Fix: zero result buffers in MatrixUtils to prevent neural network output corruption

[clubsnake]

Summary

The three matrix multiplication functions in MatrixUtils.cs (MatrixProduct × 2, MatrixProductTranspose) accumulate into a result buffer using += but never zero it first. When a caller passes a pre-allocated buffer — as FeedForwardNetwork.CalculateOutputs() does with tempResults on every physics tick — values from previous forward passes accumulate indefinitely instead of being computed fresh.

Root cause

// FeedForwardNetwork.cs, lines 106-108
float[] tempResultVec = tempResults[i];          // allocated once, reused every tick
result = MatrixUtils.MatrixProductTranspose(
    layerWeights, result, tempResultVec);         // += into stale values

tempResults is allocated in the constructor and never cleared. Each call to MatrixProductTranspose adds the new dot-product results on top of whatever was left from the last tick.

Why it's hard to notice

Sigmoid activation saturates quickly. After ~5-10 ticks of accumulation, every pre-activation value is so large (positive or negative) that sigmoid(x) maps it to ≈0 or ≈1 regardless of the actual inputs. The network still "works" — it just outputs near-constant values that ignore sensory input, which is why creatures converge on simple repetitive motions (e.g., rhythmic flexing) rather than developing adaptive, input-responsive gaits.

Observable impact

• Creatures plateau early and cannot learn complex behaviors
• Larger/deeper networks don't improve performance (their extra capacity is wasted)
• The genetic algorithm struggles because fitness differences between chromosomes are flattened by the saturated outputs

Fix

Zero each result element before the inner accumulation loop — one line per function:

result[i] = 0f;   // added before the k-loop in each function

This is a minimal, safe change. When result is freshly allocated (the null path), the zeroing is redundant but harmless. When result is reused (the bug path), it's essential.

Testing

• Verified the bug exists in the current master source
• Confirmed FeedForwardNetwork.CalculateOutputs() passes reused tempResults buffers
• Tested the fix in a compiled build — creatures now produce input-responsive outputs and evolve more diverse behaviors

---

Bug discovered and fix authored with assistance from Claude (Anthropic).