I had someone recommend Julia to me over MATLAB and they did it for performance reasons. When I looked about a year ago the stiff ODE solver in Julia gave the wrong answer for the kinds of problems I do. I stopped looking at that point.

Julia will mature in time but right now having something run faster but wrong is not useful.

Octave on the other hand is pretty nice I just like MATLAB more due to the whole IDE that is part of it but I agree with the drolli that in some circumstances octave is a better choice.

I agree that MATLAB can be expensive and that in many ways it would be better if everyone used FOSS. However the reality is that right now my choices are between MATLAB and python and for many tasks MATLAB allows me to get the job done faster and since the job involves human lives it makes the choice fairly simple.

I would like better FOSS tools to replace MATLAB and especially better documentation and defaults and the sad thing is I have the skill to do a lot of that work but not the time. Choosing between making a drug available and making numpy better is a fairly simple choice for me.

This is the kind of reply I like.

There are so many different kinds of tasks in the world that sometimes you find a language or library just seems to fit one of them very well and other times while you could do it the fit is not as good. In the end what matters is getting the job done.

There are just a few things in the MATLAB global optimization toolbox and some of the pde/ode solving built into it that make my life much easier and faster to develop with. However for many other tasks I use python. Right now the toolchain is mixture of python, c++, matlab and excel and strangely enough it works very well and it allows scientists to setup simulations that can then run on clusters and bring data back in a way that they can analyze it and massively speed up development.

How is R a comparison to MATLAB for engineering work? From what I have looked at it does not have a development environment to compare to MATLAB, it also lacks all the solving and non-linear optimization methods that MATLAB has built in. R may be very nice but solving systems of hundreds to hundreds of thousands of coupled ODEs, PDEs and doing non-linear optimization just does not look like something it is very good at.

I am doing very little statistics and mostly writing simulations.

I don't do much in the way of statistics and R just does not seem like a good fit for what I do at all. MATLAB has better PDE solvers, better non-linear optimization and stiff ODE solvers.

Most scientists I have seen publish their MATLAB code however I am more concerned about industry than open research.

How much slower do you think it is okay to get work done in order to put it in a completely free software framework? How many people is it okay to have die from the additional time?

Some things are free and done very well like OpenMP and MPI however for many other tools the free version is just not as good.

I have been a professional python developer for about 10 years now but when writing matrix based simulations and doing data visualization numpy, scipy and matplotlib are not viable competition to MATLAB.

Most free software projects have HORRIBLE documentation and epicly horribly defaults. The problem is that the people that know how to change these things are also too busy doing other work. Yes I do have the skills to fix many parts of matplotlib and numpy but I can also just use MATLAB and get my work done.

Since the work I do is on writing computing simulations for drug manufacturing the more time it takes me to solve a problem the more people DIE. I like free software a lot and have used it for a very long time but I am passed the point of caring much about the license or the cost of the software.

The problem is secret courts and that they have been caught spying on everyone multiple times already.

If he was arguing that they should be able to get a court order at a NORMAL court not the FISA one and with probably cause have the right to decrypt the data and only the data covered by the search warrant then I would support him.

I can definitely agree with this.

I know many people, including myself, that don't drink alcohol because it tastes bad. I have tried various kinds of alcohol and in all cases it has tasted worse than other drinks I could get.

There is supposed to be a mutation in a very small percentage of people and it gives alcohol a pretty nasty taste.

I have NO ethical problems with alcohol, I don't mind being around others that are drinking it I just don't like it myself.

He could probably find one for transfats being bad for you though.

However, that is a completely different kind of problems. There are 3 transfats that humans commonly encounter and we have special enzymes to deal with the trans position so we can process it.

The only problem with artificial trans fats is they have the trans bond at a different position and our body does not process it correctly. This ends up causing malformed cholesterols which then aggregate on your artery walls and cause damage. It is not the cholesterol that is bad but misshapen molecules that aggregate.

Any transfat that is not trans in one of the positions that we have an enzyme to handle should not be allowed in food. They are just incompatible with human enzymatic processing and that is the only reason to ban them. There is at least one trans fat in milk and another in beef and those are fine since we have special enzymes to process them.

Air is NOT an ideal gas at ALL. You can't use the ideal gas law and have it work.

However you are in luck though since engineers made tables long ago of air properties at a huge range of temperatures, pressures etc and you can just look up the properties of air. However the properties of the material of the football would have to be tested.

The only time you can use the ideal gas law is with a nearly pure gas at high temperature and no chemical reactions.

It does suck that so much of the stuff we teach people in chemistry is not actually useful.

Just because humans evolved with nature does not mean that is the best way for us to learn and grow.

Nature is just how we started. We have the capacity to learn and exceed it. There is no reason to believe that we can't do better than how we learned in the past.

I also don't see music as a critical skill to learn during development.

I nave not seen any kind of standardized test so far that I thought was a remotely accurate prediction of skill.

Overall humanity has a huge problem with education at this point. We have done the research and we know that memorization does not work for actual learning. However, no amount of research seems to turn into actual changes.

At this point I think we are going to have to just destroy the entire education system from grade school through grad school. They won't change and they live in their own world divorced from reality.

Even when you see a university publish major papers on how ineffective their own memorization based systems are they refuse to change. I have talked with some university professors about this and usually the reasons that are given for keeping the memorization based systems are politics, culture, history etc. None of which have anything to do with education.

The human race is being held back by the education system at this point and since they won't evolve they need to be replaced.

The way we teach calculus is based off of rote memorization. You need all the rules to solve the integrals. However, functional analysis is an almost entirely different kind of skill. Functional analysis is based on the theory that underlies calculus but that is usually skipped in order to just teach straight problem solving.

I see skills like functional analysis as more important since you learn what to expect from functions and why. The exact answer a computer can give you but a better understanding of functions will tell you very quickly if you made a major error in setting up the system on a computer, or if there are multiple answer how to determine which is the correct one for your system.

There is just not enough time to teach understanding (since it takes so much experience to gain it) and the memorization of rules for solving integrals and derivatives. Since any cellphone, laptop, tablet, etc can solve integrals and differentials but they can't give you understanding I think we should be spending time on the parts that computers can't do. As a result you can solve more realistic (and FAR harder) problems and you learn far more valuable skills in problem solving.

I can't even imagine dong that for my subject. It is impressive that you managed to do it and I am thankful I don't have to go down that path.

My Master's thesis will be on chromatography simulations at industrial concentrations with industrial bio-molecules.

Overall I think that computers have helped a lot if used wisely and have enabled entirely new areas of research that are saving hundreds of thousands of lives every year.

They can also definitely be abused but that is a reason to learn how to integrate them effectively that is not a reason to ban them.

Tech to read textbooks is great for engineering texts!

With an ebook you can SEARCH. Trying to find out where a table of values I need to pipe roughness or viscosity relationships vs temperature for a certain chemical is so much easier to do with searching. Most engineering books seem to have about a hundred pages or so of just tables, graphs etc at the end.