{"id":"0b1829ea-8144-47e6-a993-a294f034df18","arxiv_id":"2501.07060","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"An in-place quantum adder-by-constant circuit is constructed with n-3 ancilla qubits and T-count 4n-5, improving on prior adder-based constructions.","lead":"This paper presents a quantum circuit that adds a classical number to a quantum register in place, using fewer helper qubits and T gates than previous approaches. Such addition is a core building block in Shor's factoring algorithm and other quantum arithmetic, so reducing its cost matters for practical fault-tolerant quantum computers.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Controlled-adder T-count 11n−15 is not justified: Sec. 3.4 never analyzes how Gidney's measurement-based AND-uncomputation corrections behave under the external control, so the count may be too low.","rationale":"The reader's weakest-assumption and my stress-test both target Sec. 3.4's controlled T-count. I agree that the derivation is missing a treatment of AND-uncomputation corrections, though I would not phrase it as 'X on I/O becomes a Toffoli'; a classically-conditioned X, when controlled by the external qubit, is a CNOT. The underlying gap is real: the paper does not enumerate the corrections or show that all remain Clifford after adding the control. This is load-bearing because the abstract explicitly claims 11n−15 for the controlled version. The uncontrolled circuit and its 4n−5 count are supported by the detailed construction and by the availability of Q# code, so the appropriate verdict remains conditional pending a resource audit. I suggest the concrete check above because it settles both correctness and the exact T-count using the author's own implementation.","tokens_in":117,"tokens_out":24284,"duration_ms":377043,"concrete_test":"Run the supplied Q# implementation [10] for the controlled adder at n=5 and n=6: in a state-vector simulator, verify for every constant a and every b that the controlled circuit maps |0>|b> to |0>|b> and |1>|b> to |1>|a+b mod 2^n>, then use Q#'s ResourceEstimator to count T gates. If the T-count is 11n−15 and the semantics hold, the concern is resolved; if the count exceeds 11n−15 or the semantics fail, the controlled resource claim must be revised.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central resource claim for the controlled circuit (Sec. 3.4, T-count 11n−15) counts n−2 AND/AND† pairs at 4 T each and n−1 Toffoli gates at 7 T each. Missing is any treatment of the measurement-based uncomputation of the AND gates from [5]. In Gidney's construction, an AND† is implemented by measuring the target and applying classically conditioned Clifford corrections; the correction set is not specified in this paper. In the controlled adder, each correction that acts on an I/O qubit must additionally be conditioned on the external control qubit. If the correction is an X or Z, the controlled version is a CNOT or CZ and adds zero T; the reader's worry that an X becomes a Toffoli appears to be a miscount, since the measurement outcome is classical. But if any correction is a phase gate of controlled-S type, its controlled version is non-Clifford and adds T gates, and the paper gives no bound on the number of occurrences. Without this analysis, 11n−15 is an unverified upper bound rather than a proven resource count. The uncontrolled 4n−5 claim is not affected.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proposes a quantum circuit for in-place addition of a classical n-bit constant to an n-qubit register modulo 2^n. The construction generalizes an incrementer circuit from [9], replaces certain Toffoli gates with Gidney's temporary logical-AND/AND† pairs, and applies several optimizations. For the uncontrolled adder the paper claims n−3 ancillas and T-count 4n−5; for the controlled version it claims n−2 ancillas and T-count 11n−15. The circuits are implemented in Q# and compared against several known quantum adders used with a constant-loading construction.","tokens_in":3980,"tokens_out":6984,"duration_ms":73543,"significance":"If the claims hold, the uncontrolled circuit improves on the standard constant-loading approach by using substantially fewer ancillas (n−3 versus about n plus the adder's ancillas) while matching or slightly improving the best T-count (4n−5 versus 4n−4 for Gidney's RCA). The paper's concrete gate counts, explicit T-count formulas, and release of Q# code are useful strengths. The controlled-adder claim, however, is not yet supported by a complete gate-level analysis, so the significance of the controlled result is conditional on that analysis.","major_comments":[{"comment":"The T-count 11n−15 for the controlled adder assumes that n−2 AND/AND† pairs cost 4 T each after imposing an external control. This is not justified, because the AND† gate from [5] is implemented by measurement and classically conditioned corrections. The sentence 'make all the gates that target input/output qubits (but not ancillas) controlled' does not specify how these