/Lecture-Notes

Repository for figures and tex files of the book "Modern applications of machine learning in quantum sciences"

Primary LanguageJupyter Notebook

This is a repository for figures and tex files of "Modern applications of machine learning in quantum sciences"

Open In Colab

List of our figures:

  • 1.3 % of ML-based articles in the selected fields in years 2000-2021 - A. Dawid
  • 2.1 Plots of the (a) binary cross-entropy and (b) mean-squared error - A. Dawid
  • 3.1 Ising model (to reproduce, go to the Notebook A1 from the school GitHub) - R. Koch
  • 3.2 IGT (to reproduce, go to the Notebook A1 from the school GitHub) - R. Koch
  • 3.3bc PCA (to reproduce, go to the Notebook A1 from the school GitHub) - R. Koch
  • 3.6 Learning by confusion (to reproduce, go to the Notebook A3 from the school GitHub) - R. Koch
  • 3.7 Pediction-based method (to reproduce, go to the Notebook A3 from the school GitHub) - R. Koch
  • 4.1 Toy example of a labeled two-dimensional data set - A. Gresch
  • 4.3 The kernel form makes a difference - A. Gresch
  • 4.6 Selection of new candidate points via BO using Upper Confidence Bound acquisition function - K. Nicoli
  • 4.7 Search for the optimal kernel - A. Dauphin
  • 6.4a Parameter update for a random walker - B. Requena
  • 7.4 Inverse Schrödinger problem solved using dP - J. Arnold
  • 8.4 Perceptron capacity by Cover - M. Gabrie
  • 8.11 Illustration of a quantum circuit (only pdf) - P. Stornati
  • 8.16 Variational quantum simulation - P. Stornati

List of the figures by FESIDO Studio Graficzne in folder graphical_designer:

  • 1.1 Traditional programming vs ML
  • 1.2 AI vs ML vs DL
  • 1.4 Interplay between AI, quantum computing, many-body physics, and quantum chemistry
  • 1.5 Contents of these Lecture Notes
  • 1.6 Tree of dependencies between chapters (added in v2)
  • 2.2 Learing rate as a hyperparameter
  • 2.3 Under- and overfitting
  • 2.4 The bias-variance trade-off
  • 2.5 Geometric construction of SVMs
  • 2.6 Neural network (modified in v2)
  • 2.7 Convolutional filter
  • 2.8 Autoencoder
  • 2.9 Recurrent neural network
  • 2.10 Backpropagation (added in v2)
  • 3.3a Phase classification with PCA
  • 3.9b Interpretation of neural networks via bottlenecks
  • 4.2 A linear SVM applied to non-linearly separable data
  • 4.4 Bayesian neural network
  • 4.5 Bayesian optimization
  • 4.8 Three main classes of problems tackled with BO and GPRs
  • 4.9 BO and GPRs for feedback loops
  • 5.1 Scheme of a restricted Boltzmann machine
  • 5.2 Autoregressive neural quantum state
  • 5.3 Recurrent neural-network architecture as a neural quantum state
  • 5.4 Expressive capacity of neural quantum states
  • 5.5 Schematic representation of the of various ansätze
  • 6.1 Overview of the basic reinforcement learning setting
  • 6.2 Short-term and long-terms rewards in reinforcement learning algorithms
  • 6.3 Schematic representation of the episodic and compositional memory of various projective simulation agents
  • 6.4b Evolution of various walker policies
  • 6.5 Performance of AlphaGo and AlphaGo Zero
  • 6.6 Reinforcement learning for quantum feedback of an optical cavity
  • 6.7 Reinforcement learning for circuit optimization
  • 6.8 Reinforcement learning for quantum error correction
  • 6.10 Reinforcement learning to find optimal relaxations
  • 7.1 Machine learning influences physics
  • 7.2 Standard vs differentiable programming
  • 7.5 Sketch of a normalizing flow (modified in v2)
  • 7.6 Volume transformation (added in v2)
  • 7.14 Illustration of the Hamiltonian learning of a one-spin system
  • 7.17 Automated design on experiments (added in v2)
  • 8.1 Physics influences machine learning
  • 8.2 Statistical physics toolbox for understanding machine learning theory
  • 8.3 Generalization error in classical and modern regimes
  • 8.6 Schemes of a committee machine and random feature model
  • 8.11 Illustration of a quantum circuit diagram
  • 8.12 Quantum machine learning
  • 8.13 Realization of the famous Shor algorithm in a real quantum computer
  • 8.14 Quantum support vector machine enhanced by a quantum device
  • 8.15 Variational optimization of quantum circuits

Folder tex_files contains:

  • arXiv_v1.zip - zipped complete set of tex files and associated ones being a basis for the arXiv v1 submission (we recommend loading it with Overleaf).
  • arXiv_v2.zip - version 2.

Moreover, folder colors contains:

  • colors_dict.pkl - pickled dictionary with our RGB-coded five main colors (green, purple, yellow, orange, blue) and their three shades (dark, medium, light),
  • colors_1D.pkl - the same colors in 1D array,
  • colors_2D.pkl - colors in 2D array,
  • Jupyter notebook that shows how to unpickle them.

Finally, folder fonts contains:

  • set of fonts called New Hero used for text in plots,
  • Jupyter notebook that shows how to use them with Python.

Version 2 update (22.06.2022)!

  • We wrote a new section 2.5 on backpropagation in NNs (with a new fig. 2.10)
  • We expanded section 7.2.2 on normalizing flows (with a new fig. 7.6)
  • We wrote a new section 7.3.4 on automated design of experiments (with a new fig. 7.17) and expanded the outlook of 7.3 (ML for experiments).
  • We added the appendix C concerning kernel methods.
  • We added a tree of dependencies between chapters to allow the reader to choose what they want to read in a more informed way (fig. 1.6)
  • We modified slightly two figures: 2.6 NN and neuron and 7.5 Sketch of a normalizing flow.
  • We added new references following feedback from the community.