28

I'm just starting my PhD, and during a sort of informal journal club, my professor told us that we should always try to read some new papers in our field (high energy physics, and general relativity and quantum cosmology; hep-th and gr-qc on arXiv). How does one practically do this? On the arXiv "new" section? Via the arXiv newsletter, with its terrible formatting?

I would be surprised if there aren't any "latest cool things in this particular area of physics" with easy-to-read formatting: an Instagram post, a podcast, or even just a rich text email.

8
  • 1
    On the technical side, there's dailyarxiv.com that provides a nicer interface for the list of new submissions on arxiv for a certain date range. Then there's arxiv-vanity.com which reformats pdf to a somewhat more mobile-friendly format.
    – And R
    Commented May 10, 2023 at 13:01
  • 4
    Frame challenge: as an overworked PhD student, passively staying on top of "developments in my field" beyond active literature searches related to the studies I am performing seems like a huge workload even though it is just 15-30 minutes per day reading titles and looking through abstracts. If I spent those 15 minutes working on my thesis instead I'd probably finish a lot sooner. My 2¢, don't know if it deserves an answer. Commented May 10, 2023 at 13:03
  • 8
    Does this answer your question? How to stay on top of recent literature? Commented May 10, 2023 at 13:04
  • 2
    My advice: Try to read one key paper that you can use as a reference to build your research and dissertation on (your dissertation must be novel but it also must relate to past research) and try to understand it really well. If relevant, program up the algorithm or equations if you can. I also tend to agree with @FerventHippo's comment as well. Commented May 10, 2023 at 19:26
  • 2
    What's hep-th and gr-qc? Commented May 11, 2023 at 4:11

11 Answers 11

14

While the suggestion in other questions to use Google Scholar might be good in other fields, I take from your question that you are in hep-th/gr-qc. In this case, keeping up with new postings on the arXiv is the way to go. Anything worth your time will be posted to arXiv and this is generally the first place it will appear (Google Scholar alerts tend to be at least a couple of days behind the curve).

I personally just follow the daily digest e-mails. It is easy enough to do, and the mail sits in your inbox as a reminder that you haven't read the arXiv yet that day. However, there are more fancy solutions. One some of my colleagues are very enthusiastic about his Benty Fields. Besides simply reading the daily digests in a nicer format, the site can track your interest and recommend papers you are likely interested in. I've tried it for a bit, and found it was very nice. Nonetheless, I have reverted to just using the plain digest e-mail, as they fit my daily work flow better.

5
  • 2
    Google Scholar alerts tend to be at least a couple of days behind the curve - I also follow the arXiv (math) rather rely on than Google Scholar, but is this really the main reason for you? Is seeing a paper a few days late that big of a deal in your field?
    – Kimball
    Commented May 10, 2023 at 12:46
  • 5
    @Kimball, I think the difference is due to a social effect not the "research gap" of 1 day. On the day a paper appears people will say: "Hey have you read this paper. What do you think?" It is nicer to not always be the one who isn't in the know.
    – Kvothe
    Commented May 10, 2023 at 12:50
  • 1
    @Kvothe I agree it's good to be able to discuss papers on the fly, and sometimes this situation arises, but my experience is usually when people bring up new papers, either they've been public for a little while, or I know about the results from talks/word-of-mouth already. Maybe this is a difference between math and physics?
    – Kimball
    Commented May 10, 2023 at 14:21
  • 2
    @Kimball Yes, this ends up being more of a social thing than a research thing. E.g. if you want to congratulate someone in your network on getting a nice paper out, it works better the day of, not a week later.
    – TimRias
    Commented May 10, 2023 at 15:21
  • I come back a month later to remark that I've fallen in love with Benty Fields. Now all of my group uses it for JC purposes: I find it much easier to read than arXiv and much less invasive than Google Scholar email. Commented Jun 7, 2023 at 9:04
25

One strategy that I can recommend: find your advisor/advisors on GoogleScholar, click Follow and tick the boxes "New citations" and "New articles related to this author's research". Google will then send you email notifications with papers that tend to be the most relevant to you.

Later, as you identify other important names in your field, follow them too.

1
  • 9
    +1 | GoogleScholar #Follow #NewCitation #NewArticle Commented May 9, 2023 at 18:47
16

I just check my field, math/combinatorics daily on arxiv. No newsletters or emails, just check as if it was a news site. I read perhaps 2-4 abstracts every day, and add the interesting ones to a todo list. Once I have some time, I read a bit more carefully, and add the results to a website I manage. This process has helped me immensely.

3
  • 2
    Is the website public, and if yes, do you mind sharing it?
    – Neinstein
    Commented May 10, 2023 at 8:54
  • 1
    +1, individual field and subfields have their own options, by their own people. Sorry, but what runs in my field/subfields won’t help you in your subfields. We have quasi-personal/professional blogs where new papers of interest are posted regularly… in fact, that’s the point of those blogs. We also have a paper repository (the ADS, at adsabs.harvard.edu ) that really helps by auto-listing and auto-linking the referenced papers, and inversely, the later citations that reference a past paper. Too bad if your field doesn’t have an equivalent. Commented May 10, 2023 at 12:34
  • 3
    @Neinstein Sure, it is at symmetricfunctions.com Commented May 10, 2023 at 14:19
11

I've also found the arXiv daily mailings to be terrible; most the papers are irrelevant for me and the formatting isn't very pretty. My solution is to use RSS feeds to filter out the content I actually want.