measurement-based corrections are controlled. If any correction becomes a controlled non-Clifford gate when conditioned on the external control, the T-count increases. Please provide the explicit controlled implementation of each AND/AND† pair, or prove that all corrections remain Clifford after conditioning; otherwise 11n−15 is an unverified upper bound rather than a proven resource count.","section":"Section 3.4"},{"comment":"Correctness and T-count for arbitrary n are asserted from the n=5 diagrams and the analogy to [9], but no inductive proof or machine-checked verification is given. The Q# repository [10] is referenced but not auditable from the text. Since the central claim is an exact T-count of 4n−5, please add a rigorous correctness proof for general n, or a formal verification trace, so that the gate cancellations and the final gate inventory can be checked for all n.","section":"Sections 3.1–3.3"},{"comment":"The comparison methodology is ambiguous: the text says the fig. 1 construction 'adds n ancillas', but the table's 'Ancilla' column does not state whether those n ancillas are included in the listed counts. Without this clarification, the table cannot be reproduced and the claim that the proposed adder uses fewer ancillas than all compared constructions is not precisely supported. Please state the convention explicitly and, if the listed numbers exclude the constant-loading ancillas, provide the total ancilla counts.","section":"Section 4, Table I"}],"minor_comments":[{"comment":"The displayed formula for MAJ(x, y, a) contains a typo: the second conjunct should be (y⊕a), not (x⊕a), so the expression should read ((x⊕a)∧(y⊕a))⊕a.","section":"Section 2"},{"comment":"The sentence 'We denote this Xa(q)... as shown on fig. 3' appears to refer to Figure 2, not Figure 3; please correct the cross-reference.","section":"Section 2, Figure 3 caption"},{"comment":"Optimization item 3 ('Eliminate the first ancilla, because c1 = b0') is not self-evident from the text; a short explanation or relabeled diagram would make the ancilla count n−3 transparent.","section":"Section 3.3"},{"comment":"The GitHub implementation link should include a commit hash or version, so that the code can be audited reproducibly.","section":"Section 3.5"},{"comment":"The statement 'more efficient than any circuit that can be constructed using any known quantum-by-quantum adder' is a strong claim; it would be more precise to say 'among the constructions compared here, to our knowledge' unless an exhaustive survey is intended.","section":"Section 5"}],"recommendation":"major_revision","confidential_remarks":"The manuscript is short and the central uncontrolled construction appears plausible, but the controlled-adder T-count is not proven at the gate level. I recommend requesting an explicit treatment of the measurement-based AND uncomputation under external control, and a formal correctness argument for general n, before considering publication."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short take: this is a believable improvement to adder-by-constant: n-3 ancillas and T-count 4n-5 for the uncontrolled version, generalizing Li et al.'s incrementer and using Gidney's AND gates. The main soft spot is the controlled version: the T-count 11n-15 is asserted without analyzing how Gidney's measurement-based AND-uncomputation corrections behave when every gate on I/O qubits is made controlled. Until that's filled in, treat the controlled count as an upper bound that might be too low.\n\nWhat's genuinely new: the uncontrolled adder-by-constant construction and its resource counts. The idea of building adder-by-constant directly, rather than loading the constant into ancillas, is not new in principle, but the specific generalization to arbitrary odd constants and the use of AND/AND† pairs to get 4n-5 T-count appears to be new and is arithmetically consistent. The circuit diagrams are helpful, and the author provides Q# code. The uncontrolled circuit's counts are plausible and the derivations check out modulo the typo in Sec. 2 where the MAJ formula should be ((x⊕a)∧(y⊕a))⊕a.\n\nThe weakest part is Sec. 3.4. The controlled adder's T-count of 11n-15 assumes n-2 AND pairs at 4 T and n-1 Toffolis at 7 T. But Gidney's AND† is measurement-based; its corrections are Clifford for the uncontrolled circuit, and when you add an external control, each correction on an I/O qubit must itself be controlled. The paper does not specify the correction set or bound the number of non-Clifford corrections. The specific fear that an X correction becomes a Toffoli is probably wrong—measurement outcomes are classical—but the broader issue remains: controlled-S (or similar) corrections would add T gates, and the paper gives no argument they don't occur. So the controlled T-count is unverified, not proven. It can likely be fixed by checking the Q# implementation, but the paper as written doesn't do it.