The Gold Standard of Universal Algebra | Tracking Conditional Laws Through Subalgebras, Products and Quotients

Begin with the integers under addition, negation and the constant zero. Consider the statement: if x+x=0, then x=0. Every integer satisfies this implication. Doubling a positive integer is positive, doubling a negative integer is negative, and doubling zero gives zero. There is no nonzero integer whose double is zero.

This statement is different from the identity x+x=0. That identity would require every integer to have double zero, which fails already at x=1. It is also different from asking for the solutions of one equation without making a claim about every possible input. The conditional law says something about all inputs, but only demands its conclusion on inputs satisfying the premise.

Write it as x+x=0 ⇒ x=0. The arrow means implication. It is not an operation belonging to the group, and it does not instruct us to divide by a stored element called two. Both sides of the premise and conclusion are terms in the declared signature: addition, negation and zero suffice.

At x=4 the premise is false, because 8 is not zero. That input therefore cannot refute the implication. At x=0 the premise and conclusion are both true. A genuine counterexample would require the premise to be true and the conclusion false at the same input.

This gives the first practical test. Before announcing a failed conditional law, record both evaluations. Showing that its conclusion is sometimes false is insufficient. The law never promised that conclusion on inputs where the required condition does not hold. Keeping that distinction visible will explain the quotient example later.

Contents · Next chapter

An identity has the form s=t, with all its variables understood universally. Associativity, for instance, requires (x+y)+z=x+(y+z) for every triple. It has no additional hypothesis that selects only some assignments. Every allowed triple is a test case for the equality itself.

A quasi-identity permits a finite list of equations as hypotheses, followed by one equation as conclusion. A typical form is: if s1=t1 and s2=t2, then u=v. All the variables still range over the whole algebra. An assignment becomes a possible counterexample only when every premise equation holds and the conclusion equation fails.

The finite conjunction matters. If one premise fails, the assignment does not violate the implication, even if another premise holds. For example, x+y=0 and x=z ⇒ z+y=0 follows simply by replacing z with x. Testing x+y=0 alone would ignore part of the statement being checked.

Ordinary identities can be regarded as the special case with no premise equations. The empty list of conditions places no restriction on assignments, so the conclusion must hold everywhere. This convention makes the relationship between the two forms precise without pretending they have identical preservation behaviour.

The usual algebraic treatment of quasi-identities concerns equations in one fixed signature. Inequalities, existential requirements and disjunctions need separate treatment. The assertion that a group has a nonidentity element, for example, is not an equation conditioned on a finite conjunction of equations in this form. Translate the actual sentence before invoking a preservation rule; a familiar-looking “if” in ordinary language is not enough.

Contents · Previous chapter · Next chapter

We use nonempty algebras whose basic operations are total. For additive groups the signature contains a binary addition, a unary inverse and a distinguished zero. The expression x+x is a term made from addition. The equation x=0 uses the named constant, so the language is part of the claim.

Different algebras can interpret that same signature differently while satisfying the group identities. The integers use ordinary addition. Residues modulo six use addition followed by reduction modulo six. The common symbols let us state the same conditional law in both, but do not guarantee that it has the same truth value.

Distinguish an integer from its residue class. The integer 6 is not zero. Its class modulo six is the zero class. This is not an inconsistency: equality is being evaluated in two different carriers. A quotient deliberately changes which original values count as equal after observation.

For a finite example, write the carrier explicitly as {0,1,2,3,4,5}, with arithmetic interpreted modulo six. These labels represent classes. Calling the operation ordinary integer addition would be incorrect, because 4+4 would then leave the displayed carrier. The actual group operation returns 2.

An audit should therefore begin with the carrier, each operation and the intended equality. Then write the premise terms and conclusion terms. This short specification prevents a common mistake: calculating a premise using quotient equality while judging the conclusion using original equality. Both parts of one test assignment must be interpreted in the same algebra.

Contents · Previous chapter · Next chapter

Map every integer n to its residue q(n) modulo six. This is a surjective group homomorphism. It preserves addition because reducing a sum agrees with adding the reduced residues, preserves negation, and sends zero to the zero class. Every residue is reached by an integer representative.

The integers satisfy x+x=0 ⇒ x=0. The quotient does not. Choose the residue x=3. Adding it to itself gives 3+3=0 modulo six, so the premise is true. But the residue 3 differs from the residue 0, so the conclusion is false. This is a complete counterexample to the conditional law in the image.

Check the original representative 3. In the integers, 3+3=6, which is not zero. Thus the premise was false before quotienting. There was no original obligation for 3 to equal zero. The quotient made the premise true by identifying 6 with zero while leaving 3 distinct from zero.

