Table of contents

Back to the main page 1. How to write the paper 2. Grading 3. Figures

How to write the paper

The report in this course will be written as one big research paper using the LaTeX template (also availiable on Overleaf if you want to use this) for the Astronomy and Astrophysics journal giving you some experience in writing a proper paper. If you prefer another journal style like Physical Review D (Latex files; Overleaf) or Monthly Notices of the Royal Astronomical Society (Latex files; Overleaf) you can also use that. You will hand in the code and paper after each milestone and get feedback. The paper should have sections similar to the ones listed below:

  • Introduction: Write a short introduction (and an abstract) for the whole project and for each of the milestones. For the milestones this serves as a motivation for the milestone (put it in context of the greater project).
  • Theoretical background: Write a short subsection for each milestone with the relevant theory. You can copy the equations from the website if you need to (but all text should be your own). You don't have to add derivations of the equations.
  • Implementation, numerical methods and tests: Write a short subsection (if relevant) for each milestone on the numerical implementation you have done. Describe briefly what you have done in the code. What are the inputs needed, what are the steps you are doing and mention if you used any special algorithms etc. (if you just used the ODE solver or a spline just briefly say for example that we used a Runge-Kutta 4 ODE solver and created a qubic spline, you don't have to add much more than that. You don't have to comment on every single function you have in the code. This is meant to be a brief overview with just the most important details).
  • Results: Write a short subsection for each milestone presenting and discussing the main results. Try to use the physics we have learned in class to explain the plots: e.g. what does it mean that $\frac{d\mathcal{H}}{dx}$ crosses zero, what are the different regimes seen in the evolution of $\delta_b(x,k)$, why does the matter power-spectrum have a peak, etc.
  • Conclusions: Write a short conclusion section where you summarize and discuss the results obtained in the whole project, possible way forward (improvements) etc.

You are free to structure the paper as you want as long as it contains all the stuff listed above. The paper will only be graded in the end, but you can get feedback after each milestone to get an idea of where you are at.

Figure: Example on how one can structure the paper. This example can be found in this latex file (requires the A&A LaTeX package to compile).

You should also hand in the code. For the numerical work you get max score for correctly having completed the full assignment. For (bigger) mistakes or if there are things that you did not manage to do then there will be points subtracted. How much I subtract will depend on how the results are discussed in the report (getting wildly wrong results and not commenting on it is really bad). I don't subtract or add points for how the coding is done, what algorithms are used etc. but I will give feedback for those who are interested. The final point is something I should not have to mention as it should be obvious: but you don't copy text from anywhere (from Wikipedia, fellow students or whatever). All words should be your own. The only exceptions are equations - you can copy the TeX of the equations from this website, you don't have to retype them yourself. Also this is an individual project. You can (and should) help each others if you have problems, but at the end of the day its you that should do your own coding.


In each of the milestones there will be a score associated with every deliverable. The total amount of points for each milestone is 100 points. The grade is computed from A = 92-100 %, B = 77-91 %, C = 58-76 %, D = 46-57 % and E = 40-45 %. We do not score the code implementation separately, but together with the results you obtain (so having minor bugs leading to not perfect results is much better than having perfect results gotten from a code that is not supposed to give correct results).

For all the deliverables in the milestone you can get up to 55 points. Thus if you just do the coding, getting correct results and just presenting the bare results then that is enough to get you a very weak C. If you want to do better you need to show that you understand the physics of the results that you are computing and present/discuss everything in a good way. 45 points are given for how you present things in the paper. For each milestone we will grade as follows:

  • Introduction and theory background: 10 points
  • Implementation, numerical methods and tests: 5 points.
  • Discussion of the results/plots: 30 point.

For the abstract/introduction/conclusion to the whole project there is an additional 15 points. There is 5 points for having relevant citations (and used in a correct/sensible way). There is also a pot of 30 points that you can get as extras in the end (if you did something very good; if you did something extra; etc). There is a total pot of 450 points to be gotten on the whole project, but you get a full score with just 420 points.


When you produce a plot remember that it should be formatted such that its easy to "get information" out of the plot. To get full score for a plot it should have:

  • Axes labels and labels for the different curves (alternatively explain what the different curves are in the caption). Use colors that are easy to distinvish.
  • Use a sensible scale for the $x$ and $y$ axes. E.g. if you are showing something that varies over many orders of magnitude then you have to use a log-scale. If you are showing something that varies between 1 and 2 and you should use a linear scale.
  • Use a sensible unit for whatever is plotted. E.g. giving the age of the Universe in seconds is not very sensible, using years would be better and using giga-years would be best. Likewise specifying cosmological distance scales in meters is not very informative.
  • Use horizontal or vertical lines to specify special values/times where relevant to discuss the plot better (e.g. if you show a plots of the density parameters $\Omega_i$ then showing the equality times would be very relevant; if you show the optical depth then showing $\tau=1$ would also be very relevant).
  • If relevant you might want to scale the quantity you are plotting (e.g. if plotting something that varies between 1 and 2 together with something that varies between 1 to 1000 then scaling the second curve might be a good idea).
  • Have a caption that explains what the plot is of.