/sci/ - Science & Math

bumping, also in similar boat

>>10671234
Yes do a bachelors in computer engineering or math. Never do CS, unless you are desperate to work a code monkey.

>>10671248
Would you mind looking at the degree I am specifically contemplating? It's called Computer science and engineering but I'm not sure if it is more like CE or CS in the US.

Again it is 5 years (standard in my country) as it is Bsc.+Msc. with the latter being chooseable (see syllabus for year 4).

https://www.kth.se/student/kurser/program/CDATE/20192/arskurs1?l=en

I'd greatly appreciate it truly!

stfu you haven't done shit in uni yet
sick of these kids thinking they're going to do xyz without having experienced anything yet.
Oh yeah you're going to do a masters in math or cs? you havent even taken undergraduate analysis or algorithm classes.
walk before you run kid.
also don't get sucked into meme fields.
"data science for financial markets'
you haven't even gone through the classes and understood what you do and don't like, you're basing your supposed career of what you've heard about and is trendy.
go to uni, find out which sub fields you like and then direct it further.
if you haven't even begun university you have no clue what you're talking about.
t. someone who works in financial markets with a stats BS

>>10671263
Most schools that have a computer science degree within the faculty of engineering, make it into more of a software engineering degree. If you are really interested in learning about machine intelligence and algoritims. I would suggest going for math or stats in your undergrad rather than a CS degree. Most of what they will teach you are just how to do this and can be learned in your own.

>>10671276
As I mentioned before Masters is standard in my country as in you already get admitted to a master's degree when applying for a bachelor's. It's basically a 5 year degree with the 2 last years being specialised. I specifically said math or CS because those are the only options in broad terms.

The specific soecializations are:
Applied and computational math
Communications systems
Computer science
Embedded systems
Machine Learning
Software Engineering and Distributed Systems
System, control and robotics

And a few more I left out as I am completely sure I don't want to study.


You are right that data science is trendy right now which is why I usually avoid using the word but it is a very close description on what I actually am interested in.

The reason why I want to study more math on the side is because I have already completed a three mathematical courses in the program so I have more time to study math more. I have already taken linear algebra, discreet math and single variable calculus so I want to take multivariable calculus and perhaps a couple courses in stats or numerical methods.

>>10671284
Honestly this was my first impression too but I don't know if I have much of a choice.

I was originally going to study Bsc. Engineering Physics while simultaneously taking a couple programming classes and then doing Msc. In CS. Engineering physics is pretty much the most mathematical degree engineering degree in my country while also including physics which I have a good aptitude for.

Unfortunately it's not looking like I will get admitted as my grades are slightly below last years admissions limit (hardest admissions in the country).

The reason why I don't want to study a Bsc. in math is because it is unjustifiably looked down upon by employers. I honestly don't know what to do except taking a degree in CS and compensating for it by choosing a more math heavy Msc.

Du kommer inte palla dubbla data med mat-stat, mat-stat är svintungt. Om du måste läsa massa matte så välj F istället, du kan inte göra upp planer på att dubbla då det är omöjligt att veta på förhand om du kommer klara det eller inte.

Man dubblar om man söker Data, upptäcker att man har A i alla ämnen och massor med fritid men såhär blir det inte för 99.99% av studenterna, du kan inte veta på förhand om du har det som krävs eller inte. D är en svintung linje, låt inte /sci/ lura i dig något annat.

>>10671314

fortsättning;

de som dubblar KTH och något annat kör oftast KTH och ekonomi, detta funkar (för vissa) eftersom ekonomi är pisslätt, mat-stat är inte pisslätt och du kommer få plugga dygnet runt för att få det där att funka.

>>10671234
If you're stupid enough to take advice from a bunch of salty math and engineering majors who don't even know what computer science is you don't deserve a computer science degree. Its your loss.

Ask yourself this, which field is currently responsible for the majority of research in data analysis, machine learning, and algorithms? Which degree do companies look for when hiring someone to one of those positions?
If you still can't figure it out and choose to blindly believe salty /sci/ fags and choose your academic field because of /sci/ fags then you where simply too stupid for a cs degree.

