Search results

preconditions are easiest and most common.. assert(... insert clever type traits...) object invariant ...trickier? postcondition, like PREC and also OK?

You could benchmark/reverse engineer Boost C++ Contract source code haha. Not Rocket Science.

Tests // Normal use cases CheckValve v1(false); IDevice<CheckValve> dev(v1); // Default not open dev.Open(); // dev.Open(); dev.Close(); dev.Open(); dev.Close(); CheckValve v2(true); // hit ground running IDevice<CheckValve>...

here's a test case for Rust // InterfaceContracts.hpp // // Associating contracts with C++ Concepts. // // (C) Datasim Education BV 2018-2025 // #include <boost/contract.hpp> template<typename Device> concept IOpened = requires (Device d) { d.Opened(); }; template<typename Device>...

You'll need a bigger avatar :-)

Design by Contract // Test101.cpp // DJD 2025-10-6 // // "Design by Contract" (based on Hoare logic) was originally supported in Eiffel. // Attempts in .NET did not reach fruition it seems. You can insert code for Contracts // in C# but the compiler just IGNORES it. // C++ does not support...

All the rock stars (Paul Weller, Rory Gallagher, George Best etc.) stayed there. Paradiso and Milky Way rock temples nearby. https://magazine.thalesians.com/2025/01/21/an-interview-with-daniel-j-duffy-the-origins-of-c-in-quantitative-finance/

Cool! A relevant piece of information: the first C++ course in Nederland was given in the Hotel American across from my office in the Leidsekade in 1990.

Thank you :) More info on C++, PDE, MC https://www.datasim.nl/books

'entirely' ? A good idea could be to learn stuff by taking Hull's book as the 'critical path'. Of course, there are others. // Those are my principles, and if you don't like them... well, I have others. Groucho Marx

Get John Hull's book and program the models therein in Python and C++. A clear project. A prominent "learn by doing" quote comes from Aristotle, who said, "For the things we have to learn before we can do them, we learn by doing them". Other well-known versions include Richard Branson's "You...

And maybe critical questions based on versions C++98/C++03 C++11 // C++14 C++17 C++20 just an idea

Python won't help much as a stepping stone to C++. I would learn C++ and not couple it with numerics. Single Responsibility Principle (SRP). Numerics is independent of the language; learn the maths first. https://www.datasim.nl/onlinecourses

C++ Numerical maths

1. (First Encounters with Smart Pointers: Unique Pointers) Consider the following code that uses raw pointers: { // Block with raw pointer lifecycle double* d = new double (1.0); Point* pt = new Point(1.0, 2.0); // Two-d Point class // Dereference and call...

Even higher level We now wish to implement Rectangle and Square in an object-oriented language. The options are: Rectangle is a subclass of Square . Square is a subclass of Rectangle. Rectangle delegates to Square. Square delegates to Rectangle. Design Rectangle and Square as classes...

Survival pack 1. Review C++ Syntax: “Survival Syntax’’ We give a list of language features in C++ that we introduced in section 2.7. The goal of this exercise is to investigate what they are and what they mean. In particular, you should know the following: . Header and code files; what are...

I did not say "numerical analyst" (like me) but more "quant meets numerics", which is slightly different.

2025 someone at 2018 level requests a 1970-style answer from ChatGPT.

The best way to learn C++ is to program an application until it works. Then those cute multiple choices fade into the background.

Cargo cult programming https://en.wikipedia.org/wiki/Cargo_cult_programming The term cargo cult programmer may apply when anyone inexperienced with the problem at hand copies some program code from one place to another with little understanding of how it works or whether it is required.

See this .. all in there https://www.datasim.nl/books

A layered approach.

So many?

I don't (need to) program for a living in the strict sense but jumping into Rust (which is almost like C++) would have little value added for me. C++ is more powerful, anyways. Maybe interfacing Rust and C++ is enough?

Questions to challenge the developer's intelligence beyond the somewhat 1d syntax questions.