The map has not failed any homomorphism requirement. Indeed q(3+3)=q(6)=0, while q(3)+q(3)=3+3=0 in the quotient. These two results agree exactly. The failure concerns a conditional property, not preservation of the basic operation.

This distinction is worth saying aloud: a homomorphism preserves the result of a calculation after applying the map; it does not necessarily reflect equality back to the original inputs. Equal images can come from unequal original values. Here that permitted collapse creates a new premise instance and explains the entire failure.

Contents · Previous chapter · Next chapter

Why do ordinary identities survive surjective homomorphisms? Take an assignment of values in the target algebra. Surjectivity supplies a preimage for each variable. Evaluate an identity s=t on those preimages. It holds in the source by assumption, so applying the homomorphism gives equality of the target term values.

The term evaluation commutes with the homomorphism because each basic operation is preserved. One proves this through the term tree: variables map as assigned, constants are preserved, and an operation applied to already-correct subterms remains correct. The argument needs no additional premise to be established on the chosen preimages.

Try the same proof for a conditional equation. A target assignment may satisfy its premise. Choose preimages of the variables. To use the source conditional law, those preimages must satisfy the premise in the source. That is the step which can fail. Equality of the target premise values only gives equality after applying the map.

For our example, the target premise says q(3+3)=q(0). It does not say 3+3=0. The source implication cannot be invoked from the weaker statement. Writing these two equations beside each other exposes the unsupported inference without any appeal to mysterious quotient behaviour.

Surjectivity solves the problem of finding preimages of values. It does not solve the different problem of finding preimages that simultaneously satisfy a system of equations exactly. A conditional-law proof may require that stronger lifting property. Never substitute the first guarantee for the second merely because both are described informally as “lifting.”

Contents · Previous chapter · Next chapter

Let B be a subalgebra of A with the same signature. Its carrier is a subset closed under every operation and containing the required constants. Operations on B are restrictions of those on A. Consequently, evaluating any term on inputs from B gives the same value whether the calculation is viewed in B or A.

Suppose A satisfies a particular quasi-identity. Choose an assignment from B for which all the premise equations hold in B. They also hold in A, since the same inputs and operations produce the same values. The source law gives the conclusion equality in A, and that is the same equality between the term values in B.

Notice what did not happen. No new equality was introduced merely by restricting the carrier. Two elements of B equal in A are the same element of B, and unequal elements remain unequal. The subalgebra may remove possible assignments, but it does not merge the values of surviving assignments.

For a concrete check, restrict the additive integers to the even integers. This subset contains zero and is closed under addition and negation. If an even integer doubles to zero, it is zero, just as in the larger group. There are fewer candidate inputs, but the interpretation of each retained calculation is unchanged.

An arbitrary subset is not enough. Positive integers are closed under addition but omit zero and additive inverses, so they are not a subalgebra in the group signature used here. One may study them in a different signature, but that is a different construction. The preservation proof assumes a genuine subalgebra, including all its operation and constant requirements.

Contents · Previous chapter · Next chapter

Now take a family of algebras of the same signature, each satisfying the chosen conditional law. Their direct product performs every operation coordinate by coordinate. Equality of two product elements means equality in every coordinate. These two facts let us check the implication separately in each factor.

Suppose an assignment of tuples satisfies every premise equation in the product. Then each premise equality holds in every coordinate. In any chosen factor, the conditional law forces the conclusion equality for that coordinate assignment. Since this works in every factor, the two conclusion tuples agree throughout the product.

For the doubling law, use the additive product Z×Z. If (a,b)+(a,b)=(0,0), then 2a=0 and 2b=0 as integer equations. The integer property gives a=0 and b=0. Thus the original tuple was zero. This is a direct coordinate calculation, rather than an inference from the product merely looking like a grid.

The same reasoning works for any indexed product that is formed: each coordinate checks the same finite list of premise equations and the same conclusion equation. It does not require the factors to have identical carriers, only to interpret the common signature and satisfy the same law.

Be careful with a weaker observation. If you check the premise in only one coordinate, you have not established it in the product. For example, (0,4) doubles to (0,8), which has a zero first coordinate but is not the zero tuple. Coordinatewise reasoning is reliable when “every coordinate” remains attached to each equality being claimed.

Contents · Previous chapter · Next chapter

Consider the additive monoid of nonnegative integers, with addition and zero. It satisfies cancellation: if x+y=x+z, then y=z. This is another quasi-identity. It uses three variables and one premise equation, rather than the single-variable doubling example. Ordinary addition on nonnegative integers permits the cancellation step.

