All notable changes to Pulselib will be documented in this file.
- Better rendering of ramp in hres mode for ramp of 1 sample and duration < 1 ns.
- Fixed rendering of ramp in hres mode for ramp of 1 sample and duration < 1 ns.
- Fixed pulse shaping with scipy windows
- Improved waveform partitioning in Qblox compiler to reduce memory usage.
- Added range check on Qblox chirp.
- Refactored/cleaned Qblox pulse generation
- Update for qcodes 0.49
- Optimized off resonant pulses for Qblox to use less memory.
- Added optional, loopable kwargs to add_MW_pulse for AM and PM.
- Require Python >= 3.10 and numpy >= 1.24, < 2.0
- Remove HVI variables from segments.
- Fixed bug in amplitude of pulses on real channels introduced in v1.7.29.
- Added update_snapshot() to MeasurementParameter to allow custom snapshot data.
- Added arguments of fast_scan and read_input to snapshot.
- Fixed corner case bug in Qblox backend.
- Fixed bug in time trace with parameter sweeps.
- Small performance improvement in virtual matrix processing.
- Big performance improvement in virtual matrix processing during compilation.
- Fixed Keysight_QS for acquisitions > 2.1 s.
- Fixed read_channels for Keysight_QS with t_measure > 2.1 s or > 4000 samples. Limit is now 42 s.
- Added check on Qblox IQ marker timing.
- Automatically adjust RF frequency for Keysight_QS resonator drive (3.8 kHz resolution).
- Added checks on limits in read_channels for Qblox and Keysight.
- Fix Keysight_QS with RF source and video mode
- Removed accidentally committed print statements
- Improved HW/SW threshold checking with configuration: pulse_lib.acquisition.measurement_converter.MeasurementConverter.ALLOWED_RELATIVE_THRESHOLD_DEVIATION pulse_lib.acquisition.measurement_converter.MeasurementConverter.ALLOWED_FRACTION_THRESHOLD_DIFFERENCES
- Fixed error on feedback event insertion for Qblox.
- Fixed timing of rf_source in continuous mode for Qblox.
- Fixed exception in sub-nanosecond ramps for Qblox.
- Fixed unit radians for phase in measurement parameter.
- Allow multiple integration times on channel (if not using feedback on Qblox).
- Fixed RF generation in video mode with Keysight_QS
- Fixed adding (simultaneous) segments with multiple looping variables.
- Fixed releasing jobs from uploader.
- Small correction to DC compensation in Qblox fast scan 2D.
- Added high resolution sine wave to Qblox.
- Added RF generator to Keysight QS.
- Fix define_digitizer_channel_iq: add RF parameters.
- Fix error check on RF frequency sweep
- Added scan_resonator_frequency for fast digitizer/rf_source frequency sweep on Qblox.
- Log warning when sequence run duration > 3 s.
- Use AWG channel attenuation for Keysight QS sequences.
- Added oscillator triggering for QS mode.
- Clear sequencer memory in FPGA when Keysight_QS is configured.
- Performance improvement in MW pulse processing. (30% gain in precompile of randomized benchmarking.)
- Modified Keysight_QS backend to drive qubits without MW source using plunger and barrier gates.
- Added RF parameters frequency, source_amplitude and phase
- Added set_default_hw_schedule for Keysight and Tektronix
- Validate channel names
- Fixed unit of phase setting for Keysight RF demodulation [rad]
- Fixed acquisition to allow simultaneous acquisition by combining segments (QConstruct simultaneous)
- Added update_end(stop) to segment to extend segment till at least
stop
.
- Fixed Qblox feedback for multiple measurements (requires Q1Pulse v0.11.5)
- Added high resolution rendering of ramps on voltage channels (Qblox)
- Added hres rendering of sine waves on voltage channels (Keysight and Tektronix)
- Added checks on IQ channel configuration.
- Added Keysight AWG RF oscillator control
- Allow n_rep = None also for Keysight
- Fixed issue with bad aligned custom_pulse followed by ramp of > 1000 points.
- Added pre-pulses to Qblox fast scan.
- Corrected fix for mixed measurement assignments and raw measurements in measurement processing.
- Fix mixed measurement assignments and raw measurements in measurement processing.
- Fixed looping of keyword arguments. (It was broken by the performance update of v1.7)
- Removed digitizer triggering using add_HVI_marker (Keysight)
- Fixed Qblox hardware/software threshold data check.
- Raise clearer exception when a state threshold is applied on time trace data.
- Set defaults
selectors=False
andaccept_mask=False
forget_measurement_results
andget_measurement_param
- Added sequencer.recompile to quickly update an existing sequence when the virtual gate matrix has changed
- Corrected qblox version check for marker inversion.