>>10671314
>>10671320

Tack! Jag var inte medveten om matstats svårighetsgrad vilket var därför jag var osäker på om jag skulle klara det eller inte men nu vet jag. Jag vill verkligen inte ge intrycket av att jag tror att jag helt obegrundat klarar av att dubbla matstat och D för jag vet att högskolestudier inte är jämförbara med gymnasiet.

Med det sagt borde det väl inte vara omöjligt att ta några mattekurser parallellt så som flervariabelsanalys eller vad tror du? Jag har redan läst tre mattekurser på ett annat lärosäte så arbetsbördan första året bör väl ändå vara lägre för mig även om kurserna inte går att tillräkna?

Jag vill egentligen läsa F just eftersom programmet innefattar mer matte och fysik vilket är ett ämne jag gärna lära mig mer av. Tänkte att jag kunde få in programmering och datateknik senare i utbildning så som under mastern.

Problemet är att jag inte kommer gå ut med full merit så kommer hamna på runt 21.7 istället för 22.2 i slutbetyg. Jag kommer därmed nog inte bli antagen till teknisk fysik vilket är synd.

En lösning är att jag stryker allt jag sökt till förutom teknisk fysik och öppen ingång. Kan därmed välja F efter ett års studier och därmed läsa teknisk fysik utan att fördröja studierna med ett år så som det skulle bli ifall jag tog ett sabbatsår.

Vad tycker du? Borde jag köra på open? Det som gör mig osäker är det här med klassbyte och betygshets.

>>10671324
I agree with you, I see a huge potential in computer science as a field and I do enjoy the problem-solving behind it. I just fear that I have a CS degree will make me more of a software developer rather than a CS developer. I've heard that CS is mostly math. What do you think?

>>10671263
That looks like an average to below average school in the US. Granted I can't judge the rigor of the classes and while a lot of important topics are covered I can't help but feel the degree (at least for the bachelor's part) is lacking but that might be since I'm used to a different countries education model.
I'd say its fine but I really can't judge a degree just by looking at class names. Hear opinions from people that actually attended. Not from people on /sci/ trying to justify their major of choice.

>>10671284
>Most schools that have a computer science degree within the faculty of engineering, make it into more of a software engineering degree
Terrible advice based on anecdotal evidence that means nothing in the long run.

>>10671312
>The reason why I don't want to study a Bsc. in math is because it is unjustifiably looked down upon by employers.
No, its justifiably looked down upon for a reason. In a math degree you don't learn any applicable skills. Maybe you take a couple applied math classes here and there but they won't compesate for the knowledge deficiencies you still lack. In a masters degree yes you might learn some useful applicable skills which will make you a more appealing candidate but a computer scientist will still have an edge over you fir stuff like ML and algorithms. The real world isn't at all like what /sci/ thinks it is.

>>10671234
If you think you have the ability to do any of these things, study maths instead and teach yourself computer science. I don't know about your specific course and I'm not going to look into it. I got a Bsc in maths and taught myself computer science afterward in my spare time and had no issue getting into software development. not far into it yet but the future is looking bright

>>10671375
The school is one of the best in my country but it isn't MIT exactly. I don't know anyone who's studied there and neither do I know anyone planning on studying computer science but I will take you advice on asking people (online) who have at least studied there.


Sorry for being unclear but when I'm referring to a Bsc. in math I'm specifically referring to "mathematical statistics" or "mathematics with computer science". Obviously a pure math degree is useless but I don't even think one exists in my country. The reasons why employers look down on math degrees in my country is because they don't make you an "engineer" and are taught at stockholm university which is average here.

>>10671371
>I just fear that I have a CS degree will make me more of a software developer rather than a CS developer.
CS is a giant field. If for your masters you choose to go the software developer route that's where you'll be but CS at its core is math. By CS developer I assume you mean researcher and yes that's possible based on your masters options. I mean do you really think a math degree would be more useful for a CS researcher instead of a CS degree? That logic makes no sense.

>I've heard that CS is mostly math.
Again it depends what you study. If you focus on embedded systems for your 4th year you won't learn much math.
Based on your degree the most mathematical MS programmes are computer science, Machine Learning, Applied Computational Mathematics, and maybe Information and Network engineering. They all should teach a good deal of math.
But still ask around to get a better sense of your options.

>>10671403
The classes look fine it just feels lacking since in the US you often have more requirements like chemistry and physics and it just feels weird looking at a degree like that. Its a personal bias.

