Why measurements are made of effects

This paper introduces generalized measurement theories (GMTs) as a complementary framework to general probabilistic theories to rigorously demonstrate that measurements must be composed of effects whenever probabilistic states can distinguish between different measurements, a condition argued to be physically well-motivated.

The Gist

Here is an explanation of Tobias Fritz's paper, "Why Measurements Are Made of Effects," translated into simple, everyday language with creative analogies.

The Big Question: Why Do We Measure Things This Way?

Imagine you are a detective trying to solve a mystery. In physics, the "mystery" is the state of a particle (like an electron), and the "detective work" is measurement.

In almost every physics textbook (from classical mechanics to quantum theory), there is a strict rule for how a detective works:

1. You have a list of possible outcomes (e.g., "Spin Up," "Spin Down," or "Red," "Green," "Blue").
2. For each outcome, you assign a number (a probability) representing how likely it is to happen.
3. The Golden Rule: If you add up all those probabilities, they must equal 100% (or 1).

In the language of physics, these "numbers" are called effects, and a full measurement is just a list of effects that sum to 1.

The Paper's Big Question: Why is the universe built this way? Why can't we have a measurement where the numbers don't add up to 1? Or where the rules are totally different? Is this just a quirk of our current math, or is there a deep physical reason?

The New Toolkit: Generalized Measurement Theories (GMTs)

To answer this, the author builds a new mathematical playground called Generalized Measurement Theories (GMTs).

Think of standard physics theories (like Quantum Mechanics) as a Lego set with very specific instructions: "You can only build towers if you use these specific bricks."
The author says, "Let's throw away the instruction manual."

In this new playground:

• We don't assume measurements are made of "effects" (probabilities) at all.
• We don't assume states (the things being measured) exist first.
• Instead, we start with the measurements themselves as the primary building blocks.

The Analogy: Imagine you are studying a new language.

• Standard Physics: You assume the language has nouns, verbs, and grammar rules, and you study how sentences are formed.
• Fritz's Approach: He says, "Let's just look at the sounds people make. We won't assume they form words or sentences yet. We'll just see what happens if we treat the sounds as the fundamental thing."

The Discovery: When Do Measurements Become "Effects"?

The author asks: "If we start with this free-form playground, under what conditions do we end up with the standard rule (that measurements are lists of probabilities)?"

He introduces a concept called Probabilistic Separation.

The Analogy: The Taste Test
Imagine you have two different soups, Soup A and Soup B.

• If you have a panel of tasters (the "states"), and for every possible taster, Soup A tastes exactly the same as Soup B, then for all practical purposes, Soup A and Soup B are the same soup.
• However, if there is even one taster who can tell the difference, then they are distinct soups.

In physics, a "state" is like a taster. A "measurement" is the soup.
The paper proves a powerful theorem:

If every measurement in your theory can be distinguished by some "taster" (state), then that measurement must be a list of probabilities (effects) that sum to 1.

The "Aha!" Moment:
The reason measurements in our universe are made of effects isn't because of some arbitrary math rule. It's because we assume that physical states can tell the difference between different measurements. If two measurements gave the exact same results for every possible state, we would treat them as the same thing. Once you accept that "states can distinguish measurements," the math forces the measurements to look like probabilities.

The "Classical" vs. "Quantum" Distinction

The paper also explores what makes a theory "Classical" (like everyday life) versus "Quantum" (weird, spooky physics).

The Analogy: The Master Key vs. The Puzzle

• Classical World (Strongly Classical): Imagine you have a set of keys. If you have a key for the front door and a key for the back door, you can easily make a "Master Key" that opens both at the same time. In math terms, any two measurements can be combined perfectly. This corresponds to Boolean Algebra (true/false logic).
• Quantum World (Not Strongly Classical): Imagine you have two locks. Sometimes, trying to open one lock jams the other. You cannot combine certain measurements perfectly. This is called Contextuality.

The author shows that if a theory allows you to combine any measurements perfectly (Strong Classicality) AND allows you to "zoom in" on specific parts of a measurement without breaking it (Projectivity), then that theory is mathematically identical to a Boolean Algebra. In other words, it's a classical world where everything is either True or False, with no weird quantum overlaps.

Summary: What Did We Learn?

1. The Setup: We usually assume measurements are lists of probabilities. But why?
2. The Experiment: The author built a theory where measurements don't have to be probabilities.
3. The Result: He proved that if you believe that "states" (the physical reality) can distinguish between different measurements, then the measurements automatically become lists of probabilities.
4. The Conclusion: The structure of quantum mechanics (and general physics) isn't arbitrary. It is a logical consequence of the fact that physical states exist and can tell measurements apart.

In a Nutshell:
Measurements are made of effects because reality is good at telling things apart. If you can't tell two measurements apart using any physical state, they are the same measurement. Once you accept that, the math forces the measurements to behave like probabilities.

Technical Summary

Here is a detailed technical summary of the paper "Why measurements are made of effects" by Tobias Fritz.

1. Problem Statement

In standard physical theories, particularly General Probabilistic Theories (GPTs) and Quantum Theory, measurements are modeled as $n$-tuples of effects (linear functionals) that sum to the unit effect (or identity operator). This structure relies on the "no-restriction hypothesis," which posits that any mathematically valid effect corresponds to a physical measurement.

The paper addresses a foundational question that has not been rigorously formalized in the literature: Why must measurements be composed of effects? Specifically:

• Is it possible to construct meaningful physical theories where measurements are not tuples of effects?
• Under what conditions does the structure of "measurements as tuples of effects" emerge naturally from a more primitive framework?
• How can one distinguish between "classical" and "non-classical" measurement theories within a generalized framework?

