Stratified Sampling: Neyman Optimal Allocation versus Proportional and Simple Random Variance
Split a population into three strata and decide how many units to sample from each. Compare simple random, equal, proportional and Neyman optimal allocation, change the shares, spreads and the gap between strata, and see the variance of the estimate fall. A Blender-built plotting easel and layered cake, with Monte Carlo, brute-force allocation search and variance-decomposition checks built in.
In multi-task learning, a single shared parameter vector must satisfy every task's loss at once, and when two tasks' gradients point in opposing directions the naive sum of their gradients can partially cancel — or even move backward on one task. This simulator renders each task's quadratic loss as a bowl on a shared 3D landscape, lets you dial a conflict angle between task optima, and compares plain gradient summation against PCGrad's "gradient surgery": projecting out the conflicting component of one task's gradient onto the other's normal plane before summing. Run the optimizer, watch the instanced trail of past positions, and read live cosine-similarity and combined-loss telemetry to see exactly when and why gradient surgery helps.
Split a population into three strata and decide how many units to sample from each. Compare simple random, equal, proportional and Neyman optimal allocation, change the shares, spreads and the gap between strata, and see the variance of the estimate fall. A Blender-built plotting easel and layered cake, with Monte Carlo, brute-force allocation search and variance-decomposition checks built in.
3D · Three.js / WebGL renderer · 60 FPS target · runs fully client-side, no install