Okay mathematical statistics and math +CS would be pretty useful for data analysis, more so than a CS degree but as for the other two (ML and algorithms it depends). For ML it should be useful enough albeit lacking granted you can pick up what you lack. It really depends what they teach you but a CS degree would have an advantage albeit not an advantage you cant overcome. In some ways stats would even be seen as preferable, it depends. The hardest part is convincing employers you overcame it. For algorithm it just wouldn't be as useful as CS. The bias still makes sense since usually stats/math guys are good with math but have difficulty implementing it.

Just do what makes you happy. The math options aren't bad you just have to do more depending on what field you go to.

>>10671371
I did a bachelors in CS at Chalmers (tror det är CS iaf, datateknik) and yes, that will make you mainly a software developer.

>>10671443
Why would you need chemistry and physics for a CS-degree? Sounds weird to me.

>>10671422
>By CS developer I assume you mean researcher

Exactly!

> I mean do you really think a math degree would be more useful for a CS researcher instead of a CS degree? That logic makes no sense.

This is comforting to hear. I initially thought that a CS degree would be more suitable but I've become more unsure as I've seen people saying that most computer scientists have a background in math while CS degrees only lead to code monkeying. Simultaneously the CS degree at KTH doesn't include multivariable calculus while engineering physics (the other degree I'm contemplating on) is more math heavy while having the same options for Msc. but at the cost that it teaches physics instead of programming during the first years.

I greatly appreciate your help especially because you read the syllabus and while I am not completely sure on if I should study engineering physics or computer science (master's options are the same) I at least now know that CS research is possible with a Bsc. In CS.

Why CS>EngPhy is not completely obvious to me is because, as previously mentioned, EngPhy has more math. Furthermore I have a good aptitude for Physics and enjoy the subject so I'd gratefully study it for the pleasure.

Thank you again, truly thankful. I've lately been extremely indecisive.

>>10671443
>it just feels lacking since in the US you often have more requirements like chemistry and physics.
Your probably right as in Europe a Bsc. Is only 3 years but high school at least in Sweden is more difficult I think to compensate for it. Additionally a Msc. is pretty much standard here while I the majority in the US don't take a Msc. if I'm not mistaken.

Thank you for help! I am very thankful for it and will reread all replies I've gotten in this thread at a later time.


>>10671463
Datateknik at KTH is informationsteknik att Chalmers. I am not sure why it is this way but it is. Does informationsteknik make you a software developer at chalmers?

>>10671471
>Does informationsteknik make you a software developer at chalmers

I'm not sure actually, my very shallow understanding of that program is that it's focused on things like cellphones, internet-communication and that many women choose it for some reason (at least more than data).

>>10671293
you said it is standard however not required. Again the fundamental point is there isn't any point theorising like this so early. unless have to choose.
also the amount of math you have taken is very small and pale in comparison to what awaits. relating to my point is you haven't done proof based mathematics's or anything beyond beginner uni Math. it is very hard to say whether you would actually enjoy the content of higher math this early, the emphasis greatly changes from computation to proof. (computation and as actually doing things, not related to cs). i was like you and tried to plan things out but unless you have to don't shoehorn yourself early, try things out and see what you enjoy and are good at.
also it doesn't matter too much your selection all this shit is similar, you'll either do cs and take math electives or vice versa.
regardless of which degree you do you should cover.
data structure and algorithms + maybe algorithm analysis class
AT LEAST one class about statistics and application with R or python.
numerical methods.
multi variable calculus.

additionally would recommend taking intro to real analysis, optimisation, a ML meme course depending on the quality at your uni. more stats always is good, DE's are helpful, a database computing class using SQL, applied probability shit using stochastic processes and etc etc there is always more shit

ps if you really want advice on a specific one applied and computational maths with some stats classes sounds good, or machine learning if you just want it for the buzzword factor.

