add computational scientific problems

[knightron0]

• n-body simulations
• heat diffusion
• fluid dynamics simulations

i'm sure there are more possibilities here -- this is a live list

[tugrul512bit]

Maybe following are allowed too?

Collision detection: useful for physics engines, raytracing, etc (may require tree build/traversal)

(Generally two types)

• broad-phase: compares AABBs only to see if they collide each other and returns an i,j collision pair list (AABB[i], AABB[j]) .Each AABB contains a complex object geometry.
• fine-grain phase: everything vs everyhing collision checking inside a pair of AABBs

---

Molecular force calculation: useful for protein-folding

• grid of density cells: uses electromagnetic fields, etc to define forces on a volume
• particle in cell: uses particles directly but in an optimized way to find them
• brute force (only for small groups)

---

Collision response calculation (linear algebra required): useful for game engines, physics simulations (finds a good/fair update point for each particle)

• global solution: uses a big matrix (mat inversion, mat-vec multiplication, etc) to solve the linear system at once (not scalable but finds solution at once)
• local solution + iteration: uses single particle based updates (6x6 matrix) and leaves the global solution to multiple repeatations

---

Falling-sand simulation. (this can be easy)

[tugrul512bit]

Also a "sorting" problem would be nice. 1D sorting, segmented 1D sorting, 2D sorting, sorting for arrays with large elements. Because when doing physics calculations, sorting the particles can improve the performance.