\n\nAlso minor: the comparison table is a bit confusing. The text says the constant-loading procedure adds n ancillas, but the table entries for the known adders look like the adders' own ancilla counts, not the totals after adding n. Clarifying this would help.\n\nVerdict: the uncontrolled part is worth knowing and citing; the controlled part needs more work before I'd rely on it. The paper deserves a serious referee, and I'd accept it for review with a request to fix the controlled analysis and the table.","headline":"A solid incremental improvement to adder-by-constant, but the controlled-adder T-count needs a missing analysis of AND-gate uncomputation.","tokens_in":4488,"tokens_out":4675,"would_cite":true,"duration_ms":39796,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"A new circuit performs in-place addition of a classical n-bit constant to a quantum register with only n-3 ancilla qubits and 4n-5 T-gates, beating all previously known constructions built from quantum adders.","keywords":["quantum adder","addition by constant","T-count","ancilla qubits","temporary logical-AND gate","Toffoli gate","quantum incrementer","Clifford+T"],"falsifier":"Count the T-gates in the controlled circuit for a specific n (for example n=5) using the provided implementation: if the total exceeds 11n-15, the assumed Clifford corrections are not all Clifford in the controlled setting. Equivalently, compile the circuit for a small n with a tool that resolves the measurement-based AND uncomputations and count the resulting Toffoli and AND pairs.","tokens_in":3491,"feed_emoji":"⚛️","tokens_out":11018,"duration_ms":82258,"temperature":0.7,"pith_summary":"This paper proposes a new quantum circuit for adding a classical n-bit constant to an n-qubit register in-place, modulo 2^n. The constructed circuit uses n-3 ancilla qubits and a T-count of 4n-5, and the controlled version uses n-2 ancillas with a T-count of 11n-15. The construction generalizes an existing quantum incrementer circuit and replaces Toffoli gates with pairs of temporary-logical AND gates to cut T-count. The author claims this is more efficient than any adder-by-constant built from any known quantum-by-quantum adder, the standard approach for this problem. If correct, the result lowers the resource cost of a common arithmetic subroutine used in Shor's algorithm and other quantum computations.","feed_headline":"Quantum constant adder uses n-3 ancillas and 4n-5 T-gates","feed_subtitle":"A generalization of the quantum incrementer cuts the cost of in-place constant addition for Shor's algorithm.","key_machinery":"The circuit is built on three components. First, a classically controlled X gate, denoted X_a(q), applies X only when the classical bit a equals 1, so the constant is embedded for free with no T-gates. Second, the temporary logical-AND gate, which behaves like a Toffoli but requires the target to be in the zero state, is paired with its adjoint AND-dagger for uncomputation; a pair costs 4 T-gates instead of the 7 needed for one Toffoli. Third, the majority gate MAJ(x,y,a)=floor((x+y+a)/2), implemented with one Toffoli and five classically controlled X gates, propagates carry bits through the standard carry recurrence. The circuit is then optimized by cancelling adjacent inverse blocks (the same pattern used to build a quantum incrementer), merging and eliminating classically controlled X gates, using b0 directly as the first carry to drop one ancilla, and pushing the final Toffoli onto an output qubit to drop another. The uncontrolled version therefore contains n-3 AND-pairs and one Toffoli, giving a T-count of 4(n-3)+7=4n-5. The controlled version skips the final-ancilla elimination to avoid a doubly-controlled gate, so it contains n-2 AND-pairs and n-1 Toffolis, giving 4(n-2)+7(n-1)=11n-15.","core_discovery":"The paper's central claim is that in-place addition of a classical constant to a quantum register does not require the usual overhead of loading the constant into n ancilla qubits and then running a quantum-quantum adder. Instead, by generalizing a quantum incrementer, it constructs a circuit that directly computes the carries using majority (MAJ) gates, cancels adjacent adjoint blocks, and replaces every Toffoli that targets an ancilla initially in the zero state with a temporary-logical AND/AND-dagger pair. The resulting uncontrolled circuit has exactly n-3 ancillas and a T-count of 4n-5; the controlled version, which keeps one extra ancilla to avoid needing a doubly-controlled gate, has n-2 ancillas and a T-count of 11n-15. The author states that, to his knowledge, this is more efficient than any circuit constructed using any known quantum-by-quantum adder, and provides an implementation of both circuits.","pith_inferences":["The same carry-recurrence simplification could be applied to other arithmetic circuits, such as comparison-by-constant circuits, which the author mentions implementing alongside this work, potentially lowering their T-count similarly.","The paper counts T-gates but not T-depth or the cost of logical-level measurement and feedforward needed for the measurement-based AND uncomputation; in an error-corrected setting