>>10671234
I can only comment on data science (MSc in CS with the focus on natural language processing). First of all, if you plan on focusing on DS/ML, I would definitely suggest going for Master's as the standard CS Bachelor's curriculum will serve as the introduction to the field, but at the same time doesn't go in-depth enough to give you the skills required to actually work in production.
Secondly, the amount of math needed depends on your goals. The essential subjects like calculus 1, linear algebra, and probability theory 1&2 are a must and it's completely possible to work in the field of data science having no further mathematical education (although you wouldn't be working on any cutting edge stuff). If you also want to appreciate the mathematical proofs of the algorithms, optimization methods and calculus 2&3 should also be studied. If your goal is to rise above the junior developer position and actually derive the new results, stuff like numeric (optimization) methods, Monte-Carlo methods, Bayesian statistics, and functional analysis 1 would be extremely beneficial. The last rung of the ladder is truly advanced treatments or applications/doing a PhD in DS/ML. For that it's all of the above plus stuff like measure theory, calculus of variations, functional analysis 2, statistical learning theory, robotics, stochastic optimization to name a few. As it can be seen, the amount required to get by largely depends on your own goals. However, if I had to give advice I would suggest minoring in statistics as opposed to mathematics (as is usually suggested) as it can be seen from the above that basically all the math required for DS/ML is actually statistics classes and it won't bore you down in unnecessary stuff. Having taken stuff like real analysis and discrete mathematics/graph theory, analytical geometry, there have been basically no scenarios over the years where any of that stuff became useful and everything I use day-to-day is either CS or statistics.

>>10671234
Yes, it's a meme, if you want to do theorical CS you do math, if you want to do systems you do ECE.
>data analysis, machine learning
That's just fancy stats, knowing measure theory will give you an edge over CS majors who might not even do mathematical stats.
>algorithms
CS majors don't do these in depth till grad school, a math BS can easily learn undergrad algorithms in a few days.
>Honestly I find software engineering such as creating operating systems and UI's very boring. Programming in or itself is not attractive only the computer science theories and applications that surround it

it depends on the school really. if you go to a top school like stanford, CS is definitely worth it.

CS programs at most state schools are garbage though so don't waste your time and money with that.

>>10671371
We speak English here you retarded foreigner.

ITT retarded mathfags who think "coding" is easy shit that you learn in a day. Ever wonder why physicists and mathfags have a reputation for atrocious code (phds especially)?

If you want to do software, do software. Take the CS curriculum. If your at a shitty state school all the majors will be shitty regardless. If your school has bad AI classes, learn things yourself online. CS is lucky in that it has the most free resources online for learning things on your own compared to any other major due to open source culture.

>>10671248
>bachelors in math

>>10673433
it has been years since the last time i saw that face. thanks for being gay anon

>>10673289
>ITT retarded mathfags who think "coding" is easy shit that you learn in a day. Ever wonder why physicists and mathfags have a reputation for atrocious code (phds especially)?

I don't get this either. There's this sentiment going around like math/physics students would have extremely easy time mastering the entire CS curriculum *if they really wanted to*. In my university, math majors also had to take the A&DS class with CS students and while I do agree that math majors had easier time with the theory part of the subject, most of them struggled hard with the actual implementation where, on the contrary, CS students excelled. It seems like some people, whose experience with the field is superficial at best, think that doing CS pretty much equals to watching Udemy web development tutorials for a couple of months, when in reality it's a really small part of the curriculum. In my own experience, physics/math majors who have tried out software engineering are usually pretty mediocre as they can technically write code, but cannot write code well. This can directly be attributed to the lack of CS classes (testing, system design, clean code practices, classes that cover creating an entire project, etc) or in even worse cases to the honest belief that it's already good enough if the code simply works and disregarding how it fits into the bigger system.

>>10673827
>>10673289
The opposite of Op's question:
I'm a CS undergrad don't get me wrong I like math but honestly don't want to work in research areas and prefer to become a software engineer. How to learn to write better code/ be better at building software?

It depends on your school. I won’t name where I went to undergrad (though from the description, you’ll probably be able to guess it), but it was highly ranked for both math and theoretical CS. CE people will tell you yada yada because they believe that knowing semiconductor physics + codemonkeying somehow illuminates some superiority. If your school has a proper CS program with the proper math and systems (they collaborate / come out of the appropriate academic traditions; in our case the theory department was started by mathematicians and the systems by EE and physic departments back in the day), then there’s no need to go to CE, since CE requirements generally ask for data structures and basic algorithms, with a side of some systems classes, but not much more. So basically, they’re telling you to do an engineering degree + codemonkeying.

