As others have said, Matlab is definitely the most powerful and probably the most used tool in the industry for mathematics in engineering/computing.
However, I'd also like to note that Functional programming languages are a HUGE bonus.
Scala and Haskell are some of the best due to the fact they can be used on Linux systems.
A fun bonus is F#, which is a very powerful .NET functional-programming language. (It's perfectly compatible with Mono, but I'd recommend programming on Windows using MS Visual Studio with "Xamarin Studio" and "Mono for Windows" 3rd Party extensions to develop cross platform applications straight from the MS Visual Studio IDE)
C# is a fun one, it's not a functional programming language (It's an OOP language), but it has a lot of functional programming features. C# is often compared and contrasted to F#, and most people that go to learn F# will find that authors of articles and books will treat F# as more of a conversion from C#. Not everyone does this, but a good portion do.
C and C++ are used a lot in scientific industry simply due to the amount of power they provide. You should screw around with some stuff. While Matlab is awesome, I personally think you should try a functional programming language first.
I'm a Visual Studio and .NET guy.. So I'm going to vouch for C# and F# over most things, but like I said, Haskell and Scala are also very powerful (I'm more on Scala's side, Haskell is C++ while Scala is C#.. I personally think C# is the most well developed programming language ever; it's like Java, Python, and C++ all had a baby named Beyonce).
Everyone has their own opinion
Haskell does not run on the Java Virtual Machine (JVM), but Scala does. Frege is basically a Haskell imitation that runs on the JVM though if you were interested in that.
Summary:
Pure math: Matlab
Functional Programming Language: Haskell
FP (Haskell imitator) on JVM: Frege
OOP/FP Hybrid on JVM: Scala
FP .NET Language: F#
OOP/FP Hybrid .NET Language: C#
Give those a shot. Functional programming is what is used a lot in Machine Learning, which is often very heavily tied into complex mathematical algorithms involving anything from basic statistics to chaos theory and advanced combinatorics.
(Disclaimer: I hate python, so if I left some Py languages out, that's why; sorry).