Code review

[martinradev]

I will start the code review. Additional comments will be done accordingly:

• Transpose matrix computation is correct
• Matrix multiplication is correct (Kertolasku)
• Matrix reduce (? isn't it just subtraction) - Vahennyslasku - it is fine. However, maybe you could change += to =. It is not like += is a more heavy operation, but it is a small change. Also, here you rely on the fact that the matrix is initialized with 0s. In some languages this is not the case. Would be a good practise for the future.
• Multiplication by a constant factor (Skalaaritulo). It is fine. Again the above comment. Presumably = would be better than +=. Also, since you have two types of multiplication - by some scalar and by a matrix, then maybe you should use method overloading.
• Sumation (Yhteenlasku). Clearly, the same operation is matrix reduce (Vahennyslasku). You could combine them in one class. Just check whether it is summation or subtraction. In any case, you could still remove the += and set =

The project seems good, but there are a few things which may be good to consider: the array may be best to be repesented by objects and in the class of the object have those 2 dimensional arrays. Accordingly, this will give you the possibility to support vector operations like Cross product, dot product, etc. My review will continue, but have those in mind. I will be very interested to check the Straessen algorithm the following days.

[martinradev]

Tests have been run. All passed. The project runs on Ubuntu 12.04 with netbeans 8.0

[martinradev]

Normally, when using divide and conquer algorithms (Strassen's algorithm) it is a good idea to use the default (naive) way to compute the answer for smaller dimensions of the input. For example, if the matrix is of type 3x3 or smaller, then you could use the normal algorithm. This is also not the fastest solution you can use. Maybe you could consider the "Coppersmith–Winograd algorithm" which would be maybe too hard to implement.

[martinradev]

In some cases, it is very valuable to compute the power of a matrix. You could check exponentiation by squaring so that you compute it faster. Maybe you could also improve it by a constant factor for this case.

[martinradev]

Think also about singular value decomposition. It is quite useful in many advanced algorithms which do some kind of classification like LSA.

[martinradev]

I believe this is the end of the issue. Generally, the code is nice. I would like you to think about the topics I have given you and think about doing some work on them. It was hard to do a better review since everything was in Finnish (even variables and methods which is generally maybe a bad idea). I think that in the future it would be better for your career to focus on writing your code in English since possibly those are standards in most companies. Generally a nice project and possibly different than the others. If you need some kind of reference I would suggest Poole's Linear Algebra book. It has a lot of ideas regarding not only mathematics, but how to do approximations and computations of specific things in comp sci.