I did a double major in math and CS. I went heavy on the analysis in the math department, alongside the regular suite of year long algebra sequence, topology, geometry, combinatorics, and of course the basic requirements like differentials, etc. On the CS side, I took up to an OS design class and an embedded class for systems (past the required architecture class and systems class), and then I focused all on theory, so I did 2 higher algorithms classes after the required one, computablity and complexity courses (the latter being in the grad department), topics in theoretical CS (which was basically a suite of really hard problems / proofs from nontrivial results), and more. I never had trouble landing good, non codemonkey internships. In fact, the two I took were at research labs for future technology alongside other math, physics, CS, and engineering students. I ultimately ended up doing grad school in math kinda working jointly with a CS theory group, bridging the gap between continuous phenomena and computation, but also writing in mathematical physics. It’s really exciting, truth be told.

>>10674596
Basically, the tl;dr from this blog post is that dont stress about labels too much. I acknowledge there are many codemonkey schools out there, but not all are. If you’re in one, go to math for research or engineering for industry. Otherwise, go to town in the CS program and take all the tryhard classes with all the math students (many of which who double major anyway). Double majoring only gave me a richer degree. Don’t listen to 4chan much because it’s mostly undergrads or grads looking for validation via shitting.

>>10674596
>>10674606
How'd you manage to pull off a double major?

>>10674630
Lots of time in the library, easy online classes for most of my general requirement classes, and taking a summer during research while I could.
Pro tip; talking to a professor for long term research that yields nontrivial results oftentimes looks way better than a 9 week REU. If you do undergrad at an R1 school and want to do grad school, you have your work cut out for you.

>>10674651
*a bunch of summer classes
It was a lot of work, but I made it so I only took 4 classes at a time for most semester, only taking 5 for two semesters. This let me max out on grades and focus on research during the year

>>10674155
To write cleaner code, I would suggest the books of Robert C. Martin such as "Clean Code: A Handbook of Agile Software Craftsmanship" (he has also written a bunch of others in related areas). As for building better software I would say there isn't really a silver bullet and it mostly comes with experience. Internship/working is obviously the best way to learn, but I would also suggest hackathons as they force you to get a working product up and running in a short span of time (as opposed to personal projects where you can fuck around for three months and still have nothing to show for it).

>>10674155
>>10674788
Books may help learn some common patterns, but won't really help you code better.

To make cleaner/better software it just takes time and experience. I'd recommend looking through open source projects on github from companies like Google, Facebook, Microsoft, etc. It won't always be the world's "cleanest" code but it'll show you how people actually manage and architect complicated software with possibly 100s of engineers in practice.

I personally wouldn't recommend hackathons as good practice for clean code. You learn how to shit together something in a weekend that 90% time you will never touch again. There's no incentive in hackathon to write good maintainability/extendable code. Hackathons are just for fun/social/networking aspects.

I recommend having a longer term (several year) side project to work on because that forces you to think about maintainability, how to migrate to new libraries, and suffer multiple refactoring.

>>10671234
>Honestly I find software engineering such as creating operating systems
You don`t even see this anymore. People don`t want to make new OS, they want to make clones of UNIX. Fuck this gay earth.

>>10671467
It adds a level of rigor and the more you know the better. I took 3 physics classes for my CS degree. I feel better off for taking those classes.
Plus physics can be very useful when taking embedded systems

>>10671324
I think it more has to do with the number of retards who do compsci to make bideojames.

>>10671234
I've been accepted for compi sci, ele eng, and mech e at the same school. Have a few days to decide. Fuck.

>>10675173
What school?

>>10674596
Thank you for your suggestion but I find it difficult diciphering between an adequate CS degree and one that is not. I assume that more math classes are indicative of a more theoretical but I can be completely wrong.

Keep in mind that in my country you get admitted to a Msc. simultaneously as the Bsc. but you get to choose the specific area for the former later on. What I miss in Bsc. I can through the Msc. Compensate for it to some extent.

Anyways, the specific degree I am considering has the following math courses:
Linear algebra and geometry
Discreet math
Single variable calculus
Modelling and simulations
Numerical methods
Probability and statistics

I know you don't know the specific courses but what is your first impression? I have already read 3 of these courses during high school so I will most likely also go with multivariable calculus and some relevant stat courses (am mostly interested in data analysis).

I would be extremely grateful if you could give your opinions on another degree because honestly if I can go with this one I want to but not many people outside of my country have even heard of it let alone respect it.