  1. Download an RSS reader on your device (I prefer QuiteRSS)

  2. Set up filters using different keywords, authors, etc. via the arXiv search API. Here's an example of an RSS feed URL that filters this way:

    http://export.arxiv.org/api/query?search_query=((all:boson+AND+(all:stars+OR+all:star))+AND+(cat:hep-th+OR+cat:gr-qc+OR+cat:hep-ph))&sortBy=lastUpdatedDate&sortOrder=descending&max_results=100

    The above link would find the 100 newest papers on hep-th, hep-ph, and gr-qc with the keyword "boson" and "star"/"stars" somewhere in the title or abstract (you can also filter by author, date, and so on... see the arXiv documentation for more options). Add these types of links to your RSS reader.

  3. If you wish, you can also just subscribe to the daily mailings over RSS (without the filter functionality) using the instructions here.

As an added bonus, you can also keep up with your favourite Stack Exchange posts via RSS. Most published journals also have RSS feeds you can use.

3
  • 4
    RSS is my go-to method as well. Makes it very easy to filter out stuff of little interest, flag interesting works for follow-up, and read things on your own schedule. (Of course, it is also easy to build up a backlog of papers to read...)
    – Anyon
    Commented May 9, 2023 at 19:13
  • 2
    Thunderbird works as an RSS reader as well (though very poorly documented and with an unintuitive user interface). One can organise/group the RSS feeds into different folders/inboxes instead of having either all RSS feeds mixed up or having to check each RSS feed manually (did I say this powerful feature is very poorly documented?). Commented May 10, 2023 at 7:11
  • 1
    Zotero doubles as RSS reader, too. One advantage is that this allows you to directly copy any articles of interest into your own bibliographic database. One downside is that it's impossible to export one's subscriptions. Another downside is, surprise, poor documentation. Commented May 10, 2023 at 10:02
6

As mentioned in a comment, I honestly find Google Scholar the most useful. You can search by tag, i.e., label:theoretical_nuclear_physics, search the profiles of favorite authors in a field for the latest publications, or most cited ones, and explore new research fields using labels, etc.

Other than Google Scholar, I find Research Gate also a good source of information, it's sort of like the Facebook of Academia.

Moreover, it's good to create a list of the most popular journals in your field and utilize their web pages (for example Physical Review Letters is a good source of top-notch physics research).

Finally, if you like coding, you can even create your own web scraping tool!

6

I'd like to add four possible paths that others have not mentioned yet.

First, all ethical (Musk) issues etc. aside, at least in some fields, Twitter is actually a very good way to stay up to date on the current literature. Follow the important figures in your field; many of them will regularly post about their own preprints or comment on other people's work. In some fields (probably most pronounced in machine learning), this really is where much of the field's conversations and controversies happen. During the recent partial Mastodon migration, a colleague complained that he's been seeing significantly less interesting papers since moving away from Twitter.

Second, conferences! Going to them is obviously ideal, but it can also just be a very good strategy to occasionally skim the program of high-quality conferences in your field for interesting papers.

Third, while Google scholar alerts have been mentioned (I use them), I actually find those to be less useful than simply going through the list of recent publications of people by whom I've read a great paper. Chances are high they have a few other great publications on the subject already from the last few years, but it might be quite a while until they produce new work in this area.

Fourth, there are now various customized paper recommendation systems. Typically, you can enter a few papers you consider relevant / interesting, and then they will periodically send you updates with new work you might consider useful. Some examples include SemanticScholar, ResearchRabbit and LitMaps; new competitors are popping up all the time right now.

Every kind of arxiv or journal newsletter I ever tried basically just had an awful signal/noise ratio, and I usually stopped using them quite quickly.

1
  • 2
    Yes, +1 to conferences. You should, at minimum, be cognizant and up to date on the relevant conferences of your own field. Even if you never go, check the proceedings after the fact- and if possible the programs beforehand, if posted- to stay current on the topics that are on your field’s ‘floor’, and the names that appear. Certain names will keep appearing (for a reason, some good, some bad), and new names tied somehow to a paper or talk you find compelling might be noted for future reference. Commented May 10, 2023 at 12:41
5

For arXiv, there's the subscription option to section of interest: receive regular daily listings of the abstracts of new submissions by email

  • hep-th
  • gr-qc

You create filters in your email client to pipe to a folder. You can check the folder periodically in your email.
PS: hopefully subscription isn't what you implied by arXiv 'newsletter'.

Kindly give @kamilazdybal Google Scholar's new citation approach a try.