2. Methodology

The author develops a new mathematical framework called Generalized Measurement Theories (GMTs) to investigate these questions. The methodology proceeds as follows:

• Categorical Abstraction: Instead of defining measurements via vector spaces or convex sets (as in GPTs), a GMT is defined as a functor $M: \text{FinSet} \to \text{Set}$.
  • $M(X)$ represents the set of all measurements with outcomes in a finite set $X$.
  • For a function $f: X \to Y$, the map $M(f): M(X) \to M(Y)$ represents post-processing (coarse-graining).
  • The framework assumes only post-processing as the algebraic structure, deliberately omitting "mixing" (convex combinations) to test the minimal requirements for deriving standard measurement structures.
• State Definition: The paper introduces deterministic states (natural transformations $M \Rightarrow \text{id}$) and probabilistic states (natural transformations $M \Rightarrow \Delta$, where $\Delta$ is the probability distribution functor).
• Separation Condition: A crucial condition introduced is probabilistic separation: a GMT is probabilistically separated if distinct measurements can be distinguished by at least one probabilistic state.
• Classicality Analysis: The paper defines classicality through categorical properties:
  • Weak Classicality: All finite families of measurements are compatible (can be obtained as post-processings of a single joint measurement).
  • Strong Classicality: Joint measurements exist uniquely as categorical products (marginalization is a bijection).
  • Projectivity: The theory preserves equalizers (measurements are supported on the subset where post-processing functions agree).

3. Key Contributions

A. The Framework of Generalized Measurement Theories (GMTs)

The paper formalizes GMTs as functors from finite sets to sets. This allows for the construction of "toy" theories where measurements do not necessarily arise from effects.

• Example of Non-Effect-Based Theory: The author constructs a specific GMT (Proposition 4.2) where a system admits a single genuinely ternary measurement that becomes completely uninformative (trivial) upon any binary coarse-graining. This theory cannot be embedded into any GPT defined by effects, proving that the "tuple of effects" structure is not a logical necessity for all measurement theories.

B. Derivation of Effects from States

The central theoretical result (Theorem 4.5) establishes that if a GMT is probabilistically separated, it is isomorphic to a sub-theory of a GPT.

• Mechanism: Given a probabilistically separated GMT, one can construct a vector space of "quasi-probabilistic states." In this space, every measurement $\alpha$ in the GMT induces a tuple of linear functionals (effects) $(\hat{\alpha}_x)_{x \in X}$.
• Result: The normalization condition $\sum \hat{\alpha}_x = \text{tr}$ is automatically satisfied due to the naturality of the state maps. Thus, measurements are made of effects if and only if the measurements are distinguishable by probabilistic states.

C. Characterization of Classicality

The paper provides a rigorous categorical characterization of classical theories:

• Strong Classicality + Projectivity: A non-trivial GMT is isomorphic to a theory based on a Boolean algebra (i.e., classical deterministic or probabilistic logic) if and only if it is both strongly classical (joint measurements exist uniquely) and projective (measurements respect equalizers).
• Quantum Distinction: The paper demonstrates that while projective quantum measurements form a projective GMT, general POVMs (Positive Operator-Valued Measures) do not. This highlights a structural difference between projective measurements and general quantum measurements regarding the "support" of measurements on outcome sets.

4. Key Results

1. Theorem 4.5 (Main Result): Every probabilistically separated GMT is a sub-GMT of a General Probabilistic Theory. Consequently, in any physical theory where states can distinguish measurements, measurements are necessarily represented as tuples of effects summing to the unit.
2. Counter-Example: There exist valid GMTs (e.g., the "ternary measurement" example) that are not embeddable into any effect-based theory, proving that the "effect" structure is an emergent property of the separation condition, not a fundamental axiom.
3. Theorem 5.12: A non-trivial GMT corresponds to a Boolean algebra (is classical) if and only if it is strongly classical and projective. This links the algebraic structure of classical logic directly to the categorical properties of measurement compatibility and support.
4. Contextuality: The paper uses the GMT framework to re-prove and generalize contextuality results (Kochen-Specker, Gleason). It shows that theories like POVMs and projective measurements in dimensions $\geq 2$ or $3$ lack deterministic states, and certain "unknown function" theories lack even probabilistic states.

5. Significance

• Foundational Justification: The paper provides a conceptual explanation for why quantum mechanics and GPTs use effects. It argues that the "tuple of effects" structure is not an arbitrary choice but a necessary consequence of the physical requirement that measurements must be distinguishable by the states of the system. If two measurements yield identical statistics for all possible states, they are operationally indistinguishable and should be modeled as the same mathematical object.
• Unification of Frameworks: By placing GPTs and Quantum Theory as specific instances of the broader GMT framework, the paper clarifies the relationship between "states" and "measurements." In GPTs, states are primary; in GMTs, measurements are primary. The paper shows how the standard GPT view emerges when the separation condition is imposed.
• Refinement of Classicality: The distinction between "weak" and "strong" classicality, and the introduction of "projectivity," offers a more nuanced way to classify physical theories. It suggests that classicality is not just about the existence of joint measurements, but about the uniqueness of their construction (categorical products) and the behavior of measurements under equalizers.
• Future Directions: The framework opens the door to exploring physical theories where the "no-restriction hypothesis" is violated in non-trivial ways, or where uncertainty is modeled via possibilistic (non-probabilistic) structures, as illustrated by the "unknown functions" example.

In summary, Fritz demonstrates that the standard formalism of measurements as tuples of effects is a robust, emergent feature of any physical theory where states provide sufficient information to distinguish between different measurement protocols.