Hi

[this question refers specifically to the C version of NR, and possibly the C++ version, though I've never used that]

Am I correct in thinking that a function requiring a vector input (eg. the ODE solvers) could actually be passed a 2D matrix (as allocated using the funcs in nrutil.c) by passing the address &array[0][0] since the 2D arrays are contiguous chunks of memory? Of course the number of vector elements would need to be calculated correctly. I know that for 0 offset arrays and vectors this is true, I'm just wondering if the unti offset for 2D arrays destroys the contiguous nature of the memory area?

Thanks,
Jonathan

Hello, Jonathan.
This response refers to NR in C, not to NR in C++.
The answer to your question is basically "yes", but with one caveat.
If you allocate a matrix using the **matrix(), **dmatrix(), or **imatrix() routines in Appendix B of NR in C, then the storage is allocated by a single
malloc((size_t)((nrow*ncol+NR_END)*sizeof(--whatever--)))
All the nrow*ncol data values are in the addresses &array[nrl][nrc] and sequentially. Note this is not necessarily &array[0][0] unless **matrix has been created with values for zero offset. If it is unit offset, then the address of the first data point is &array[1][1], but you can still do what you want to do.
Hope this helps!
Cheers,
Bill P.