CCP5 Summer School 2023

Organised by CCP5 and sponsored by CECAM, the School is intended for newcomers to the science of molecular simulation and will provide a comprehensive introduction to the theoretical background as well as practical sessions on computational methods and research seminars to illustrate the versatility of simulation in modern research. There will also be opportunities for participants to present their own research.

The Summer School starts with a two-day programming course, where students can opt to take either Python or modern Fortran. After this preparation, the first five days of the main School will cover the basics of molecular simulation, and the remaining three days will be devoted to more advanced courses with options in mesoscale, ab initio, machine learning for interatomic potentials and biomolecular simulation. Course notes will be provided in electronic format. In addition to the lectures, there will be extensive practical sessions in which students will undertake computational exercises to reinforce and further explore the material.

The school will take place between 16th and 27th of July 2023 at Durham University.

A fee of £650 to cover part of the expenses will be charged to successful applicants. The school has 70 places available.

The fee partially covers accommodation, breakfast, lunch and dinner for the duration of the school. It also covers school gala dinner and poster session refreshments. The rest is covered by our sponsors. Successful candidates will need to cover their transport costs.

Please note The school can be recognised towards your doctoral training in UK, also upon request we can provide a letter for ECTS credits for your school.

Accommodation will be provided in student accommodation, single en-suite rooms for the duration of the school. If you need extra days of accommodation please let us know as soon as possible and you will have to pay for them.

NOTE Initial emails for acceptance have been send… please confirm your place by 10th of April(no need to pay at this stage). Unconfirmed places will go to the waiting list. Please check your SPAM/JUNK folder. If you have not received an email at this stage you are probably on the waiting list.

Key dates

• Application deadline: 15th of March 2023
• Acceptance decision: 31st of March 2023
• Accept place: 10th of April 2023
• Fee payments: 1st of May 2023

Organising Committee

• Dr Colin Freeman, University of Sheffield
• Prof Neil Allan, University of Bristol
• Dr Mark Miller, University of Durham
• Dr Alin Elena, STFC Daresbury Laboratory

Sponsors

Code of Conduct

We value the participation of everyone and want to ensure that everyone has an enjoyable and fulfilling experience, both professionally and personally. Accordingly, all participants of the CCP5 Summer School are expected to always show respect and courtesy to others. The CCP5 and its partners strive to maintain inclusivity in all of our activities. All participants (staff and students) are entitled to a harassment-free experience, regardless of gender identity and expression, sexual orientation, disability, physical appearance, body size, race, age, and/or religion. Harassment in any form is not acceptable for any of us.

We respectfully ask all attendees of the CCP5 Summer School to kindly conform to the following Code of Conduct:

• Treat all individuals with courtesy and respect.
• Be kind to others and do not insult or put down other members.
• Behave professionally. Remember that harassment and sexist, racist, or exclusionary jokes are not appropriate.
• Harassment includes, but is not limited to, offensive verbal comments related to gender, sexual orientation, disability, physical appearance, body size, race, religion, sexual images in public spaces, deliberate intimidation, stalking, following, harassing photography or recording, sustained disruption of discussions, and unwelcome sexual attention.
• Participants asked to stop any harassing behaviour are expected to comply immediately.
• Contribute to communications with a constructive, positive approach.
• Be mindful of talking over others during presentations and discussion and be willing to hear out the ideas of others.
• All communication should be appropriate for a professional audience, and be considerate of people from different cultural backgrounds. Sexual language and imagery are not appropriate at any time.
• Challenge behaviour, action and words that do not support the promotion of equality and diversity.
• Arrive at the events punctually where possible.
• Show consideration for the welfare of your friends and peers and, if appropriate, provide advice on seeking help.
• Seek help for yourself when you need it.

please report any issues to alin-marin.elena@stfc.ac.uk

Registration is open, please follow the following link to register. No payment request will be made until acceptance when you will receive an official email. If in doubt contact Alin Elena at alin-marin.elena@stfc.ac.uk

