Reading
Spend enough time reading!
General
- A. Zagoskin: Quantum Theory of Many-Body Systems, Springer 1998.
Nice introduction. Chapters 1 and 3.1 are mandatory reading for most of what we do. - J. W. Negele and H. Orland: Quantum Many-particle Systems,
Writing
A BSc thesis should not be too long. Aim for 12–20 pages.
The intended audience is past you, i.e., a hypothetical student of your level before you started the work.
Structure
A good structure for a thesis (or paper) is:
- Introduction: where you summarize what was known a priori, i.e., where things stood before your work. You should:
- embed your topic in a wider scientific concept ("zoom in" to your problem)
- describe the problem and why it is interesting
- what attempts have been made before
- System and Method: document the system and which methods and techniques you have used in tackling the problem. Resist any temptation to write a textbook, instead heavily cite appropriate sources! You should give the information needed to reproduce and understand the results, no more, but also no less. You should:
- describe the system under study, and the relevant properties needed;
- give definitions of all the quantities that are measured/computed;
- expose the exact method and setup used to measure/compute these quantites.
- Results and Discussion: the main part of your work. Give enough evidence for supporting your conclusions, but do not include irrelevant info. You should:
- make high-quality figures/tables for your results (see later).
- describe what you are seeing in each figure in the text
- discuss what these results mean in the text.
- Conclusions and Outlook: where you summarize what is now known a posteriori, i.e., where things stand after your work. Describe:
- What new things have you learned from your work,
- How these conclusions agree or disagree with previous or related work,
- which new questions arise from your work and are interesting topics for follow-up studies ("zoom out")
Your work should present the shortest, clearest, least cluttered path from your problem (introduction) to your conclusions. Detours, e.g., technical discussions, codes, parameter dumps, tangential observations, etc., go into appendices.
The abstract is a short version of each of these parts. Life hack: abstract = all of the parts above, each one condensed into 1–2 sentences. An abstract must be specific: e.g., "we observe a X% increase in quantity Y for system Z", not "we observe a significant increase in relevant quantities".
Do not start from the introduction, start writing the results section first.
Text and Language
If you want to be a writer, you must do two things above all others: read a lot and write a lot.
– Stephen King, On Writing
We recommend writing your thesis in English since it is the lingua franca in physics today. Some advice particularly relevant to German speakers:
- Simplicity is a virtue: complicated language typically confuses rather than impresses.
- Use the active voice ("I/We/They show that ..." rather than "It was shown that ...").
- Prefer short sentences, avoid complicated relative clauses.
Every single claim in your work has to be sourced, i.e., it either follows from other people's work, in which case you cite that work, it follows from your own data, in which case you have to present that data, or it follows from your own derivation, in which case you present that derivation.
Everything you took from somewhere else has to have a citation attached to it. Citation style but must be consistent for all references. In our field Phys. Rev style is commonly used such as e.g. M. Wallerberger et al., Comput. Phys. Commun. 235, 2 (2019) https://doi.org/10.1016/j.cpc.2018.09.007. Cite liberally – "citations are free" – but only cite relevant work.
Figures and tables
Figures and tables for your results are the centerpiece of every paper/thesis. You should:
- have clearly labelled axes, as well as a legend (either in the caption or in the figure) if there is more than one curve per plot
- add a detailled caption: you should again summarize which exact quantities are plotted for which system/parameters using which method, as well as any "oddities" that may trip up the reader. The figure should almost be "self-contained", i.e., understandable from the caption alone. It is not uncommon for a caption to be longer than the figure.
- use panels or multiple lines per plot to present things that belong together, but avoid clutter.
Table of contents