Define a four-element monoid with carrier {0,1,2,3} and operation a⊕b=min(3,a+b). Its top value represents a capped total. Zero is an identity. Associativity follows because both (a⊕b)⊕c and a⊕(b⊕c) equal min(3,a+b+c), using nonnegativity of the inputs.

The map h(n)=min(3,n) is an onto homomorphism from nonnegative-integer addition to this capped operation. If either input already reaches the cap, both sides of the preservation equation reach it. If neither does, capping their sum once gives the same result as the stated target operation.

Cancellation fails in the target. Take x=2, y=1 and z=2. Then 2⊕1=3 and 2⊕2=3, but 1 differs from 2. The premise has become true because two different totals have been capped to the same value. The conclusion remains false.

This example shows that the issue is not special to negative integers or modular arithmetic. A lawful compression of an associative operation can lose a cancellation property. If a later argument cancels a common input after such compression, it needs an independent justification. Associativity and operation preservation alone do not supply one.

Contents · Previous chapter · Next chapter

Write the capped operation table row by row. Adding zero returns 0,1,2,3. Adding one returns 1,2,3,3. Adding two returns 2,3,3,3. Adding three returns 3,3,3,3. Every entry belongs to the declared carrier, so closure is visible before any law is discussed.

Several properties survive. The table is symmetric, so the operation is commutative. The earlier minimum-of-total calculation proves associativity. Zero remains an identity, and three is absorbing: a⊕3=3 for every a. These facts coexist with the failed cancellation implication.

There is no valid subtraction operation supplied by this monoid signature. From 2⊕1=2⊕2, writing “subtract two from both sides” introduces a procedure the structure has not provided. Even if one uses ordinary integer subtraction on the displayed labels, that calculation is not an inverse for capped addition.

For example, the common capped result is 3. Ordinary subtraction gives 3−2=1, which recovers y=1 but not z=2. The operation discarded the distinction between uncapped totals 3 and 4. A single common output cannot determine which of those two second inputs was used.

This is an algebraic explanation of information loss, not a recommendation for any particular measurement system. A cap may be exactly the intended observation. The question is whether the subsequent calculation asks it to preserve a distinction it deliberately forgets. Recording the failed input triple makes that boundary precise and testable.

Contents · Previous chapter · Next chapter

Return to an additive group A and a congruence θ. Two original elements become equal in A/θ precisely when they are θ-related. The quotient satisfies the doubling implication exactly when, for every a in A, the relation a+a θ 0 forces a θ 0.

This formulation replaces quotient labels with a condition entirely on the original carrier and the chosen identification relation. It is not the same as requiring a+a=0 to imply a=0 in A. The equal signs have become θ-relations, which may hold for more pairs.

For the integers modulo six, θ means that the difference is divisible by six. At a=3, the double 6 is θ-related to zero, while 3 is not. The failed quotient assignment and the failed congruence condition are the same witness written in two languages.

For the integers modulo five, suppose 2a is divisible by five. Because two and five are coprime, a is divisible by five. Thus a θ 0 follows, and the quotient retains the conditional law. The original source is unchanged; choosing a different congruence changes the answer.

This is a useful repair strategy. Instead of asking whether quotients preserve all conditional laws automatically, translate the particular law into a property of the chosen congruence. Then prove or refute that property. The result is more informative than rejecting every quotient, and more reliable than assuming that any homomorphism will preserve the desired implication.

Contents · Previous chapter · Next chapter

Let the quotient be the additive group of residues modulo a positive integer n. We ask whether 2x=0 implies x=0. If n is even, the nonzero residue n/2 is a counterexample: doubling it gives n, which is zero modulo n, while n/2 is not the zero residue.

If n is odd, write n=2k+1. The integer k+1 satisfies 2(k+1)=n+1, so multiplying the equation 2x=0 by k+1 gives x=0 modulo n. Equivalently, two has a multiplicative inverse as a coefficient acting on this additive cyclic group.

Therefore the law holds exactly for odd n, including n=1, whose group has only zero. This is a full classification for this particular family and this particular implication. It is stronger than checking moduli five and six, because both directions have been proved for arbitrary positive n.

For n=9, the doubles of residues 0 through 8 are 0,2,4,6,8,1,3,5,7. Only zero maps to zero. For n=8, the doubles are 0,2,4,6,0,2,4,6, so residue four supplies the nonzero solution. The two finite lists provide independent checks of the general reasoning.

Do not turn this into the claim that every conditional group law is preserved by odd-order quotients. The argument depends specifically on the coefficient two and the cyclic structure under discussion. Changing the premise to 3x=0 changes the relevant divisibility condition, which becomes the transfer exercise near the end of the lesson.

