On Algorithms: Metropolis, Heat Bath, Molecular Dynamics and Hybrid Monte Carlo

For simplicity, we will start with the matrix harmonic oscillator. The action is given explicitly by
S_{\rm Bo}=\int_{0}^{\beta} dt {\rm Tr}\big[\frac{1}{2}\dot{\Phi}^2+\frac{1}{2}B\Phi^2\big].
In statistical mechanics (quantum field theory) we calculate the partition function (Euclidean path integral) given by
Z=\sum_ne^{-\beta E_n}=\int {\cal D}\Phi\exp\big(-\frac{S_{\rm Bo}[\Phi]}{\hbar}\big).
This corresponds to the canonical ensemble where the temperature $\beta\equiv 1/\hbar$ is kept fixed. Here we usually employ the Metropolis algorithm, which is the most general of all Monte Carlo algorithms, to calculate this partition function. This is given explicitly by the probability distribution
\begin{equation}P(\Phi_n\longrightarrow \Phi_n^{\prime})={\rm min}(1,\exp(-\Delta S_{\rm Bo}))~,~\Delta S_{\rm Bo}=S_{\rm Bo}[\Phi^{\prime}]-S_{\rm Bo}[\Phi].\label{metropolis}\end{equation}
In more detail, the Metropolis algorithm consists of the following steps:

  • We start from some initial configuration ${\Phi}_0$.

  • We propose a new configuration $\tilde{\Phi}$.

  • We compute the variation of the action $\Delta S_{\rm Bo}=S_{\rm Bo}[\Phi]-S_{\rm Bo}[\Phi_0]$.

  • We accept the new proposal with a probability $P$ given by (\ref{metropolis}).

  • We repeat starting from step $2$ until we reach thermalization.

  • We measure thermalized configurations which are used to calculate expectation values of physical observables.

The meaning of equation (\ref{metropolis}) is as follows. After we compute the variation of the action in step $3$ we check its sign. If ${\Delta} S_{\rm Bo}<0$ then we should accept the proposal since it had resulted in a decrease of the action and thus getting us closer to the minimum. Otherwise, if ${\Delta} S_{\rm Bo}>0$ we accept the proposal only with a probability given by the Boltzmann distribution $\exp(-{\Delta} S_{\rm Bo})$. This is the part which is simulating quantum mechanics and it is implemented numerically via the Von Neumann method. In other words, choose a uniform random number $r$ between $0$ and $1$ and compare it to the Boltzmann distribution $P=\exp(-{\Delta} S)$. If $r<P$ then accept the proposal otherwise reject the proposal.

In many circumstances such as non-local theories it is found that the micro-canonical ensemble is much more favorable than the canonical ensemble. In this case it is the energy that is kept fixed and the preferred algorithm in this case is the hybrid Monte Carlo algorithm which synthesizes together the Metropolis algorithm, the molecular dynamics algorithm and the heat bath algorithm. This alternative formulation relies on the partition function
  Z=\int {\cal D}\Phi{\cal D}P \exp\bigg(-\frac{1}{2}\sum_{n=1}^{\Lambda}{\rm Tr}P^2_n-\frac{S_{\rm Bo}[\Phi]}{\hbar}\bigg).
The Euclidean action $S_{\rm Bo}$ acts therefore as a potential term while the new fields $P_n$ act as conjugate momenta associated with the generalized coordinates $\Phi_n$, i.e. $\sum_n{\rm Tr}P_n^2/2$ is a kinetic energy. The Hamiltonian is then given by
  H_{\rm Bo}=\frac{1}{2}\sum_{n=1}^{\Lambda}{\rm Tr}P^2_n+\frac{S_{\rm Bo}[\Phi]}{\hbar}.
This Hamiltonian defines a canonical evolution in a fictitious time denoted by $\tau$. Indeed, Hamilton equations of motion read explicitly
  \frac{d({P}_n)_{ab}}{d\tau}=-\frac{\partial H_{\rm Bo}}{\partial (\Phi_n)_{ba}}\Leftrightarrow -\frac{d({P}_n)_{ab}}{d\tau}=\frac{\partial S_{\rm Bo}}{\partial (\Phi_n)_{ba}}\equiv(F_n)_{ab}\nonumber\\
   \frac{d({\Phi}_n)_{ab}}{d\tau}=\frac{\partial H_{\rm Bo}}{\partial (P_n)_{ba}}\Leftrightarrow  \frac{d({\Phi}_n)_{ab}}{d\tau}= ({P}_n)_{ab}.
The force $F_n$ is given explicitly by

These equations are solved using a molecular dynamics algorithm. In particular, the so-called leap-frog algorithm, which preserves the phase space volume and reversibility and only break Hamiltonian conservation, gives us explicitly the equations  
  &&(\Phi_n)_{ab}(\tau+\delta\tau)=(\Phi_n)_{ab}(\tau)+\delta\tau (P_n)_{ab}(\tau+\frac{\delta\tau}{2}).\label{lf2}
We remark that the force $F_n$ at the instant $\tau$ is needed to advance all the configurations $\Phi_n$ from $\tau$ to $\tau+\delta \tau$. Thus, we need to calculate the force $F_n$ for all $n$ at instant $\tau$, then apply equations (\ref{lf1}) and (\ref{lf2}) to advance all $\Phi_n$ from $\tau$ to $\tau+\delta \tau$, after which we calculate again the force $F_n$ for all $n$ at instant $\tau+\delta \tau$, before we finally  apply equation (\ref{lf3}) to advance all $P_n$ from $\tau+\delta \tau/2$ to $\tau+\delta \tau$.

