qucat / qucat

Quantum Circuit Analyzer Tool

QuCAT: Quantum Circuit Analyzer Tool in Python

Maintained by Mario Gely

QuCAT stands for Quantum Circuit Analyzer Tool. This open source python library provides standard analysis tools for superconducting electronic circuits, built around Josephson junctions.

QuCAT features an intuitive graphical or programmatical interface to create circuits, the ability to compute their Hamiltonian, and a set of complimentary functionalities such as calculating dissipation rates or visualizing current flows in the circuit. QuCAT currently supports quantization in the basis of normal modes.

If you find QuCAT useful, please consider citing it in your publications, see: https://qucat.org/citing.html.

If you have questions concerning the usage or installation of QuCAT, please ask questions on our forum.

Installation

The recommended way to install qucat is via pip by opening a terminal and running pip install qucat, more information here.

Documentation

Visit https://qucat.org/ for all information about installation, a documented lists of functions, tutorials and more.

Tutorials

A selection of tutorials is available here.

Build status and test coverage

Master branch:

Contribute

You are most welcome to contribute to QuCAT development by forking this repository and sending pull requests.

Filing bug reports, proposing changes or additional features can be done on our issues page

You can also contact Mario at [email protected]

qucat's Issues

Right click only occurs when clicking wheel on mac

Remove dependence on QuTiP

There should be an option to return Hamiltonians as an array rather than a QuTiP object if QuTiP is found to not be installed.

Opening file from GUI on mac leads to warning

Warning is of the form:

Class FIFinderSyncExtensionHost is implemented in both /System/Library/PrivateFrameworks/FinderKit.framework/Versions/A/FinderKit (0x7fffaea88210) and /System/Library/PrivateFrameworks/FileProvider.framework/OverrideBundles/FinderSyncCollaborationFileProviderOverride.bundle/Contents/MacOS/FinderSyncCollaborationFileProviderOverride (0x11e584dc8). One of the two will be used. Which one is undefined.

Circuit components disappear when moving out of the editor window, and cannot be created from outside the editor window

Occurs on old mac versions: MacOS HighSierra 10.13.6

GUI issues on mac

Operating system: MacOS HighSierra 10.13.6

Possibly related issues:

• https://github.com/mgely/qucat/issues/2
• https://github.com/mgely/qucat/issues/6

I have the following issues on my mac:

• the writing is too small, see screenshot
• the GUI closes unexpectedly when trying to scroll using two fingers
• except for the wires and capacitors, none of the circuit element icons show up. They appear to have been placed with the corresponding values and are drawn in the jupyter seem, but don't appear in the GUI

Text file containing the circuit:
felix-test-shorting.txt

Dependencies not installed with 'pip install'

Steps to reproduce:

conda create -n qucat
conda activate qucat
pip install qucat
python -c 'import qucat'

This produces the following traceback:

Traceback (most recent call last):
  File "<string>", line 1, in <module>
  File "/home/jbw/.local/miniconda/lib/python3.7/site-packages/qucat/__init__.py", line 2, in <module>
    from .core import *
  File "/home/jbw/.local/miniconda/lib/python3.7/site-packages/qucat/core.py", line 1, in <module>
    import sympy as sp
ModuleNotFoundError: No module named 'sympy'

This can be easily fixed by specifying install_requires in setup.py

Add list of contributors to website

GUI closes unexpectedly when trying to scroll using two fingers

Occurs on old version of mac: MacOS HighSierra 10.13.6

Add list of relevant attributes of Qcircuits/Components in the docs

Add code formatting with Black

Black is a good Python formatter. It would automatically format the existing code and can be used to check for format errors on new pull requests.

In edit value, enter is not bound to ok on linux

Toolbar not giving consistent behavior

Feature request: Plot, but don't break circuit

When omitting junctions or inductances, qucat returns a ValueError: There should be at least one junction or inductor in the circuit. It would be nice if the corresponding circuit could still be plotted, in case you wanted to just plot circuits

Add developer guide to the documentation

To be included:

• What commenting or docstring style to adopt
• How to edit and run tests
• How to edit and build the documentation

Enable plotting functionalities for programmatically created Qcircuits

echos issue #43

Certain circuit elements are not displayed or disappear when zooming in/out

Occurs on an old versions of mac: MacOS HighSierra 10.13.6

Open circuit not really open?