AFAIK a lot of quant work is numerical (most problems don't have closed solutions).

I've been doing PDE/FEM/FDM for 50 years. Never did manifolds. https://www.datasim.nl/books

Unfortunately, algebraic topology is not used a lot in finance. Might need to learn additional "kinds of maths" skills. C++ and Python. @Valc

What kind of "pure" mathematics?

Welcome, I'm from Dundalk :) // UCD, DCU, TCD?

no hurry; you have your whole future ahead of you :-)

Nice weekend QUIZ

We give a short overview of several programming languages that support the object-oriented paradigm. This section also describes the personal journey of Daniel J. Duffy in OO land since 1989. An early example of what would be considered an object-oriented programming language is Sketchpad...

You're welcome. Back then, most developers got OOP wrong except myself of course (a long story, but basically I trained people + wrote production software..) but OOP is applied concept/semantic modelling. I used Eysenck, M. W. and Keane, M. T. (2000) Cognitive Psychology. Psychology Press...

My 2 cents. I am sure my online course AMP is hands-on, rigorous and unique (I did it numerical analysis in academia and industry since 1974). Your initial list could include more of the "hard" maths IMO. It is future-proof.

I agree; DL is not fundamental knowledge; will DL be around in 5 years time? "DL, (numeric) maths inside" I have two numerics online offerings.

ODE/PDE Numerical Analysis, e.g. Numerical Linear Algebra. QN C++/Python https://www.datasim.nl/onlinecourses

I am the originator of the QN C++ courses Some background on how it evolved https://magazine.thalesians.com/2025/01/21/an-interview-with-daniel-j-duffy-the-origins-of-c-in-quantitative-finance/ // I was the first C++ developer + company in the Netherlands.

Author: Daniel J. Duffy, Datasim Education BV Publisher: Datasim Press Multiparadigm Software Design in C++, Python and C# delivers a repeatable and step-by-step approach to software design. We address the software lifecycle from problem description through to design and implementation in...

Thanks, Andy @Andy Nguyen

Andy @Andy Nguyen Would it be possible to change the title to New Book: Multiparadigm Software in C++, Python and C# by Daniel J. Duffy (Datasim Press) @Andy Nguyen

I have written and copy edited everything. So, all chapters have been written by me. We are now entering the final phase, i.e. printing.

You're welcome. C++ is the gold standard in finance and QN C++ as well (warning: I am the originator :)) @APalley @Paul Lopez Most of my books discuss finance/numerics and C++ https://www.datasim.nl/books Remark: my ODE/PDE online course for finance is on UCB list...

Maybe more numerical analysis. applied maths ODE/SDE/PDE C++ // Topological Data Analysis(TDA), a new technique which uses algebraic topology to analyse data (basically extremely fancy feature extraction). Not sure if this is used a lot. Very esoteric..

This uniqie ODE/PDE course for finance is on the UCB recommended list. https://www.datasim.nl/onlinecourses/97/distance-learning-ordinary-and-partial-differential-equations https://www.datasim.nl/onlinecourses BTW I am the orignator of both Quantnet C++ courses.

.

School's out for summer

This course indeed has wide applicability to many applications..

No quant knewledge is needed with the possible exception of Monte Carlo at the end. But it would be easy to choose a PDE/FDM/FVM case instead and the same design (my domain architectures) for it. I wrote 3 books on PDE/FDM (my own background) and all schemes in C++. This course goes all the way...

This robust ODE/PDE course is on the UCB recommended list. The materiial is specifically geared up to computational finance. https://www.datasim.nl/onlinecourses/97/distance-learning-ordinary-and-partial-differential-equations

https://www.datasim.nl/onlinecourses

.

@Ken Abbott

Some one-liner remarks on how to learn ODE/PDE and apply it. In general, we can get students with Calc3 knowledge to 1 and 2 factor PDE in finance (Fokker Planck, Black Scholes, Hull White, CIR, XVA etc.) in about 2 or 3 months. Some objective criteria to measure the quality of student...