Contents · Previous chapter · Next chapter

Failure of automatic preservation does not mean that every quotient damages a conditional law. A source might fail the law while a quotient satisfies it. Identifying elements can remove an existing counterexample by making its previously false conclusion true. Quotients can both create and eliminate possible failures.

For example, the additive group modulo six fails 2x=0 ⇒ x=0 at x=3. Map it onto the additive group modulo three by reducing residues modulo three. This is well-defined and operation-preserving because three divides six. The offending residue three now maps to zero.

The target group modulo three satisfies the doubling implication, since three is odd. The original counterexample has disappeared, and the target contains no new one. Checking only that one old witness disappeared would not prove the target law; the odd-modulus argument or exhaustive target calculation supplies the remaining evidence.

The one-element quotient provides an extreme case. Every term evaluates to its sole element, so every equation is true. Consequently every quasi-identity made solely from equations is true there as well. This explains why such laws alone cannot require the existence of at least two elements.

The right conclusion is therefore directional and limited. Subalgebras and products preserve a conditional law when their starting algebras satisfy it. General homomorphic images offer no comparable guarantee. They may retain it, lose it or acquire it from a source that lacked it. Each proposed quotient needs its own argument about the law being used.

Contents · Previous chapter · Next chapter

A collection of algebras defined by ordinary identities is called a variety. The Universal Algebra subject guide explains its closure under homomorphic images, subalgebras and products. Here we are studying the broader form of specification in which conclusions are equations conditioned on finitely many equations.

A class axiomatized by quasi-identities is called a quasivariety in the standard fixed-signature setting. Our direct proofs explain its closure under subalgebras and products. The integer-to-modulo-six example shows why homomorphic-image closure cannot be added to that list without further information.

There is a technical boundary worth retaining. Closure under subalgebras and products by itself is not the complete general characterization of classes axiomatizable by quasi-identities. The standard characterization also involves ultraproducts and the trivial algebra, with closure under isomorphism understood. The references develop that theorem; this lesson does not construct ultraproducts.

We do not need the full characterization to use the elementary tests. If a class is explicitly described by a collection of quasi-identities, the premise-by-premise subalgebra and coordinatewise product arguments apply to each law in the collection. Their conclusion is obtained directly from the stated specification.

Conversely, finding that a few examples are closed under a few constructions does not identify an axiomatization of an entire class. Keep the logical direction visible: known laws yield preservation proofs. Recovering laws from structural closure conditions is a separate theorem with its own hypotheses, rather than an inference licensed by a small experimental sample.

Contents · Previous chapter · Next chapter

One false repair is to call the failed conditional law an ordinary identity. The expression 2x=0 ⇒ x=0 contains a premise whose truth varies with the assignment. Removing the implication changes the assertion. Neither 2x=0 everywhere nor x=0 everywhere expresses the original property of the integers.

A second false repair is to demand surjectivity again. The modulo-six map was already onto. Every target element had a preimage, including the counterexample residue three. What was missing was a preimage satisfying the premise exactly, not a preimage as an element. Repeating the weaker condition cannot fill that logical gap.

A third false repair is to test only zero. The zero assignment satisfies both premise and conclusion in every additive group, so it is almost guaranteed to look reassuring. The useful search looks for nonzero elements whose double is zero. A diagnostic test should target the possible failure pattern.

A fourth false repair is to infer a quotient's truth from its size alone without stating its structure. Our odd-modulus result concerns the cyclic additive group of residues. A bare set with nine labels does not specify addition, zero or doubling. A different operation table requires a new calculation, even if the labels and cardinality look familiar.

Finally, an analogy about compressed information should not replace the proof. Compression explains why a failure is plausible, but some compressed structures retain the desired law. The congruence condition from Chapter 10 decides the issue. Use the analogy to ask a better question, then answer it with the actual equations.

Contents · Previous chapter · Next chapter

For a finite algebra, list the variables appearing anywhere in the premises or conclusion. Enumerate every assignment of those variables to carrier elements. Evaluate each term using the declared operation tables, including constants. Do not evaluate a quotient operation using unreduced integer arithmetic by accident.

For each assignment, first test all premise equalities. If any fails, record that the assignment is outside the counterexample search for this implication. If every premise holds, compare the conclusion values. A mismatch is a witness that disproves the law, and the checker can stop once it reports that witness clearly.

For the doubling law on six residues, the assignments with true premise are x=0 and x=3. The first passes the conclusion; the second fails. A useful report contains the input, the evaluated premise values, the evaluated conclusion values and the declared modulus. Merely printing “false” discards the explanation.

