qcodes / broadbean Goto Github PK

Hi,

I had some trouble when running the following code:

bp1.setSR(1e9)
bp1.insertSegment(0, ramp, (0, 1e-3), name='', ts=1, durs=20e-9)
bp1.insertSegment(1, ramp, ( 1e-3, 1e-3), name='', ts=1, durs=30e-9)
bp1.plot()

One gets the error:
File "C:\ProgramData\Anaconda3\lib\site-packages\matplotlib\axes_base.py", line 244, in _xy_from_xy
"have shapes {} and {}".format(x.shape, y.shape))
ValueError: x and y must have same first dimension, but have shapes (50,) and (49,)

When uploading a sequence BB doesn't check if the real channel amplitude and the one defined from this function ptseq.setChannelAmplitude() are the same.

Could it be possible to implement either of these solutions:

• raise a flag if the two are different?
• using the value of ptseq.setChannelAmplitude(i,f) to update AWG. chi_amplitude(f) when uplaoding the sequence?

The main broadbean.py file has grown to become a monster. We need to factor it out.

@Dominik-Vogel

Blueprints should be renamed to waveforms everywhere in the code base.

What is the reason for the npts argument of custom functions? E.g.

    def sine(freq, ampl, off, SR, npts):
        time = np.linspace(0, npts/SR, npts)
        freq *= 2*np.pi
        return (ampl*np.sin(freq*time)+off)

A format more user friendly seems to be:

    def sine(freq, ampl, off, times):
        freq *= 2*np.pi
        return (ampl*np.sin(freq*times)+off)

@WilliamHPNielsen @lucblom

While working with AWG5208 and using the example notebook from broadbean as the reference, using outputForSEQXFile method of the Sequence object together with makeSEQXFile method of the AWG driver like below resulted in an error 1331 on the AWG device upon loading the successfully created and transmitted SEQX file:

...
seqx_args = sequence.outputForSEQXFile()
seqx_file = awg.makeSEQXFile(**seqx_args)
...

As @WilliamHPNielsen suggested, forging the sequence should be used instead (and it indeed worked):

...
forged_sequnce = sequence.forge()
seqx_file = awg.make_SEQX_from_forged_sequence(forged_sequence)
...

ToDo:

• deprecate outputForSEQXFile method of the Sequence object with the message of using forge instead
• update the example notebook to use forge

@WilliamHPNielsen @sohailc

How should we deal with very short segments? What is closest to what the user needs and how do AWG's usually handle short pulses?

If we supply an array of voltages [0, 0, 1, 1, 0, 0, 1, 1] at a given sample rate, when do we expect voltage changes to happen?

Given the following input:

Which of the three following scenarios do we expect the AWG to (attempt to) realise? How should broadbean's plotting reflect that?

@dmtvanzanten @Dominik-Vogel @nataliejpg

Is it possible to add a Blueprint relative to another one? E.g.

bp1 = bb.BluePrint()  
bp1.insertSegment(0, ramp, (0, 1e-3), name='wave1', dur=3.001e-6)
bp1.insertSegmentOffset(relative='wave1', offset=1e-6, ramp, (0, 1e-3), name='wave2', dur=3.001e-6)

Currently PulseAtoms must have an argument.

The bundling together of N markers and one analogue channel is somewhat artificial (although some AWGs actually bundle them together in hardware). It would be a good simplification to simply have waveforms and channels, where a waveform then may be 1-bit, 12-bit, 14-bit or whatever. It should only matter when assigning a waveform to an output of an instrument, where the instrument (driver) should then complain if one tries to assign a non-binary waveform to a marker channel. Just like the driver should complain if the waveform specifies too large a voltage etc.

For the plotting, however, it certainly does make sense to distinguish a marker from a "real" waveform, so I would propose that the plotter takes a list of channels to visualise as markers. They should then get a smaller plot window or something like that.

As of now channels are not given in any particular order:

    @property
    def channels(self):
        """
        The channels that has something on them
        """
        chans = [key for key in self._data.keys()]
        return chans

Windows 10 in an environment installed with the qcodes-dk installer script which has installed PyQT5.6 from Conda

pip install git+https://github.com/QCoDeS/broadbean.git
Collecting git+https://github.com/QCoDeS/broadbean.git
  Cloning https://github.com/QCoDeS/broadbean.git to c:\users\qcodes~4\appdata\local\temp\pip-r0ge38lm-build
Requirement already satisfied: numpy>=1.12.1 in c:\users\qcodes-davewecker\anaconda3\envs\qcodes-qdev-master\lib\site-packages (from broadbean==0.9)
Requirement already satisfied: matplotlib>=2.0.1 in c:\users\qcodes-davewecker\anaconda3\envs\qcodes-qdev-master\lib\site-packages (from broadbean==0.9)
Collecting PyQt5 (from broadbean==0.9)
  Using cached PyQt5-5.8.2-5.8.0-cp35.cp36.cp37-none-win_amd64.whl
Requirement already satisfied: six>=1.10 in c:\users\qcodes-davewecker\anaconda3\envs\qcodes-qdev-master\lib\site-packages (from matplotlib>=2.0.1->broadbean==0.9)
Requirement already satisfied: python-dateutil in c:\users\qcodes-davewecker\anaconda3\envs\qcodes-qdev-master\lib\site-packages (from matplotlib>=2.0.1->broadbean==0.9)
Requirement already satisfied: pytz in c:\users\qcodes-davewecker\anaconda3\envs\qcodes-qdev-master\lib\site-packages (from matplotlib>=2.0.1->broadbean==0.9)
Requirement already satisfied: cycler>=0.10 in c:\users\qcodes-davewecker\anaconda3\envs\qcodes-qdev-master\lib\site-packages (from matplotlib>=2.0.1->broadbean==0.9)
Requirement already satisfied: pyparsing!=2.0.4,!=2.1.2,!=2.1.6,>=1.5.6 in c:\users\qcodes-davewecker\anaconda3\envs\qcodes-qdev-master\lib\site-packages (from matplotlib>=2.0.1->broadbean==0.9)
Collecting sip<4.20,>=4.19 (from PyQt5->broadbean==0.9)
  Using cached sip-4.19.2-cp35-none-win_amd64.whl
Installing collected packages: sip, PyQt5, broadbean
  Running setup.py install for broadbean ... done
Successfully installed PyQt5-5.8.2 broadbean-0.9 sip-4.19.2

Allow named arguments for PulseAtoms.
Current call signature

BluePrint.insertSegment(pos, func, args=(), dur=None, name=None, durs=None)

does not allow to use PulseAtoms with named arguments. I my opinion the named arguments should even be the default in order to avoid human error when calling a method like

bp.insertSegment(-1, monteCarloZeta, args=(12.3, 12.5, 59.76, 21.4 , 43.56, 3*pi))

Currently the sample rate has to be specified for every Blueprint. It is desirable to build entire sequences without having to specify the sample rate and set it once it needs to be forged.
We should think about the implications of setting the sample rate at that later stage.

BluePrints are assigned to channel depending on order in which they were added using addBluePrint, instead of channel numbers specified in addBluePrint.

Eg.

element = bb.Element()
element.addBluePrint(2, pulse2)
element.addBluePrint(1, pulse1)
seq = bb.Sequence()
seq.addElement(1, element)
...
seq.outputForAWGFile()
awg.make_send_and_load_awg_file(*awgfile[:])

will assign pulse2 to channel 1 and pulse1 to channel 2.