Core concepts
The mental model behind the workspace: a single editable model that carries layout, connectivity, hierarchy, and physics at the same time.
Qfactr treats a photonic circuit as a physically grounded object, not a schematic abstraction. Components sit at real positions, waveguides are real geometry, and the things you connect are real optical ports. A handful of concepts hold this together. Once you have them, the rest of the workspace follows naturally.
This page defines those concepts and how they relate. If you are new to photonic integrated circuits, the PIC 101 primer covers the underlying physics; here we focus on how Qfactr represents a design.
Physical coordinates
The canvas is a true physical plane measured in micrometers (µm), not a grid of abstract tiles. Every component you place has a real position and a real footprint from the moment it lands, so the geometry you see is the geometry that will be fabricated.
This matters because in photonics, geometry is physics. The distance light travels, the radius of a bend, and the spacing between devices all change how the circuit behaves optically. A tool that only tracks connectivity throws this away; Qfactr keeps it from the first placement, which is why loss and transmission can be derived directly from what is on the canvas.
Components, pins, and ports
A component is a real device with a physical footprint, a set of pins (its optical ports at fixed physical positions), and S-matrix (S-parameter) data describing how light at each port relates to light at the others. Parts come from PDK-aware libraries, so they behave like the devices they represent rather than as generic symbols.
Pins are where routing attaches. Because a pin has a real location and orientation, a waveguide connected to it starts as genuine geometry, not a logical link. Pins also carry visible connection state: a pin renders red when unconnected and green when connected, so the canvas tells you at a glance what is still open.
Explicit routes
Connections between components are real waveguide paths, not abstract nets. When you route from one pin to another, Qfactr draws an actual S-bend or Manhattan waveguide between those physical pin locations. The route has length, bends, and a shape that you can see and edit.
This is the central difference from schematic-style tools, where a wire is just a statement that two nodes are the same node. A photonic route cannot be drawn for free: every bend and detour costs optical power, and waveguides cannot cross without loss. Because routes are explicit geometry, the workspace can surface that cost directly. The routing guide covers how to draw and refine connections.
The representation layer
These pieces add up to the core idea: one editable model that holds layout, connectivity, hierarchy, and simulation-aware context together, in the same place. This is the representation layer. It is not a schematic that you later translate into a layout, and it is not a layout file that has lost its connectivity. It is a single model where placement, routing, and physics stay consistent with one another as you edit.
Keeping everything in one model is what makes the design legible to both you and the AI assistant: edits respect physical constraints, and results stay checkable against the actual geometry. The same model is what gets handed off when you export to downstream tools.
Hierarchy
Real systems are too large to manage as one flat collection of parts, so you can group components and their routes into a reusable block. A block exposes a set of external pins and has a frozen layout internally, then you instance and compose it into larger systems much as you would reuse a cell or subcircuit in electronic design.
For example, a wavelength filter built from a ring resonator and its couplers can become a block with just its input and output pins exposed. You place that block wherever you need filtering and route to its external pins, without re-deriving the internal layout each time. The hierarchy guide goes deeper.
Simulation-aware design
Because routes are real geometry and components carry S-matrix data, transmission and loss are derived from the actual path rather than estimated from a schematic. The workspace can surface transmission and loss for the current design directly, giving you fast, design-time feedback on how clean a layout is as you build it.
This in-canvas feedback is meant for rapid iteration, not as a replacement for full electromagnetic or circuit simulation. For deeper analysis you export the design to dedicated simulators. The simulation page explains the scope, and placing components shows how geometry enters the model in the first place.
The concepts at a glance
| Concept | What it means | Why it matters |
|---|---|---|
| Physical coordinates | Components live at real positions in µm on a true physical canvas. | Geometry is physics; the layout you see is what gets fabricated. |
| Components | Real devices with footprints, pins, and S-matrix data, from PDK-aware libraries. | Parts behave like the devices they represent, not as generic symbols. |
| Pins / ports | Physical optical ports with fixed positions; red when unconnected, green when connected. | Routes attach to real locations, and connection state is visible at a glance. |
| Explicit routes | Waveguides drawn as real S-bend / Manhattan geometry between pins. | Every bend, crossing, and detour costs optical power, so the cost is real and visible. |
| Representation layer | One editable model holding layout, connectivity, hierarchy, and physics together. | Edits stay consistent across all four; the design stays checkable and exportable. |
| Hierarchy | Reusable blocks with external pins and frozen layouts. | Reuse and compose subcircuits to manage complexity and scale. |
| Simulation-aware | Transmission and loss derived from the actual path geometry. | Fast design-time feedback on layout quality as you build. |