For the cancellation law on the four-element capped monoid, there are 4³=64 assignments to inspect. The triple (2,1,2) is one failure, but many others exist. Finding one is sufficient to disprove the universal implication. Counting all failures is an optional further question, not a requirement for the counterexample proof.

Exhaustive checking proves the result for the completely specified finite algebra. It does not establish the same statement for every modulus or every monoid. Conversely, a general proof can explain a whole family while a finite checker catches arithmetic or transcription errors in selected examples. Give each form of evidence its proper scope.

Contents · Previous chapter · Next chapter

Change the conditional law to x+x+x=0 ⇒ x=0 and keep additive cyclic quotients. First test moduli five, six and eight. Predict which will pass, then calculate a witness for each failure before reading the general criterion. This changes the coefficient while retaining the logical structure of the problem.

Modulo six, the nonzero residue two has triple zero, so the law fails. Modulo five, three has inverse coefficient two because 3×2=6=1 modulo five; multiplying the premise by two gives x=0. Modulo eight, three is its own inverse coefficient because 3×3=9=1 modulo eight, so the law also holds.

The general criterion is that three and the modulus n be coprime. If they are coprime, an integer combination of three and n equals one, allowing the same inverse-coefficient argument. If three divides n, the residue n/3 is nonzero and has triple zero. Since three is prime, those cases exhaust positive n apart from the harmless one-element case already included.

Now compare doubling and tripling modulo eight. Doubling fails at four, while tripling passes. The same quotient can preserve one conditional law and fail another. It is therefore misleading to label a quotient simply “safe for cancellation-like reasoning” without specifying the exact premise and conclusion.

For an independent extension, replace three by four and use moduli six and nine. Modulo six, residue three has quadruple zero; modulo nine, four has inverse coefficient seven. The calculations point to the broader coprimality criterion, but the task here is to explain each supplied case and identify precisely which part of the earlier proof transfers.

Contents · Previous chapter · Next chapter

Does a false premise make the whole conditional law false? No. A counterexample needs true premises and a false conclusion on one assignment. A false premise means that assignment does not violate the implication. This is why the integer input three does not refute the doubling law before reduction modulo six.

Can a conditional law hold only because its premise never occurs? Yes, an implication can hold vacuously. In our additive examples the zero assignment does satisfy the premise, so the whole property is not based on an entirely impossible premise. Nevertheless many individual assignments satisfy the implication because their premise is false.

Does an injective homomorphism create the same difficulty? Injectivity reflects equality: equal images imply equal original values. Combined with operation preservation, this removes the specific premise-lifting problem for assignments in its image. An isomorphic copy therefore has the same quasi-identities, as expected from a change of labels preserving all structure.

Must every quotient of a law-satisfying algebra fail? No. The integers modulo five retain the doubling law, while the integers modulo six do not. The failure of a general preservation theorem means that some counterexamples exist, not that every instance fails. Check the chosen congruence and the chosen law.

Is this the same as solving equations numerically? Solving an equation identifies assignments satisfying it. Testing a quasi-identity asks whether every assignment satisfying a specified system also satisfies another equation. The solution sets are related, but the second task contains a universal inclusion claim. Listing a few successful solutions does not establish that inclusion for an infinite algebra.

Contents · Previous chapter · Next chapter

Begin with two columns headed source and quotient. In the source column write 3+3=6≠0 in the integers. In the quotient column write [3]+[3]=[0] modulo six, while [3]≠[0]. Explain which equality changed and why the homomorphism equation still holds. This is the central mechanism of the lesson.

Next reconstruct the subalgebra and product proofs without using their names as shortcuts. For a subalgebra, explain why term values and equality on retained elements stay unchanged. For a product, explain why every premise reaches every coordinate and why coordinate conclusions assemble into tuple equality.

Then work the capped-addition example from its table. Verify associativity through the total-before-cap formula, verify the homomorphism from nonnegative integers, and produce the cancellation counterexample. These are three different obligations. Completing only the failed cancellation calculation would leave the claimed quotient construction insufficiently justified.

Finally, compare two different coefficients and two different moduli. Record a nonzero witness when the implication fails and an inverse-coefficient argument when it holds. A useful explanation identifies the condition that changed, rather than reporting a collection of isolated yes-or-no answers with no common reason.

The capability to retain is careful movement between exact equality and equality after identification. A conditional law asks more than preservation of operations: its hypotheses must be respected in the setting where its conclusion is invoked. Once that extra obligation is visible, both successful constructions and failed transfers become easier to diagnose, prove and explain to someone else.

Contents · Previous chapter · Continue to related reading