The list feels a bit pure mathsy, needs more applied + numerical. Add Numerical Analysis (Numerical {Linear Algebra, Differentiation, Integration, Optimisation, Interpolation etc..) i.e. computational focus. And coding skills ;)

Write option solvers, hedging etc, using Pytthon odeint, Ad etc.

Actually, Mike you did a whole lot of work in those courses! I was very pleased with the effort. Normally, doing a critical subset of a course is enough for a certificate. e.g ODE/PDE has parts A-H for UCB, parts A,B,F,G for the certificate.

ABSOLUTELY!

"We already use C++26’s std::execution in production for an entire asset class, and as the foundation of our new messaging infrastructure," OK https://en.cppreference.com/w/cpp/execution.html More details? Threre is always a time delay between when software prototypes and production versions...

If C++26 is the answer, what is the question? Sounds like marketing/future ware? I need convincing. What do I know? haha Might be a bit harsh: the link sounds also like an advertorial..

I am sure Numericals are a core step in the computational finance pipeline. Problem (e.g. Black Scholes, HW, XVA,...} -> {PDE, ODE, SDE} -> Numerics -> Software Design -> {C++, Python, C#}.

I am pretty sure there are overlaps. I reckon my courses are more detailed and industry-focused. I've been around longer. 😁

Maybe Applied Numerics instead of Number Theory, the latter being a bit too far from quant work. my 2 cents.

📢 Fast matrix computation is not rocket science any more. What's the meat?

Hard to say; how much real OOP in the course? It looks very STL focused. But I can only see "bullet" sections.. "the" means QN Adv C++ course?

What's your current C++ level?

One issue worth noting is that there are two main strands associated with matrices 1. Pure maths linear analysis/algebra ... paper models. Kind of academic-only focus. 2. Numerical linear algebra .. algorithms and getting stuff into the computer. And knowing what, why and how. numpy/scipy...

That syllabus could be more ambitious. imho No one really does determinants any more .. the advent of the digital computer put them out of business.

https://www.datasim.nl/application/files/5915/4999/4170/DL_Applied_Numerical_Methods_II.pdf https://www.datasim.nl/onlinecourses/101/distance-learning-applied-numerical-methods

Not sure about VBA, it is easy to learn after C++/Python. Check if it is really used. Some MFEs need ODE/PDE.

This ODE/PDE course is on the UCB recommendation list https://www.datasim.nl/onlinecourses/97/distance-learning-ordinary-and-partial-differential-equations

Random thoughts 1. Is prompt engineering clutching at straws? A cop-out for robust requirements analysis? 2. NLP is fraught with pot-holes. It won't work in general. Wittgenstein would have a lot to say. 3.Will ChatGPT ever be a requirements analyst? Probably not, because it is a name dropper...

Crepi il lupo! @cooper

Ideally, do ODE and PDE https://www.datasim.nl/onlinecourses/97/distance-learning-ordinary-and-partial-differential-equations

One of the authors is Igor Halperin, a famous quant.

In the pantheon of digital technologies that have reshaped human behavior, artificial intelligence stands out as perhaps the most insidious threat to our cognitive independence. This essay examines how AI functions as a digital drug, creating dependencies that systematically erode fundamental...

You're welcome :-) I was half kidding about edlin. Seriously, do and check each step and you can't go wrong. Like maths. Good luck!

Just write code, line for line, and get it right first time. Imagine only having a line editor :-) https://en.wikipedia.org/wiki/Edlin

For Physics, is Fortran not a standard? It's a good way to get started. Quant work is very mathsy/numerical, so Fortran is ideal. The sooner you learn to program, the better. At some stage, C++ is important.