[For Noting]

1
  • 1
    Sadly, the subscription is precisely what I meant with newsletter. I will give it another go, but personally, the plain text format is very confusing and does not help me focus, so I was seeking another way. Commented May 9, 2023 at 17:01
4

As an upgrade to arXiv, there's arXivist. It lets you select papers on arXiv that are relevant to you and then ranks papers of the day based on how likely it thinks they'll also be relevant. It's by no means perfect, but already after a few votes it does a decent job at separating the wheat from the chaff. It can also send you the top picks of every day to your email account, which I find useful for making sure I don't miss an important paper just because I didn't check arXiv that day.

1
  • 1
    Yes, in my field (aero/astro), there is ADS, which has a “similar papers” function for any given paper. In fact, “similar papers” might actually be too good, because it can easily turn into a time sink/rabbit hole if you don’t stay on track. Commented May 10, 2023 at 12:43
3

I'll keep it real with you. I'm an econometrician. I specialize in synthetic control analysis and general causal inference. Part of what I do is keep up with new developments in SCM. Every single day, I go on Google Scholar and search ' "synthetic control" "causal" "lemma" "abadie" '. I then swap lemma with Theorem.

It is not the best way of course (I can think of other ways), but it helps me follow the developments in a narrow area of econometrics literature.

3
  • 1
    What is SCM? Supply chain management? Commented May 10, 2023 at 7:38
  • 3
    SCM = synthetic control method? Commented May 10, 2023 at 7:46
  • Ha, a board of people in one Silicon Valley building can steer the course of your career… and profit from the metadata they harvest. At minimum, I’d suggest some Plan B source, to prevent possible blind spots from relying on a sole source. Commented May 10, 2023 at 12:36
3

(Specifically replying for hep-th mentioned by the author.)

I would say that the daily arxiv read habit becomes more useful after a few years in the field by which time you understand what papers are roughly about and whether they are interesting for you just from the title, abstract and author names. Once you know what the current state of the art is, it can be a good way to keep up to date on the most recent developments.

The problem with looking at these at the start of your PhD is that these papers containing the newest developments will build on other papers and will assume you are already familiar with a lot of the "most important" papers.

Therefore I would myself instead recommend reading a review on a subject you want to study in depth (probably the subject of your first research project) and read that along with the important papers cited in it.

I would also expect that your supervisor would give you a list of important papers to read and you should read those along with the papers cited in them which you will likely also will have to read in order to understand them.

After this you will likely be able to recognize, understand, and appreciate the newest papers appearing on arXiv in this specific sub field.

As your domain knowledge grows this way you can try picking up reviews on slightly different subjects and this way slowly expand the areas of high-energy physics that you understand enough so that it actually becomes useful to read new papers on the arXiv on these subjects.

(Of course a great alternative for reviews as a starting point are lectures at your university or at summer school's and such.)

"I would be surprised if there aren't any "latest cool things in this particular area of physics" with easy-to-read formatting: an Instagram post, a podcast, or even just a rich text email."

I have never really heard about such a thing. There might be popular science things that are absolutely terrible which I would advise you to avoid. For real research I don't think this exists. I think you are overestimating how big any specific sub-field in hep-th is. There will be maybe a 100 people in the world working on the things you will work on. Not exactly enough to build critical mass for a social media audience. There are some blogs that are somewhat popular but they obviously don't discuss all important papers (probably even none within your subfield). I usually hear about http://resonaances.blogspot.com/ although I think it is closer to hep-ph. There used to be a somewhat popular blog by a very controversial ex string theorist but I think it was taken down.)

1

I would recommend SciRate, which is a website where readers can give a "Scite" to papers they enjoy and it will list papers based on the number of "Scite", from high to low.

Compared with reading papers on arXiv directly, SciRate can provide you more information about how other people think of a paper. If you see a paper with more than 10 "Scite"s (which is already a big number considering the number of active users for this website), you can believe that paper has great significance and is definitely worth a look. But a paper with few or 0 "Scite" doesn't mean that it's nonsense: It might be a bit specific or deep for people to understand it very shortly, or the research topic doesn't overlap much with the most users' research field (I feel the users of SciRate have more math-based research taste in physics). Actually, I have seen many papers published on PRL didn’t get any "Scite" when they first came out on arXiv.

5
  • What is "Scite"? Some jargon used on SciRate? A sort of voting on papers on SciRate? Site scite.ai? S + cite = scientific citation? Something else? Commented May 10, 2023 at 7:33
  • Not to be confused with SciTE. Commented May 10, 2023 at 7:51
  • A citation index? scite = s + cite = smart + citation. Commented May 10, 2023 at 7:53
  • 1
    It sounds interesting. But I just tried it in my subfield within hep-th. Due to the low number of users I think it does a rather poor job of identifying key papers. The most "Scited" paper has 4 Scites. If me and one friend would sign up and Scite papers we liked we could on our own more or less pick the winners. Moreover, it suffers from the same problems that citations usually suffer from. The audience that the paper appeals to makes much more of a difference than the importance of the paper within the field.
    – Kvothe
    Commented May 10, 2023 at 12:41
  • @Kvothe You can select the timescale to be a bit longer (e.g. 1 month) to gain more information. I admit that hep-th is less "popular" on SciRate compared with quant-ph though.
    – Yunzhe
    Commented May 10, 2023 at 14:43

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .