instead I set maxLenghts=[1] everything works fine. Running this simple version of GST takes ~1.5 mins including report generation and is a huge lifesaver to determine if data makes sense! thanks for the help.

from pygsti.

Great! I'm glad this basically worked. Changing the [0] to [1] as you did is fine. In addition to not causing an error, using [1] will cause the color box plots in the "Per-sequence detail" tab of the report to show something (i.e. not be blank).

from pygsti.

As of the just-released version 0.9.5, you can generate LGST-only reports even more easily, using the new do_linear_gst function like so:

results = pygsti.do_linear_gst(ds, gs_target, prep_fiducials, meas_fiducials)
pygsti.report.create_standard_report(results, filename="example_files/LGSTonlyReport",
                                     title="LGST-only Example Report", verbosity=2)

This is further detailed in the new example notebook:
Reports - How to create an LGST-only report.ipynb

from pygsti.

Closing since I think this issue is resolved and it's been a while without activity.

from pygsti.