What does Calculus 1A entail. i.e. topics? I reckon most MFE students are (need to be) at the level of parts B,C,D here https://www.datasim.nl/application/files/1315/5947/7165/DL_Pure_Mathematics_Foundations.pdf

I am not a university, and I offer rigorois maths courses A-Z. https://www.datasim.nl/onlinecourses https://www.datasim.nl/onlinecourses/103/distance-learning-pure-mathematics-foundations

C++ already has functionality for parallel and asynchronous programming. The QN Advanced C++ course discusses it. The issues going forward. 1. Meeting the deadline and how useful. 2. Learning all this stuff. "Thinking parallel". 3. Applying in real life. Multi-threading is for "experts"; 4...

How long did that take you?

Maybe something like storytelling using the Snorri Sturluson approach?

https://www.maths.tcd.ie/pub/ims/bull75/Duffy.pdf A 101 example, from begining to end.

In fairness, this looks like traditional 2nd year undergrad ODE theory (not a bad thiing). It does not help with understanding the DEs you meet in finance. The following does ODE/PDE and applications to finance. It is on the UCB prerequisite list. (student price)...

Have fun with C++ :)

They are all wonderful, of course! :) But seriously, I would say that numerical methods are fundamental in the short term (BTW I trained as a numerical analyst) // Generally, one focus is to develop reusable and repeatable skills and not just doing stuff from memory...

Yes.

For MFE Parts A,B,F and G are the core. The others are also useful. Parts A,B,F and G: exercises to be completed for certificate :-) Normally 2-3 months is very doable. The style is by audio/video + all docs etc. + copy of my modern 2022 PDE/FDM book for finance. + lots more. And the QN C++...

ODE/PDE course still open :)

What's that in plain English?

This agenda looks pretty good. But 4 years is needed. It's worth it because it's essentially a golden opportunity.

A maths degree in 2 years? Depends on how deep the level..

yeah, the silence is deafening .. It is crystal-ball, hype, marketing. Give it 15 years. In short, again, hype. Stick to the knitting.

Welcome back. Not many thoughts, TBH. It's futureware..

Thanks, Colin. Everything fine here!

Courses are always open.

Well, they should accept my ODE/PDE :-) because it is the only that does a A-Z and does PDE in finance. And based on > 40 years research. I have not made steps in an accredited certification process (how can I go about it?) If you like, you can DM me.

Some background on these courses etc. https://magazine.thalesians.com/2025/01/21/an-interview-with-daniel-j-duffy-the-origins-of-c-in-quantitative-finance/ https://www.datasim.nl/testimonials

Regarding languages, Python and C++ are vital in general IMO. VBA Is useful but limited to Excel. And PDEs are important AFAIR.

Hi Colin, I worked with Mikael and he is/was working expert in this area. It might be useful. https://github.com/MikeJuniperhill https://onlinelibrary.wiley.com/doi/epdf/10.1002/wilm.10708 He has some more related ones out there, and he is active on QL etc. regards Daniel

Do you have an example of what you mean?

The seeds of this course were sown in 1990 and evolved over many years in real life. To date, I have written several books. And this course is focused on finance. https://www.datasim.nl/books A short history is here...

Statistics is an _application_ of mathematics to a certain domain.

What's "Calculus" precisely?

Learning statistics after applied maths >> than the other way around.

C++ is a skill for life.

https://cython.org/

Cython == Bridging Python and C/C++. Best of both worlds.

https://www.datasim.nl/blogs/29/msc-theses-on-machine-learning-and-computational-finance-2020 See the work of Dr. Jim Gatheral as well.

Python and SQL are extremely useful. To learn programming, another language is needed, e,g, C/C++, C#. Depending on LLM is not healthy.

CMake for C++

CMake

These days C++ has support for dates and time, https://en.cppreference.com/w/cpp/chrono

If you send me an email dduffy@datasim.nl btw I wrote that book ;) please send the code + tests!

good stuff. Uchi mata is a big one. I used to do a lot of sutemi but lately sasae tsurikomi ashi is easier :-) Now doing kime no kata to keep fit.

