Hard (arraylike) parameter QtPlot setpoint linearity plotting bug

[nataliejpg]

Steps to reproduce

1. Do a Loop over a hard (arraylike) parameter where you generate the setpoints and plot it with QtPlot

loop = qc.Loop(y1.voltage.sweep(0,-10, 0.1)).each(v1.trace)
data = loop.get_data_set()
plot = qc.QtPlot()
plot.add(data.VNA_magnitude)
_ = loop.with_bg_task(plot.update, plot.save).run()

data set looks like:

DataSet:
   mode     = DataMode.LOCAL
   location = 'A:/TransmonExperiments/GoogleDrive/T2acquisition/Data/6QAK3_A2_qcodes/data/014'
   <Type>   | <array_id>        | <array.name> | <array.shape>
   Setpoint | yoko1_voltage_set | voltage      | (101,)
   Measured | frequency         | frequency    | (101, 401)
   Measured | VNA_magnitude     | magnitude    | (101, 401)
   Measured | VNA_phase         | phase        | (101, 401)
started at 2016-11-18 17:10:03

2. load the data and plot it again

loaded_data = qc.load_data(loc_provider.formatter.format(loc_provider.fmt,counter='004'))
plot = qc.QtPlot()
plot.add(loaded_data.VNA_magnitude)

Expected behaviour

Should look the same both times and have the setpoints you would expect on the x axis

Actual behaviour

Looks right when you live plot it (as in step 1) but gives a nonlinear spacing error when you load the data (as in step 2) in and discards the setoints but keeps the label which is now a lie!

from 1

from 2

with this error:

a:\qcodes\qcodes\plots\pyqtgraph.py:325: UserWarning: nonlinear setpoint array passed to pyqtgraph. ignoring, using default scaling.
  warnings.warn('nonlinear setpoint array passed to pyqtgraph. '

Note: works in matplot in terms of using the setpoint array even when loading data back (but we cant use that because of #380)

Note2: code generating setpoints looks like:

f = tuple(np.linspace(int(start), int(stop), num=npts))
self.setpoints = ((f,), (f,))
self.shapes = ((npts,), (npts,))

@giulioungaretti @alexcjohnson @WilliamHPNielsen

[nataliejpg]

Note3: also works correctly for plotting in the notebook after taking the data (ie without live update)

data = qc.Loop(v1.power.sweep(-60,-10, 1)).each(v1.trace).run()
qc.QtPlot(data.VNA_magnitude)

[nataliejpg]

Note4: this sometimes works as expected even in QtPlot (depends on the setpoints) which makes me think its something to do with rounding in the driver producing a nonlinear setpoint array but it still wouldn't really explain (to me at least) why it would always work fine with the Note3 example but not with loaded in data.

[MerlinSmiles]

I guess you could try to increase the MAXPX value in the qcodes/plots/pyqtgraph.py file to see if it is related to rounding errors?
Not that I understand how that error is actually calculated...

[nataliejpg]

@MerlinSmiles that works, thanks Merlin. Not sure whether to close the issue or leave open to flag up that this is a minor thing that should be looked at/changed in qcodes. @giulioungaretti ?

[MerlinSmiles]

@nataliejpg dont close it, its still a bug.

[jenshnielsen]

@nataliejpg I think this has been fixed some time ago ?

[nataliejpg]

@jenshnielsen agreed