Programming Courses

• Introduction to Modern Fortran (6 lectures, 4 practical sessions)
• Introduction to Python

Basic Courses

• An Overview of Molecular Simulation
• Statistical Mechanics (2 lectures)
• Molecular Dynamics (3 lectures)
• Monte Carlo Methods (3 lectures)
• Free Energy Methods (3 lectures)
• Optimisation Methods
• Introduction to Force Fields
• Long timescale methods
• Advanced Free Energy methods
• Practicals (10 sessions over 5 afternoons)

Advanced Courses

• First principles simulation (6 lectures&hands-on tutorials)
• Mesoscale Methods (5 lectures)
• Simulation of Organic and Bio Molecules (6 lectures)
• Machine Learning for Interatomic Potentials (xxx)

Lecturers

• Prof Mike Allen (University of Bristol)
• Prof John Harding (University of Sheffield)
• Prof Neil Allan (University of Bristol)
• Prof Jamshed Anwar (University of Lancaster)
• Prof Paola Carbone (University of Manchester)
• Dr Livia Bartók-Pártay - (University of Warwick)
• Dr Marcus Campbell Bannerman (University of Aberdeen)
• Dr Colin Freeman (University of Sheffield)
• Dr Alin Elena (Daresbury Laboratory)

First principles simulations

• Prof Keith Refson (Royal Holloway, University of London)
• Prof Stewart Clark (University of Durham)
• Dr Barry Searle (Daresbury Laboratory)

Mesoscale methods

• Dr Ian Halliday (Sheffield Hallam University)
• Dr Michael Seaton (Daresbury Laboratory)

Simulation of organic and biomolecules

• Dr Agnieszka Bronowska
• Dr Antonia Mey
• Dr Matteo Degiacomi

Machine Learning and Interatomic Potentials

• Dr Ioan Bogdan Magdău (University of Cambridge/Newcastle University)
• Dr Alin Elena (Daresbury Laboratory)

Programming

• Dr Marcus Campbell Bannerman (University of Aberdeen)
• Dr Alin Elena (Daresbury Laboratory)

Timetable: subject to change

Advanced Seminars may be structured different depending on the lecturers.

• Professor Christoph Dellago University of Vienna
• Professor Carmen Domene University of Bath
• Dr Natalia Martsinovich University of Sheffield
• Professor Kostya Trachenko Queen Mary University London
• Professor Halim Kusumaatmaja Durham University

to reach Durham you can get some information from the University of Durham website

Student Seminar (13+2 minutes)

Posters boards will be provided, print your poster max A0 portrait.

CCP5 Student Poster Presentation Prize

CCP5 Student Oral Presentation Prize

For more information do not hesitate to contact Alin M Elena alin-marin.elena@stfc.ac.uk

An Overview of Molecular Simulation

An overview of the current state of molecular simulation with examples of special interest taken from the literature.

Introduction to force fields

Statistical Mechanics 1

In this lecture we will begin with an important question: why bother with statistical thermodynamics? We will progress to basic statistical quantities and concepts such as averages, fluctuations and correlations and how to use them in practice to calculate the physical properties of systems. This will lead us to the determination of the true statistical error for system properties obtained by simulation. We will apply these ideas to commonly calculated properties such as diffusion, radial distribution functions and velocity autocorrelation, while also examining the physical meaning of these properties. We will conclude with a look at distribution functions: how they arise and what they mean.

Statistical Mechanics 2

In the second lecture we shall look at more theoretical aspects of statistical mechanics. Beginning with the Lagrange and Hamiltonian description of classical mechanics we shall progress to the idea of phase space and the concept of a probability distribution function. This will be followed by basic applications (and associated mathematical manipulations) of the distribution function to obtain various physical properties of a system. We will examine the common ensembles (NVE, NVT and NPT) and discuss their application and interrelation. Finally we shall look at time dependence, beginning with the Liouville Equation and its connection with other time dependent equations. We shall conclude with the fluctuation-dissipation theorem.