Measure Theory with a focus on probability. This is a real heavy course .. deep knowlegde of Real Analysis needed. MT is not as useful as it is made out to be (moi, 3 years of it at undergrad level). https://www.datasim.nl/application/files/1315/5947/7165/DL_Pure_Mathematics_Foundations.pdf

These courses seem to missing crucial components that are essential going forward, e.g. https://www.datasim.nl/onlinecourses pure versus applied? it's an artificial separation. . ode/pde/fdm . deeper numerics (beyond linear algebra) . C++ . Statistics? can be learned 'on the job'.

Cool. The course has two modules on Monte Carlo and PDE/FDM option pricing numerics/code which might be useful.

Random quote This is a textbook example of "rewrite-itis" disease. Ubuntu folks decided to rewrite the core Linux tools that have been working great for decades, in Rust. Just because they like their shiny (or, more like rusty) new programming language. Rust also started infecting the Linux...

wow! Super 📢

Sounds like a good idea, TBH I haven't studied MCMC. What about MLMC? Maybe MCMC is similar. https://en.wikipedia.org/wiki/Multilevel_Monte_Carlo_method I would say that doing a modular approach in Python pays dividends going forward in your career. A remark: the MC architecture is the test...

One idea: Monte Carlo option pricing in Python based on solutions in C++, C#. https://onlinelibrary.wiley.com/doi/epdf/10.1002/wilm.10647

AI Agents!! @Paul Lopez // we did some basic stuff on agents 20 year ago ... not easy.

I don't use/need it.. What about this new stuff!!

Nothing wrong with SW engineering. Performance can be a requirement for many applications. In a sense, the maths is not the most critical part(?) C++ is still vital I reckon.

DeepSeek's libraries https://www.deepep.org/

But maybe parallel MC ? Also done in Advanced QN C++. @MikeLawrence

Anti-pattern and bad practices in real-life software projects We present the following seven patterns: BIG BALL OF MUD THROWAWAY CODE PIECEMEAL GROWTH KEEP IT WORKING SHEARING LAYERS SWEEPING IT UNDER THE RUG RECONSTRUCTION The real problem with THROWAWAY CODE comes when it isn't thrown away.

https://www.datasim.nl/onlinecourses/77/distance-learning-advanced-c-for-computational-finance-and-derivatives-pricing

.

C# is much better than Java for quant stuff (Java is good for other things). C++ and C# are close. Many shops use C# because its many advantages. C# for Financial Markets Daniel J. Duffy and Andrea Germani C# is a modern object-oriented programming language that runs under the Microsoft .NET...

In the QN C++ course we use Euler scheme for Black Scholes. It's easy but 1st order. Now, ADE is just as easy to program. Could be an exercise next version? Let me know. https://papers.ssrn.com/sol3/papers.cfm?abstract_id=1552926 @APalley @Paul Lopez

My 2022 book on PDE/FDM for finance is useful in this context And I have C++ code for all schemes. https://www.linkedin.com/pulse/upcoming-book-pde-methods-finance-planned-date-december-daniel-duffy/?trackingId=sQvR8%2Bg6RsKMnBIebUcztA%3D%3D

Fantastic! Actually, there are 2 modules on Monte Carlo and FDM 101 to do in C++. After that you are ready for Adv FDM and I can coach you how to set up the fancy schemes and program them in "C++" if you like idea.

I am trying to help. But I need to know more. DM/PM me is an idea. I have a repeatable process/best practice. See QN C++ Testimonials.

https://onlinelibrary.wiley.com/doi/epdf/10.1002/wilm.10620

Your requirements seem to be changing.. FDM has nothing to do with C++. I wrote a book in 2022 on FDM. Here is a link to some C++, PDE, FDM. https://quantnet.com/threads/blog-articles-on-c-11-and-computational-finance-by-daniel-j-duffy.32237/

I am planning to study some C++ and then jump directly to DataSim C++ course. What's your plan, precisely? BTW you need to get to QN C++ (or equivalent) level to do Datasim C++. Otherwise coaching you would be sub-optimal for both of us. Studying "some" C+ sounds a bit a la carte TBH...