The leap-frog will need two extra parameters: the step $\delta \tau$ and the number of iterations which we call $L$. The total time of the motion is given by $T\equiv L\delta \tau$. The initial values of $\Phi_n(0)$ and $P_n(0)$ at time $0$ will be specified and then by applying the above leap-frog algorithm we will obtain the final values of $\Phi_n(T)$ and $P_n(T)$ at time $T$.

Two essential remarks can be stated now:

  • The molecular dynamics involves in an obvious way a systematic error.

  • The molecular dynamics probes only classical physics.

These two problems can be solved at once via the so-called hybrid Monte Carlo algorithm which is the most general of all Monte Carlo algorithms. This algorithm involves in an essential way the Metropolis algorithm. Indeed, the configuration $\Phi_n(T)$ obtained from the molecular dynamics algorithm is the solution which we will propose as a possible update $\Phi_n^{\prime}$ to the Metropolis algorithm (\ref{metropolis}). The probability distribution in this case becomes
P(\Phi_n\longrightarrow \Phi_n^{\prime}, P_n\longrightarrow P_n^{\prime})={\rm min}(1,\exp(-\Delta H_{\rm Bo}))~,~\Delta H_{\rm Bo}=H_{\rm Bo}[\Phi^{\prime},P^{\prime}]-H_{\rm Bo}[\Phi,P].\label{hybrid}
As it turns out, the  conjugate momenta $P_n$ should be  updated using a heat bath algorithm in order to avoid ergodic problems, i.e. to be able to reach every point in phase space. This means in particular that $P_n$ should be updated directly from a Gaussian distribution. This is indeed possible since the path integral over the conjugate momenta $P_n$ is only Gaussian. In fact, this path integral can be given by a closed-form expression of the form
  Z&=&\int {\cal D}P \exp(-\frac{1}{2}\sum_{n=1}^{\Lambda}{\rm Tr}P^2_n)\nonumber\\
  &=&\bigg(\int \prod_{i}dP_{ii}\prod_{i>j}dP_{ij}dP_{ij}^* \exp(-\frac{1}{2}\sum_{i=1}^NP_{ii}^2-\sum_{i>j}P_{ij}P_{ij}^*)\bigg)^{\Lambda}.
All these integrals are Gaussian of the form
\frac{a}{\pi}\int dzdz^*\exp(-azz^*)=\int_0^1 dv_1 \int_0^1 dv_2.
These two equations show that $v_1$ and $v_2$ are uniform random numbers between $0$ and $1$ and thus $z$ is a complex random number given by the formula
z=\sqrt{-\frac{1}{a}\ln v_1}(\cos 2\pi v_2+i\sin 2\pi v_2).
The components of the conjugate momenta $(P_n)_{ij}$ are given by Gaussian random numbers of this form.

The action $S=b{\rm Tr}P^2_n$ at each lattice point can also be rewritten as an eigenvalue problem. As it turns out, the eigenvalue distribution is given by the Wigner semi-circle law
In summary, the {hybrid Monte Carlo algorithm} is an algorithm in which two crucial extra steps are added to the Metropolis algorithm:

  • The step number $2$ in the Metropolis algorithm is implemented via the {\bf molecular dynamics algorithm}. In other words, the new configurations ${\Phi}_n^{\prime}$, $P_n^{\prime}$ are given by the solutions  ${\Phi}_n(T)$, $P_n(T)$ of the molecular dynamics problem with   ${\Phi}_n(0)$, $P_n(0)$ as the initial conditions.

  • As it turns out the path integral over the conjugate momenta $P_n$  should be sampled using the so-called {\bf heat bath algorithm} in order to avoid the ergodic problem, i.e. to be able to reach every point in phase space.

Mecanique Analytique 2021/2022

تصحيح الامتحانات مع البارام 



Série4: Équation d'Euler-Lagrange et d'Hamilton (suplément)

Série3: Équations d'Hamilton 

Cours6: Mécanique Hamiltonienne


Nouveau Empoli du Temps: Dimanche 10h30mn-12h30mn (Cours), 13h30mn-15h00mn(TD). 

Série2: Équations d’Euler-Lagrange 

Série1: Principe de Travail Virtuel de D’Alembert

Cours5: Equations de Lagrange 

Cours4: Coordonnées Généralisées et Déplacement Virtuel: Principe de Travail Virtuel de D’Alembert 

Cours3: Degrés de Liberté et Contraintes Holonomes: Travail Virtuel des Forces de Contraintes 

Cours2: Lois de Conservation 

Cours1: Particules Ponctuelles: Cinématique et Dynamique 


Emploi du temps

Dimanche: 10h30mn-11h30mn (Cours)

Dimanche: 12h30mn-13h30mn (Cours)

Lundi(G1): 12h30mn-13h30mn (TD)

Mardi(G2): 12h30mn-13h30mn (TD)


Herbert Goldstein, Classical Mechanics.