The following circuit is evaluated fine (without me being able to tell if the result is correct; this is just to test the circuit drawings):

However, when I add a piece of wire to close the loop like this:

qucat returns a ValueError: Analyzing an open circuit is impossible, even though this circuit should be more closed than before?

Symbolic calculations in a faster language

In terms of performance, QuCAT would benefit from delegating analytical calculations to a more efficient, compiled language, with the exciting prospect of simulating circuits with tens of transmon qubits and resonators for quantum computing applications.

This inevitably requires the use or development of more efficient open-source symbolic manipulation tools.

The development of the open-source C++ library SymEngine https://github.com/symengine/symengine, together with its Python wrappers, the symengine.py project https://github.com/symengine/symengine.py, could lead to rapid progress in this direction.

One should keep in mind that an increase in circuit size translates to an increase in the number of degrees of freedom of the circuit and hence of the Hilbert space size needed for further analysis once a Hamiltonian has been extracted from QuCAT.

Automatically build documentation in travis

Maybe use something like this: https://github.com/drdoctr/doctr
The problem is that when we edit and build the docs, commit and push, the description of the commit is full of html pages and it is unclear what exactly was changed.

Best supported OS?

What's the best system to run this program, Windows or Linux? I tried it on Mac Catalina, and the GUI doesn't seem to function correctly. Specifically, only capacitors are drawn. Other types of components don't have their graphical representation drawn at all. I can see the values and the labels, and I can drag them around, but they are only displayed as something invisible between two dots.

re-center dosnt work on mac

Add list of contributors to website

Improve parametrization of Josephson junctions in the GUI

Make clear in the popup menu that Josephson inductance is the default, and give the option to enter Josephson energy or critical current.

Support quantization in a different basis

.. for example the charge or flux basis, or possible optimal combinations of these, to support other types of qubits, such as flux qubits or Cooper-pair boxes.

This would go hand-in-hand with handling static offsets in charge or flux, possibly with the inclusion of DC sources as new circuit components.

Overlay arbitrary quantum states on circuit...

... rather than single-photon amplitude coherent states as implemented in the "show_normal_modes" method

Creating/pasting in mac: have to wait until mouse movement before displaying

Load GUI style network from text

I'm making a Colab demonstrating qucat for me team. I'd like to be able to show normal modes of the circuit. To do this, it seems that I need to create the Network via the GUI. It's not trivial to read from a file in Colab so I'd like to be able to create a GUI-style qucat Network by reading from e.g. the string that would otherwise be in the text file created by the GUI. Is there presently a convenient way to do that? If not, I'm sure I can add something.

Add information about performance to the docs

Current direction in normal modes is wrong.

In some circuits the direction of the current is wrong (in this example the current over the capacitator and the inductor should point in the opposite direction):
.

The voltage will have a minus sign at the corresponding branch too.
Simple_LC.txt

Documentation about plotting options

Make the dictionaries defined in plotting_settings easily editable.

Parameter sweep with initial value

When a circuit element is labelled and has a value, it is not possible to do parameter sweeps. Instead qucat returns ValueError: L_J is not the label of a circuit element, which can be inconvenient because it would require you to rewrite the circuit, instead of just reusing it

• OS: Ubuntu 18.04.2 LTS
• Example file: felix_test3.txt

Using Python 3 type annotations

Because this library targets only Python 3, it could use type annotations

from typing import List

def function(x: List[str], n: int) -> List[str]:
    """Repeat a list

    Parameters
    ----------

    x
        A list

    n
        Number of times to repeat the list.

    Returns
    -------
        The list repeated n times.
    """
    return n * x

This has the advantage that Mypy can be used to statically check the code.

---

Note that because the types are in the function signature, there's no need to put them in the docstring, which makes the numpydoc format look somewhat odd (and wastes space). That being the case, it might be a good idea to switch to Google style docstrings which can be done with automated tooling.