the relative advantage over other adders may shift.","The claimed efficiency is relative to adders-by-constants built by loading the constant into ancillas; circuits based on the QFT approach trade T-count for rotation precision and are not included in the comparison, so the optimal choice may depend on the fault-tolerance budget.","A direct test of the controlled circuit's T-count for several small n values would show whether any correction in the AND-uncomputation becomes a Toffoli when controlled, refining the 11n-15 formula if that assumption fails."],"forward_implications":["Any quantum routine that uses adder-by-constant as a subroutine, notably modular exponentiation in Shor's algorithm, can replace the standard n-ancilla construction with this circuit and thereby reduce ancilla overhead to n-3 while keeping the T-count lower.","The T-count gap over the compared ripple-carry adders grows linearly: for the Cuccaro adder the reduction is from 14n-21 to 4n-5, a saving of 10n-16 T-gates.","The controlled version gives a cheaper way to compute conditional constant additions, which appear in modular exponentiation and in many controlled arithmetic operations.","Because the circuit uses only Clifford+T gates and avoids the arbitrary rotations of QFT-based adders, it fits naturally into fault-tolerant implementations.","The included implementation lets a user verify the exact T-count and ancilla count for any n by direct gate counting."],"supporting_citations":[{"why":"Supplies the incrementer construction that the new adder generalizes to arbitrary constants.","marker":"[9]"},{"why":"Introduces the temporary logical-AND gate whose pair T-count of 4 drives the circuit's T-count reduction.","marker":"[5]"},{"why":"Baseline ripple-carry adder used for the cost comparison.","marker":"[2]"},{"why":"Baseline carry-lookahead adder used for the cost comparison.","marker":"[3]"},{"why":"Baseline Takahashi ripple-carry adder used for the cost comparison.","marker":"[4]"},{"why":"Motivates the adder-by-constant problem through modular exponentiation in Shor's algorithm.","marker":"[1]"}],"fun_headline_variants":["Constant adder cuts T-count to 4n-5 with n-3 ancillas","In-place quantum constant adder uses n-3 ancillas only","Generalized incrementer yields cheap constant adder for Shor","Controlled constant adder needs n-2 ancillas, 11n-15 T gates"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The controlled circuit's stated T-count of 11n-15 assumes that every correction gate produced by the uncomputation of the temporary AND gates remains a simple Clifford gate once the whole circuit is made controlled; if one of those corrections is a single-qubit flip on an input/output qubit, it turns into an additional Toffoli gate and raises the T-count.","fun_headline_variants_meta":{"raw":{"variants":["Constant adder cuts T-count to 4n-5 with n-3 ancillas","In-place quantum constant adder uses n-3 ancillas only","Generalized incrementer yields cheap constant adder for Shor","Controlled constant adder needs n-2 ancillas, 11n-15 T gates"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000225,"raw_usage":{"total_tokens":1392,"prompt_tokens":801,"completion_tokens":591,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":417,"completion_tokens_details":{"reasoning_tokens":510}},"tokens_in":417,"tokens_out":591,"duration_ms":5928,"temperature":1.0,"reasoning_tokens":510,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-10T20:50:03.867713+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Count the T-gates in the controlled circuit for a specific n (for example n=5) using the provided implementation: if the total exceeds 11n-15, the assumed Clifford corrections are not all Clifford in the controlled setting. Equivalently, compile the circuit for a small n with a tool that resolves the measurement-based AND uncomputations and count the resulting Toffoli and AND pairs.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the incrementer construction that the new adder generalizes to arbitrary constants."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Introduces the temporary logical-AND gate whose pair T-count of 4 drives the circuit's T-count reduction."},{"cited_title":"New Circuit for Quantum Adder by Constant","cited_arxiv_id":"2501.07060","evidence_quote":"Baseline ripple-carry adder used for the cost comparison."},{"cited_title":"The basic idea First, note that if a is even, then a = 2z · a′ where a′ is odd, so the task is reduced to adding number a′ to n − z most significant bits of b","cited_arxiv_id":null,"evidence_quote":"Baseline carry-lookahead adder used for the cost comparison."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Baseline Takahashi ripple-carry adder used for the cost comparison."},{"cited_title":"In particular, it is used to implement a modular exponentiation circuit for the Shor algorithm [1]","cited_arxiv_id":null,"evidence_quote":"Motivates the adder-by-constant problem through modular exponentiation in Shor's algorithm."}],"review_version":1}