Walter Greiner, Classical Mechanics: System of Particles and Hamiltonian Dynamics.

Landau and Lifshitz, Mechanics.

En Francais:

Amiot et Marleau, Mecanique Analytique

En Arabe:

Badis Ydri, Fundamental Physics



Equations de Lagrange et formalism Lagrangien virtuel (particules ponctuelles, symmetries et principes de conservation, degre de liberte, contraintes holonomes, coordonnees generalisees, espace de configuration, principe de travail virtuel, Lagrangien, calcul variationnel, principe de moindre action d'Hamilton, theorem de Noether, invariance de jauge).


Equations de  Hamilton et formalism Hamiltonien canonique (lois de conservation et constantes du mouvement, transformations de Legendre, Hamiltonien, moment generalise,espace de phase,  equations de Hamilton, transformations canoniques, forme symplectique, crochets de Poisson, theorem de Liouville, equation d'Hamilton-Jacobi,).

Applications physiques (chute libre, probleme a deux corps et force centrale, rotation de corps solide, oscillation simple et theorie des perturbations, electrodynamique classique, relativite restreinte et espace-temps, chaos et mecanique non-lineaire, mecanique numerique, cosmologie, theorie des champs).   




Bell's theorem: A bridge between the measurement and the mind/body problems


In this essay a quantum-dualistic, perspectival and synchronistic interpretation of quantum mechanics is further developed in which the classical world-from-decoherence which is perceived (decoherence) and the perceived world-in-consciousness which is classical (collapse) are not necessarily identified. Thus, Quantum Reality or "{\it unus mundus}" is seen as both i) a physical non-perspectival causal Reality where the quantum-to-classical transition is operated by decoherence, and as ii) a quantum linear superposition of all classical psycho-physical perspectival Realities which are governed by synchronicity as well as causality (corresponding to classical first-person observes who actually populate the world). This interpretation is termed the Nietzsche-Jung-Pauli interpretation and is a re-imagining of the Wigner-von Neumann interpretation which is also consistent with some reading of Bohr's quantum philosophy.  

Comments: This essay is a summary of the main ideas found in the book (Philosophy and the Interpretation of Quantum Physics) published in 2021 with Institute of Physics (IOP) here this https URL. The essays arXiv:2008.09500 [hep-th], arXiv:2007.04489 [physics.hist-ph] and arXiv:1811.04245 [quant-ph] contain further discussion of other related ideas found in the book.


Subjects: History and Philosophy of Physics (physics.hist-ph); Quantum Physics (quant-ph).

Quantized Noncommutative Geometry from Multitrace Matrix Models


In this article the geometry of quantum gravity is quantized in the sense of being noncommutative (first quantization) but it is also quantized in the sense of being emergent (second quantization). A new mechanism for quantum geometry is proposed in which noncommutative geometry can emerge from "one-matrix multitrace scalar matrix models" by probing the statistical physics of commutative phases of matter. This is in contrast to the usual mechanism in which noncommutative geometry emerges from "many-matrix singletrace Yang-Mills matrix models" by probing the statistical physics of noncommutative phases of gauge theory. In this novel scenario quantized geometry emerges in the form of a transition between the two phase diagrams of the real quartic matrix model and the noncommutative scalar phi-four field theory. More precisely, emergence of the geometry is identified here with the emergence of the uniform-ordered phase and the corresponding commutative (Ising) and noncommutative (stripe) coexistence lines. The critical exponents and the Wigner's semicircle law are used to determine the dimension and the metric respectively. Arguments from the saddle point equation, from Monte Carlo simulation and from the matrix renormalization group equation are provided in support of this scenario.  

Subjects: High Energy Physics - Theory (hep-th); General Relativity and Quantum Cosmology (gr-qc); High Energy Physics - Lattice (hep-lat)

On the AdS/CFT correspondence and quantum entanglement 


String theory provides one of the most deepest insights into quantum gravity. Its single most central and profound result is the gauge/gravity duality, i.e. the emergence of gravity from gauge theory. The two examples of M(atrix)-theory and the AdS/CFT correspondence, together with the fundamental phenomena of quantum entanglement, are of paramount importance to many fundamental problems including the physics of black holes (in particular to the information loss paradox), the emergence of spacetime geometry and to the problem of the reconciliation of general relativity and quantum mechanics. In this article an account of the AdS/CFT correspondence and the role of quantum entanglement in the emergence of spacetime geometry using strictly the language of quantum field theory is put forward.  


Appeared under the title (The AdS/CFT correspondence) as the final chapter of the second volume of the book (A Modern Course in Quantum Field Theory) published with Institute of Physics (IOP) in 2019. Volume 1: this https URL. Volume 2: this https URL  Subjects:                 

    High Energy Physics - Theory (hep-th); High Energy Physics - Experiment (hep-ex); High Energy Physics - Lattice (hep-lat); High Energy Physics - Phenomenology (hep-ph)