Monte Carlo 1

Basics: The system. Random sampling. Importance sampling. Detailed balance. Metropolis algorithm in the canonical ensemble. Isothermal-isobaric ensemble. Grand-canonical ensemble. Which ensemble?

Monte Carlo 2

Practicalities: Finite-size effects. Random number generators. Tuning the acceptance rate. Equilibration. Configurational temperature. Ergodicity and free-energy barriers. Measuring ensemble averages. Examples (showing ensemble independence for the Lennard-Jones fluid)

Monte Carlo 3

(Free) Energy Barriers: Quasi non-ergodicity. Vapour-liquid phase transition as an example. Removing the interface by Gibbs ensemble MC. Free-energy barrier in the grand-canonical ensemble. Multicanonical preweighting. Histogram reweighting. Parallel tempering

Molecular Dynamics 1

Molecular dynamics: the basic methodology. Integration algorithms and their derivation. Static properties: thermodynamics and structure. Dynamic properties: correlation functions and collective properties

Molecular dynamics 2

Practical aspects of molecular dynamics - Verlet neighbour list, link cell algorithm. Calculating pressure: the virial theorem and the thermodynamic method. Estimating statistical errors: the blocking method. Symplectic algorithms and the Tuckerman-Berne-Martyna approach.

Molecular dynamics 3

Extended systems: canonical (NVT) and isothermal-isobaric (NPT) ensembles. Rigid Bodies, SHAKE, RATTLE.

Free energy methods 1

Free energy, chemical potential & thermodynamics. Applications. Essential statistical mechanics. Ensemble averages, probability distributions & simulations. Free energy, the challenge. Particle insertion & removal. Energy density distributions. The perturbation method.

Free energy methods 2

Review essential statistical mechanics. Thermodynamic integration. Potential of mean force calculations. Umbrella sampling. Absolute free energies. Free energy of liquids.Free energy of solids.

Optimization Methods

The energy landscape, geometrical optimisation and saddle points. Minimisation methods (steepest descent, conjugate gradient, genetic algorithm). Saddle-points (transition state theory, harmonic theory, nudged elastic band, dimer method).

Long timescale methods

Long timescales simulations - the problems. Transition state theory and kinetic Monte Carlo. Temperature accelerated hyperdynamics. Metadynamics.

Advanced Free Energy Methods

TBD

First-principles simulation has grown to become one of the most influential and important techniques for modelling at the atomic level. With nuclei and electrons as the basic ingredients the system is modelled at a deeper level of physics than with atoms and interatomic potentials. By explicitly including the electrons in the model and treating their interactions using quantum-mechanical laws, chemical bonding arises as an emergent phenomenon of the model. All kinds of bonding - ionic, covalent, metallic, hydrogen can be treated using the same method. The price of this accurate Hamiltonian is a computational cost orders of magnitude higher than atomic potential models. Nevertheless it is possible and convenient with modern parallel computers to simulate systems of hundreds of atoms, and perform optimization and molecular dynamics in a variety of ensembles.

In this advanced course I will provide a rapid introduction to the “nuts and bolts” of first-principles simulation. In accordance with the philosophy of the CCP5 Summer School, the aim is to attempt to open up the “black box” and explain the concepts and algorithms used. The presentation will assume a familiarity with wave mechanics at the undergraduate level and Dirac notation.

In the practicals you will be able to try for yourself using an advanced density functional code. You should be capable of running realistic calculations by the end of the course, and aware of the major aspects of setup and testing that are vital ingredients for success. The practicals will consist of a series of guided exercises with the CASTEP and CRYSTAL codes.

Synopsis

An Introduction to First-Principles Simulation

• Motivation
• Quantum-Mechanical approaches
• Density-Functional Theory
• Excited states: TD-DFT
• Electronic Structure of Condensed Phases
• Total-energy calculations
• Basis sets
• Plane waves and pseudopotentials
• How to solve the equations
• Ab-initio simulations

Practical calculations using first-principles QM: Convergence, convergence, convergence

• Convergence
• Structural Calculations
• Lattice Dynamics
• Exchange and Correlation Functionals
• Summary

Further Study

The lecture notes from the CASTEP workshop held in 2007 are available from http://www.castep.org. Links to a number of ab-initio methods and resources are available at http://electronicstructure.org/.