- Use Qblox input channel selection if qblox-instruments v0.11 installed
- Compare results of hardware and software thresholding
- Fixed M4i acquisition data shape
- Qblox feedback corrected for input gain
- Qblox feedback rotation convert radians to degrees
- Added conditional segments for Qblox (requires qblox-instruments v0.9+)
- Added M4i digitizer control to pulse-lib
- Added IQ demodulation to M4iControl
- Performance improvement of 10% up to 5 times in compilation.
- Added looping on segment.sample_rate
- Added aggregate_func to aggregate n samples of 1 measurement to a single value with a user defined function.
- Improved accuracy of hres ramps for Keysight (error is now ~10 ps).
- Do not add DC compensation pulse for Qblox if it is too small. This avoids uploads of sequences that do nothing.
- Improved operations on multi-dimensional looping functions
- Fixed IQ output of qblox_fast_scan
- Removed add_HVI_marker and add_HVI_variable from segment channels. Huge performance improvement for GST and RB. Overhead O(N^2) -> O(N)
- Added reload_seq=True to Qblox fast scan 1D and 2D for external sweeps modifying pulse-lib settings
- Fixed data returned by acquisition parameter when no shot is accepted.
- Fixed attenuation qubit channel for Qblox
- Fixed baseband qubit channel for Qblox
- Removed digitizer resonator drive amplitude sweep feature for Qblox. It broke functionality.
- Fixed setting of digitizer demodulation frequency and resonator drive amplitude for Qblox.
- Fixed Qblox uploader
- Fixed retries on Keysight exceptions
- Fixed rendering stacked and/or unaligned sine and custom pulses for Qblox
- Render wave of add_sin() with phase starting at start of wave, i.e. independent of start time of pulse.
- Fixed Qblox time offset bug in previous release.
- Fixed merging of phase shift and MW pulse in conditional segment (Keysight_QS)
- Fixed Qblox uploader with unaligned long pulses and gap between pulses.
- Fixed Keysight release_all_awg_memory
- Added measure till end of sequence (time trace feature)
- Fixed Qblox uploader with unaligned pulses.
- Fixed Keysight FPGA marker upload time rounding (Keysight and Keysight_QS).
- Fixed QuantumSequencer (Keysight_QS) long pulses with post-phase shift.
- Fixed Qblox uploader for unaligned MW pulses > 200 ns
- Fixed Keysight digitizer configuration for time-traces
- Added total_time to sequence and metadata
- Fixed Qblox uploader
- Fixed digitizer configuration for time-traces with Keysight uploaders
- Added histograms to measurement parameter
- Render pulses and ramps on Qblox with 1 ns resolution.
- Removed segment_pulse.add_ramp(), because it gave too many errors.
- Renamed idle frequency and reference frequency to qubit resonance frequency.
- Allow overrule of qubit resonance frequency in sequence.
- Added FrequencyUndefined for not-coherent pulsing with NCO frequency 0.0 between pulses.
- Deprecate segment_pulse.add_ramp in favour of add_ramp_ss
- Added configuration of digitizer to Keysight_QS uploader
- Added at(index) to loop_obj to get value at sequence index.
- Corrected phase for MW pulses on Qblox.
- Added py.typed for mypy
- Merge marker pulses if < 10 ns between pulses.
- Fixed rounding errors on segment boundaries in Qblox uploader.
- Qblox uploader assign IQ marker to sequencer of first qubit on IQ output.
- Refactored IQ channel definition: pulse_lib.define_iq_channel()
- Added lp.arange() similar to numpy.arange()
- Fixed missing IQ marker around chirp
- Fixed MeasurementMajority to work with pulse_templates.
- Fixed DC compensation on Qblox.
- Fixed some edge cases.
- QuantumSequencer: split long constant waveforms in start and stop waveform.
- Fixed measurement parameter for acquisition with accept_if not None
- Round acquisition time towards zero in uploader
- Fix for measurement expressions in sequence
- Fix for simultaneous driving
- Fixed majority vote in expression
- Fixed acquire wait=True with n_repeat
- Added checks on t_measure and sample rate in read_channels.
- Fixed actual number of points and interval for acquisition: n_samples, interval = uploader.actual_acquisition_points(acquisition_channel, t_measure, sample_rate)
- Replaced logging by logger
- Fixed setpoints of measurement parameter for time traces > 2 seconds.
- Fixed simultaneous pulses with indexing of segments on looping parameter.
- Calculate timeout for get_acquisition_data from sequence.
- Added continuous mode for Keysight AWG.
- Fixed errors when slicing segments after adding dimension with update_dim.
- Remove phase shifts of 0.0 rad before rendering.
- Fixed error in KeysightQS with phase-shift after MW pulse.
- Added tests.utils.last_upload.get_last_upload to analyze upload after exception
- Fixed numpy deprecation of np.int and np.float.
- Qblox uploader overlapping pulses error
- parameter iq_mode to sequencer.get_measurement_parameter(), get_measurement_data(), Qblox fast scan
- read_channels() to read digitizer channels without pulsing AWG. (Currently Qblox only)
- Keysight AWG mock rendering at 10 GSa/s using digital filter setting.