Datasim C++ assumes C++ knowledge. I recommend QN C++, it is appropriate in your case. And great support for those w/o C/C++ knowledge. FDM is fine but ODE./PDE is an essential precondition (essentially parts A,B,F,G)...

Language-independent Software Design based on reusable domains and design blueprints example: Monte Carlo

Wise words for EVERY programmer

Some recent MSc theses on ANNs might be useful. Important not to get too carried away with the hype. https://www.datasim.nl/blogs/26/msc-theses-on-machine-learning-and-computational-finance-2019 https://www.datasim.nl/blogs/29/msc-theses-on-machine-learning-and-computational-finance-2020

We can avoid virtual destructor in a numbe of ways by considering other options: 1. Use smart C++11 pointers. 2. Parametric polymorphism (templates) instead of subtype polymorphism (inheritance). 3. Postpone physcal memory allocation for as long as possible (e.g. in main() using dedicated...

This 3-month work was written in 2014 and rightly focused on GPUs which were hype at the time. It is now 2025, so many technical and organisational requirements have evolved and changed: 1. Code is not portable ("captive user"). Porting to CPUs => rewrite. 2. Traditional OOP used; parallel...

Automation COM Adds were something completely different, especially ATL and CRTP.

Ask Irina if she plans to take QN C++ course.

Beginning of the end? A bit quite in Rustland.

Fundamentals of Structured Programming Also applicable to C++ https://en.wikipedia.org/wiki/Jackson_structured_programming

Maths is essential.

An Interview with Daniel J. Duffy: The Origins of C++ in Quantitative Finance https://magazine.thalesians.com/2025/01/21/an-interview-with-daniel-j-duffy-the-origins-of-c-in-quantitative-finance/

What about IP issues, copyright etc.?

stochastic processes, numerical methods, PDEs, convex optimization, derivatives pricing. Thinking out loud: FEM and FDM applied to option pricing etc. https://www.datasim.nl/application/files/5915/7045/5027/Matt_Robinson_Thesis_.pdf

Cool. Originally, he was called Setanta. Cuchulain stories are based in Cooley, Co. Louth where I have a cottage. My mum founded the Cuchulainn Ceili Band back then. POC FADA

Wow! Cuchulainn is my name on www.wilmott.com

#include <iostream> #include <string> #include "boost/pfr.hpp" struct some_person { std::string name; unsigned birth_year; }; int main() { some_person val{ "Edgar Allan Poe", 1809 }; std::cout << boost::pfr::get<0>(val) // No macro! << " was born in "...

// Copyright 2020 Peter Dimov // Distributed under the Boost Software License, Version 1.0. // https://www.boost.org/LICENSE_1_0.txt #include <boost/describe.hpp> #include <boost/mp11.hpp> template<class E> char const * enum_to_string( E e ) { char const * r = "(unnamed)"...

Thanks for the feedback! For MFE, Parts A,B,F,G of ODE/PDE course are sufficient for a good overview and certificate. Some middle-level calculus is needed but most students find it easy. Normally, 2-3 months is fine to complete. I support all parts if you want to do them all. The course...

Template metaprogramming in C++11 and beyond https://www.boost.org/doc/libs/1_87_0/libs/mp11/doc/html/mp11.html

We'll need a bigger desktop :)

So, pre- C++20 iterator concepts aren't actually concepts in the sense of the C++20 language feature. DD probably not With C++20, these requirements have been formalized as C++ concepts. The older STL algorithm variants have been left untouched. For instance, if you look at std::find, an...

My main concern with be getting stuff from ChatGPT, a bit like a 2nd hand car salesman.

I write books. I never use ChatGPT, just like I would not copy-and-paste someone else's code. And it is so longwinded and 2nd hand information. We did a test once, yawn.

How does your C++ knowledge compare to this C++ syllabus? This course is like judo; if you complete it then you get a brown belt. https://quantnet.com/cpp/

The maths part of the curriculum looks superficial at first glance..

A popular ODE/PDE course with MFE students is https://www.datasim.nl/onlinecourses/97/distance-learning-ordinary-and-partial-differential-equations

Might be interesting

Yes! I am now finishing the book alone. So, I have about 3-4 chapters to do.

