* Implement methods from the Runge Kutta family

* Implement multi-stage methods (Adams...)

* Try to implement methods with adaptative stages in order to verify if such an operation is possible

* Implement the management of first degree with constant coefficient linear scalar equations in the kernel: homogenous -> correct resolution; non homogenous -> adapted numerical methods

* Implement the management of second degree with constant coefficient linear scalar differencial equations : homogenous -> correct ; non homogenous -> adapted methods such as Newmark

* When the matrix are operational (especially the algorythm of Jacobi's diagonalisation, iterated powers...), we should think about the implementation of degree one and degree two linear differencial systems   