Control-keystrokes should also be enabled by command on mac

Accept arrays as inputs of functions (eigenfreqs, dissipation, etc...)

Note: when putting two kwargs array inputs with different sizes, an error appears saying that they should be of same size or different directions.

Robustly handle large differences in circuit component values

In circuits which have components with vastly different values, for examples in this one:

where the inductors and junctions have a factor billion in difference in inductance, the determinant calculation will yield a characteristic polynomial with slight errors. This then translates to a 3d order polyonomial where the coefficient in front of the third order term is very small, but still appears as a "fake mode" with a negative frequency.

Some of these cases are mitigated by commit 6b3a930, where we verify that both frequency and impedance of the circuit are not negative.

To be 100% sure of the results of the circuit above though, it may be more cautious to add the junction inductance (which is vastly different) as a symbolic value, and set it later in the calculation.

Maybe a way to solve this would be to compute the determinant with only symbolic values, but will the performance suffer?

Using arrows during creation on mac makes element disappear

Writing is too small on mac

Support Black-box components and Black-box quantization

For more information, see
[1] https://arxiv.org/abs/1204.0587
[2] https://arxiv.org/abs/1403.7341
[3] https://arxiv.org/abs/1505.04116

Make a help button in the GUI

One submenu will show all keyboard shortcuts,
Another will link to the quick-start present on the website

Can't detect difference between scrolling up/down on linux

Make keystroke for component rotation more obvious

Delete shortcut does not work on mac

Exit is not working on mac

Inability to analyze un-coupled degenerate oscillators

For example in this circuit:

If one replaces junctions with inductors, we have 3 LC oscillators connected in series, all with the same resonance frequency, resulting in a complete decoupling of the circuit elements from one oscillator to the next. With the current analysis method, the degeneracy in frequency makes it impossible to analyze the circuit. Because of numerical inaccuracies, no error will be raised, but un-realistic parameters will be returned:
mode | freq. | diss. | anha. |
0 | 7.26 GHz | 0Hz | 1.08 GHz |
1 | 7.26 GHz | 0Hz | 7.81 GHz |

Kerr coefficients
(diagonal = Kerr, off-diagonal = cross-Kerr)
mode | 0 | 1 |
0 | 1.08 GHz | |
1 | 5.66 GHz | 7.81 GHz |

By adding small resistance to at least one of the modes, one obtains the expected analysis results:

mode | freq. | diss. | anha. |
0 | 7.26 GHz | 1.59 MHz | 161 MHz |
1 | 7.26 GHz | 8.56 mHz | 161 MHz |

Kerr coefficients
(diagonal = Kerr, off-diagonal = cross-Kerr)
mode | 0 | 1 |
0 | 161 MHz | |
1 | 323 MHz | 161 MHz |

Support driven systems

Support new components: AC sources and receivers to derive the Hamiltonian of the driven circuit, and the operator for the observable measured at a receiver.

An interesting add-on would be the automatic application of a unitary transformation which converts the Hamiltonian to a time-independant one by finding the optimal rotating-frame. See
https://arxiv.org/abs/1808.01247

Add list of contributors to website

Ctrl+A not working for text in Ubuntu

Selecting text or numbers in the GUI parameter fields using Ctrl+A doesn't work, but would be nice to have

• OS: Ubuntu 18.04.2 LTS

CTRL-Z incompatible with ground elements

Steps to reproduce bug:

1. create the following circuit:
   
2. move the right-most ground to a new location
3. CTRL-Z to get back to initial circuit
4. close GUI
   -> This will return an error, and upon re-opening the circuit, one realizes that the left-most ground is missing :/

Detect short circuits

When a wire shorts a part of the ciruit, I would have expected a ValueError: node_minus==node_plus

Originally posted by @feschmidt in https://github.com/mgely/qucat/issues/16#issuecomment-491700520

Disable component selection when adding a wire in the GUI

Minimal steps to re-create the bug:

• One creates a component
• Insert>Wire
• Click on component rather than empty space