Engineering Physics (similar to engineering science):
+High prestige as admissions are the hardest and the degree itself has the most difficult coursework
+A lot of physics which I enjoy and which my aptitude is strongest in. I wish to study Physics simply for the aesthetic nature and pleasure of the subject even if it's not useful.
+Significantly more math and more proofs than CS degree above.

The major con is that it does not include a lot of programming at all. Nonetheless it includes 16 American credits and if I go with this degree I will most likely go with Msc (2 years) in CS, ML or something similar to compensate for lack of programming.

Again I mostly want to work with development of theoretical CS and data analysis and NOT creating games and practical coding.

What do you think?

>>10674596
If I was unclear both degrees will take five years

It is either:
3 year CS + 2 year relevant Msc. (Such as applied and computational mathematics)

Or

3 year EngPhy + 2 year Msc. in CS, ML or similar.

If you have heard of Engineering Science at the University of Toronto it is very similar in difficulty from what I've heard.

>>10671234
Ummm ridiculous honestly, programming is staggeringly beautiful!! It's a gorgeous analogy to science and epistemology in the ways it's tested, the inference of this underlying sturctures that are not only responsible for the language in which one programs but on the shoulders of the gians on which one labours! So that every part of it, is a language written on top of another language and expressed elegantly. The multiplicity of ways to express things is astounding, but people love to compartmentalize and to reduce to the most standard ways even though the room for the creativity of one's representations is not bounded and one can do art with the way one solves problems by choosing not to adhere to some rigorous constraints but to choose your own constraints in whatever ways you want to explore the wonderful world of the wizardry of manipulating the spirits that inhabit our computers! :D

>>10675695
I might have a wrong perception but I see programming as almost equal to language and grammar. With language you can write beautiful calligraphy or an amazing book but it's still not for me. Especially if the contents of the book is already decided by someone else.

>>10675221
Are these the only requirements? Linear algebra, basics of discrete math, and calculus (at least 2-3 semesters) are..well, the basics. At my school, you needed to take 2 semesters of calculus of a single variable, 1 sem of multivarible calculus, intro to CS, data structures, linear algebra, and 1 semester of a year long basic discrete mathematics sequence for admission into the major. I would hope that the theory track has more mathematics.
>Modelling and simulations
not really a math class, more about implementing numerical models and learning your way around various environments
>numerical methods
not really a math class, more like a bag of methods to throw at problems without analytic solutions
>probability and statistics
is this one class? At my school, we had probability theory, which covers everything from combinatorial probability, to derivation and use of distributions (continuous and discrete), hard expectation problems, moment generating / characteristic functions, etc., and then a separate mathematical theory of statistics class. I only took the former since statistics wasn't super interesting to me (though probability is interesting due to measure).

My first impression is that this is workable for a standard experience going into software devving, but it's not really the proper toolkit you'd need to be a theorist / researcher. It seems that engineering physics is a good program in your country and leads to many opportunities. Keep in mind anon, I have a fondness for physics, but if your goal is to do research in CS, you'd best go for a math degree, but it totally *can* work out if you go for that engineering physics degree. Programming is a skill/discipline you usually gain yourself after hours of writing for yourself, but I guess I'm biased since I feel I learned the most on that front by writing my own kernel level code and rudimentary compilers (that ran in shitty time, no less lol).

>>10671234
>Is a Bsc. CS a meme
Lrn2meme fgt pls

>>10675237
I heavily recommend staying away from ML as a theorist, unless you want to do ML theory. Results are heavily exaggerated as we just attained cheap computation power to test out models from the 60s-80s, but they're in need of heavy revision by theorists. Either way, it's not uncalled for to switch fields or transition to adjacent fields in STEM, since after all, everything's connected, and the best research is done on the borders between fields. Do whatever you feel will give you a strong basis in pure math and problem solving. Prestige is fleeting, but good work lasts forever. Challenge yourself and do whatever you think will make you grow most as a theorist and problem solver. Oh and whatever you do, read Lovasz's Combinatorial Problems and Exercises. The problems in that book are all big brain and will stretch your abilities to think outside the box

>>10672955
>CS programs at most state schools are garbage though
This has been repeated time and time again, and time and time again people have failed to provide proof.