- acquire(... wait=True) to add wait time for acquisition.
- Timing of Keysight_QS with 2022 firmware
- Keysight_QS rf_source bug
- Fixed n_samples in measurement_param
- Fixed Qblox sequence generation when no pulses on AWG channels
- get_acquisition_param. It was deprecated for quite some time.
- pulselib.get_channel_attenuations()
- Sequencer upload and play only retry upload after specific Keysight exception.
- Chirp is stored as an instruction to be rendered by backend.
- start_times to measurement_description
- measurement names do not have to be unique. A postfix makes them unique.
- Digitizer trigger timing has been corrected for channel delays !!
- Digitizer RF source API has changed.
- Looping has been refactored to add functionality and improve performance.
- Important internal interface changes are marked with [v1.6.0] in the code.
- Added segment_container.update_dim for looping with index
- Added numpy array operations to looping variables
- Added looping.array to loop over an arbitrary array of values
- Added digitizer_channel.delay
- Added
prolongation_ns
to RF source configuration - Added
iq_complex
to fast_scan1D_param and fast_scan2D_param
- Changed set_digitizer_rf_source
trigger_offset_ns
tostartup_time_ns
. - Refactored looping and segment indexing:
- Improved performance of looping arguments
- Improved performance of segment_container.reset_time()
- Improved performance of segment and segment_container indexing
- Improved performance of sequence pre-rendering
- Reduced memory usage
- Added sanity checks on looping arguments
- Added checks on frequency when rendering MW pulses
- Removed segment_container append(), slice_time(), last_edit
- Removed segment_IQ.add_global_phase()
- Fixed addition of 2 segment_containers to construct simultaneous driving
- Tektronix marker on unused AWG channel
- Measurement result inversion with zero_on_high
- Multiple measurements on 1 channel with down-sampling
- Fixed error in measurement parameter when no shot is accepted
- Incorrect setpoints in measurement_param.add_derived_param.
- Added endpoint to looping.linspace, logspace and geomspace.
- Added more data selection options to get_measurement_results()
- Added addition of segment_containers to construct simultaneous driving
- Added sequence_builder.add_simultaneous()
- Do not add axis 'repetition' if sequencer.n_rep == None.
- Fixed set_acquisition with t_measure and sample_rate
- Fixed get_measurement_param with iq_complex=False
- Fixed high memory usage due to unlimited waveform cache.
- Fixed looping and HVI marker
- Added retries to sequence upload and play
- sequencer.get_channel_data and get_measurement_param when using play() with index as done by core-tools.
- Added sequencer.plot()
- Fixed looping t_off of add_HVI_marker
- Fixed plotting after mk_sequence
- Support for Qblox modules: QCM and QRM
- Added sequencer.get_channel_data
- Added sequencer.set_acquisition
- Added sequencer.get_measurement_param with automatic upload/play and thresholding
- Added sequencer.get_measurement_results
- Added channel acquire(..., n_repeat= , interval= )
- Added qblox_fast_scan 1D and 2D
- Added digitizer configuring by M3202A_Uploader.
- Default channel offset is None to allow configuration outside of pulselib
- Scaling of custom pulses with virtual gates. Custom function now always called with amplitude of original pulse.
- Renamed Tektronix backend 'Tektronix_5014' after refactoring:
- Faster and allow fast switching between multiple uploaded sequences.
- Amplitude output has been corrected. It is 2x previous output. Correct attenuation per channel!!
- Use sequence.play(release=False) to call play multiple times after a single upload.
- Use infinite looping when n_rep > 65535
- Added sequencer.repetition_aligment to align the duration of the sequence with an external clock or signal frequency. (currently only implemented for Tektronix)
- Added argument reset_time to wait()
- Several pages of documentation (not finished yet..)
- Error when rendering section with low sample rate extends into segment with high sample rate.
- QuantumSequencer use waveform with low sample rate for long DC compensation pulse.
- QuantumSequencer bugs in 1.3.2
- Fixed long wait (> 160 ms) for QuantumSequencer
- Added attribute 'values' to sequence loop parameters
- Fixed rendering of segments with different sample rates
- Rendering of conditional segments with looping
- IQ correction of phase, amplitude and offset: add_channel_offset, set_qubit_correction_phase, set_qubit_correction_gain
- New interface from hardware class to pulselib: set_channel_attenuations and add_virtual_matrix
- Virtual matrix on top of virtual gates
- Added hw schedule for UHFLI with Tektronix
- Improved release_awg_memory (for Keysight AWG)
- pulse.add_channel_compenstation_limit. Correct method is add_channel_compensation_limit
- loops with t_measurement in HVI variable.
- release_awg_memory() for Keysight
First labeled release. Start of dev branch and change logging.