During undergrad in maths, most of us avoided Category like the plague. Stay focused.

those with COM experience know how. And Boost aka intrusive ptr.

Look to the maths. https://en.wikipedia.org/wiki/Monad_(category_theory)

I thought DoD were hiring Rust programmers by the bucketload?? Have developers got the run of themselves? Tulip mania (Dutch: tulpenmanie) was a period during the Dutch Golden Age when contract prices for some bulbs of the recently introduced and fashionable tulip reached extraordinarily high...

Goodbye, Rust. I wish you success but I'm back to C++ (sorry, it is a rant) …And the problem with Rust is that it just doesn't have critical mass and, frankly, I don't think it will ever have. Recently, Linus Torvalds complained somewhere that old C dinosaurs don't want to learn Rust. For...

Closely related is HOF (might be in C++ 38) https://www.boost.org/doc/libs/1_86_0/libs/hof/doc/html/doc/index.html

Boost Fusion = uses STL and mpl Choose MPL over fusion when doing pure type calculations. Once the static type calculation is finished, you can instantiate a fusion sequence (see Conversion) for the runtime part...

PFR 101 // Increment each field of the variable on 1 and // output the content of the variable. #include <cassert> #include <boost/pfr.hpp> #include <boost/type_index.hpp> #include <iostream> struct test { int f1; long f2; }; test var{ 42, 43 }; boost::pfr::for_each_field(var...

I would first concentrate on type_traits and Concepts in C++20 and take it from there. BTW We discuss C++20 Concepts + examples + exercises in QN < Advanced C++ course.> :ninja: Check out this useful stuff after that Boost Describe Boost PFR #include <boost/describe.hpp> #include...

Visitor and Multimethods

20.8 Hardware Simulation and Bridge Pattern

19.6.1 Example: The Strategy Design Pattern using C++20 Concepts This is a game changer

I can try to debug this code but a better idea is to map the TA in section 13.2.2 (13.11 and 13.12) to code or use the code in my book. mixing a and a2 is confusing for me. https://en.wikipedia.org/wiki/Tridiagonal_matrix_algorithm

Ju no kata is a bit like C++

std::vector<double> r2(r.begin() + 1, r.end() - 1); => FIRST INDEX is ONE?? r2[0] -= BCL; r2[r2.size() - 1] -= BCR; edge condition? r2(J-1)? take example p. 403 Duffy 2018.

I would first uncouple the 2 algos .. no mixing of a and a2 etc. Thomas BC taken care of in RHS r term. J = 4 DS a(J+1) TA a2(J-1), better than cryptic a2(,,,,). I use DS in many apps. It originated in former Soviet Union. It is about 20% faster than TA.

Really??

They really want you to have done real analysis as I think some of the math courses jump into some advanced probability right away (so would be good to know measure theory as well). This is 3rd/4th year undergrad honours Maths degree level.

This course is on the UCB pre-requisite list https://www.datasim.nl/onlinecourses/97/distance-learning-ordinary-and-partial-differential-equations

Herb Sutter Leaves Microsoft: C++ 26 to Revolutionize Software Development Herb Sutter, chair of the ISO C++ committee and a long-time Microsoft veteran, has left the company after 22 years to become a Technical Fellow at Citadel Securities, where he will focus on technical strategy and...

Great ideas is the easy part. There are no posts yet.. @APalley @Paul Lopez

I would never do that; strangles are more low latency and memory-friendly.

Judo Black Belt Daniel Duffy

The only problem I have is with C++ object oriented programming, the pointers and the likes. They make me fret a bit for now. How do I solve this? QN C++ course

Even I used to wear suits in the beginning: 1990 Amsterdam ... very first international C++ course using a REAL compiler. Organised by Datasim and me! We was the first OO_EVERYTHING company in the Netherlands.

Plan : using std::future, easier than threading. It's in the Adv C++ course.