TensorSpace Notes

Short notes on working with arrays and linear algebra in practice. Each one explains a single idea with a small example that you can run.

Notes

• Shapes and broadcastingHow array dimensions line up, and the bugs that follow when they line up by accident.
• Memory layoutRow-major and column-major order, strides, and why loop order changes run time.
• Matrix multiplication orderThe same product can cost ten times more depending on where the brackets go.
• Automatic differentiationForward and reverse mode, worked by hand on one small function.