Mesoscale methods of modelling are capable of tackling larger length and time scales than those available using atomistic methods. By using particles considerably larger than atoms and appropriate choices of interactions between them, these techniques can readily model bulk materials and large structures at the cost of omitting some fine atomic detail. Hydrodynamics start to become more important at these scales: these modelling techniques are thus designed to ensure correct (emergent) fluid behaviour. A mesoscale model can be set up either using a ‘bottom-up’ approach from atomistic models, a ‘top-down’ approach from continuum fluid models, or both.

In this advanced course we will provide an introduction to two mesoscale methods: Dissipative Particle Dynamics (DPD) and the Lattice Boltzmann Equation (LBE) method. We will explain the origins, concepts and algorithms of both methods, as well as their applications, continuing developments and how they can be related to material models at smaller and larger scales (including those covered by the basic lectures).

In the practicals, you will be able to try out DPD and LBE using both simple ‘hackable’ codes and the general-purpose mesoscale modelling package DL_MESO. By the end of the course, you will gain insight into the capabilities of both mesoscale modelling methods. The practicals will consist of a series of guided exercises using the provided codes.

Synopsis

Introduction to the Mesoscale

• Techniques
• Physical scales
• Mesoscale simulation strategies

Dissipative Particle Dynamics (DPD)

• DPD algorithm
• Fokker-Planck formulation
• Application to simple/complex fluids
• Boundary conditions
• Thermodynamics and DPD
• Molecular dynamics and DPD

Lattice Boltzmann Equation (LBE)

• Classical Boltzmann/Boltzmann Bhatnagar-Gross-Krook (BGK) Equations
• Lattice Gas Cellular Automata (LGCA)
• Multiple component or “diphasic” LGCA
• Lattice Boltzmann Equation method
• Lattice Boltzmann BGK Equation and kinetic theory
• LBE for multi-component flow

The molecular docking and all-atom molecular dynamics simulations is an extremely powerful combination of techniques routinely applied in modern structure-based drug discovery workflows. This course will provide an introduction to how to work with protein-ligand complexes using atomistic simulations and molecular docking/virtual screening/lead optimisation tools. We will cover the principles of protein-ligand docking, virtual screening and structure-guided optimisation. This will provide a starting point for protein-ligand molecular dynamics simulations and how to analyse them. We will follow best practices from the literature (Living J Comput Mol Sci. 2019; 1(1)) and will discuss common issues around identifying good protein structures for simulations, but also how to effectively parametrise ligands and how to use MDAnalysis for the analysis of simulations of the protein-ligand complexes. We will use a blended learning environment mixing lectures and workshops.

Instructors: Agnieszka Bronowska (Newcastle), Antonia Mey (Edinburgh), Matteo Degiacomi (Durham)

Synopsis

The two day Machine Learning for Molecular Simulation activities consist of introductory lectures, practicals, and advanced seminars. The first two days are structured similarly, starting with a lecture with a high level overview and some theoretical background, followed by an introduction to the practicals. Students will use Jupyter notebooks on the Deepnote.com cloud provider. Students will be introduced to two popular methods for fitting interatomic potentials: the Gaussian Approximation Potential (GAP) and the Atomic Cluster Expansion (ACE). On the first day, the practical is about modelling organic solvents, specifically ethylene carbonate and ethyl methyl carbonate which are relevant for rechargeable battery applications. The second day again starts with a higher level overview of machine learned potential applications in chemistry, followed by a more detailed look at the ACE framework. The practical practical session will involve modelling inorganic perovskites for solar panel applications.

Lectures (day 1):

• Gaussian Approximation Potentials (GAP)
• GAP for Mixed Molecular Liquids

Lectures (day 2)

• Machine Learned Force Fields
• The Atomic Cluster Expansion - Applications to Perovskites

to be added soon