Abstract
We consider the methodology for proving lower bounds on the query complexity of property testing via communication complexity, which was put forward by Blais, Brody, and Matulef (Computational Complexity, 2012). They provided a restricted formulation of their methodology (via “simple combining operators”) and also hinted towards a more general formulation, which we spell out in this paper.
A special case of the general formulation proceeds as follows: In order to derive a lower bound on testing the property \(\varPi \), one presents a mapping F of pairs of inputs \((x,y)\in \{0,1\}^{n+n}\) for a two-party communication problem \(\varPsi \) to \(\ell (n)\)-bit long inputs for \(\varPi \) such that \((x,y)\in \varPsi \) implies \(F(x,y)\in \varPi \) and \((x,y)\not \in \varPsi \) implies that F(x, y) is far from \(\varPi \). Let \(f_i(x,y)\) be the \(i^\mathrm{th}\) bit of F(x, y), and suppose that B is an upper bound on the (deterministic) communication complexity of each \(f_i\) and that C is a lower bound on the randomized communication complexity of \(\varPsi \). Then, testing \(\varPi \) requires at least C/B queries.
The foregoing formulation is generalized by considering randomized protocols (with small error) for computing the \(f_i\)’s. In contrast, the restricted formulation (via “simple combining operators”) requires that each \(f_i(x,y)\) be a function of \(x_i\) and \(y_i\) only, and uses \(B=2\) for the straightforward computation of \(f_i\).
We show that the general formulation cannot yield significantly stronger lower bounds than those that can be obtained by the restricted formulation. Nevertheless, we advocate the use of the general formulation, because we believe that it is easier to work with. Following Blais et al., we also describe a version of the methodology for nonadaptive testers and one-way communication complexity.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Notes
- 1.
- 2.
Added in revision: This sentence reflects the state of affairs at the time this paper was originally written. At that time, we mentioned that “A notable exception has been provided by the recent work of [5]: To streamline their proof, they take a move that is analogous to ours by replacing the simple combining operators of [4, Def. 2.3] with a ‘one-bit one-way combining operator’ (see Definition 2.4 and Lemma 2.5 in [5]).”
- 3.
Jumping ahead, we note that, with respect to the general formulation, little is lost by considering only the binary representation.
- 4.
Note that \(\mathtt{CC}_0(\cdot )\) is different from the standard notion of zero-error randomized communication complexity, since in the latter one considers the expected number of bits exchanged on the worst-case pair of inputs (whereas we considered the worst-case over both the shared randomness and the pair of inputs). Note that the difference between the expected complexity and the worst-case complexity is not very significant in the case of \(\varTheta (1)\)-error communication complexity, but it is crucial in the case of zero-error.
- 5.
In that case, \(f_i:\varSigma ^{2n}\rightarrow \varSigma \), and simple combining operators correspond to the case that each \(f_i(x,y)\) depends only on the \(i^\mathrm{th}\) symbol of x and the \(i^\mathrm{th}\) symbol of y. The assertion then is that \(\mathtt{Q}_\eta (\epsilon ,\varPi )\ge \mathtt{CC}_{\eta }(\varPsi )/2{\lceil \log _2|\varSigma |\rceil }\).
- 6.
- 7.
The non-appealing aspect of \(C'\) is that it encodes each of the two parts of the message separately. The appealing aspect is that these two parts can be nicely related; for example, consider \((P_n,S_n)=(\{0,1\}^{2n},f^{-1}(1))\) such that \(f:\{0,1\}^{2n}\rightarrow \{0,1\}\) is the inner product (mod 2) function (and recall that the communication complexity of \((P_n,S_n)\) (equiv., of computing f) is \(\varOmega (n)\) [7]).
- 8.
Formally, a nonadaptive tester T consists of a pair of algorithms, Q and D, which use the same randomness, such that Q determines the tester’s query and D its decision given the corresponding answers; that is, \(T^z(r)=D(r,z_{i_1},...,z_{i_q})\), where \((i_1,...,i_q)=Q(r)\). In our one-way communication protocol each of the two parties locally determines \((i_1,...,i_q)=Q(r_0)\), then, for each \(j\in [q]\), the first party sends to the second party the message required for the computation of \(f_{i_j}(x,y)\), and finally (after computing all the \(f_{i_j}(x,y)\)’s) the second party invokes D and outputs \(D(r_0,f_{i_1}(x,y),...,f_{i_q}(x,y))\).
- 9.
It is crucial that the two input-holding parties have access to the same shared randomness, since they cannot communicate with one another. In contrast, it is less essential that the referee also has access to this shared randomness, since one of the input-holding parties can send it along while relying on the fact that the randomness can be made logarithmic in the input length (cf. [19, Thm. 3.14]).
- 10.
The point is that in this case we cannot afford any error, regardless of the value of \(f_i(x,y)\).
- 11.
In that application, we derive a lower bound on \(\mathtt{Q}_{\eta }(\varOmega (\epsilon ),\varPi )\). This lower bound is smaller than \(\mathtt{CC}_{\eta }(\varPsi )-{\widetilde{O}}(1/\epsilon )\), which in turn is negative in the case of \(\epsilon \le \mathrm{poly}(\log n)/n\).
- 12.
Recall that this means that \(\varPsi =(P,S)\) is a promise problem such that \(P,S\subseteq \{0,1\}^{2n}\), that \(\varPi \subseteq \{0,1\}^\ell \) is a property, and that the mapping \(F:\{0,1\}^{2n}\rightarrow \{0,1\}^\ell \) satisfies the following two conditions:
(1) For every \((x,y)\in P\cap S\), it holds that \(F(x,y)\in \varPi \).
(2) For every \((x,y)\in P\setminus S\), it holds that F(x, y) is \(\epsilon \)-far from \(\varPi \).
Lastly, \(f_i(x,y)\) denotes the \(i^\mathrm{th}\) bit of F(x, y).
- 13.
We also use the fact that \(x\ge y\) implies \(\frac{x-\beta }{\log (x/\eta )}\ge \frac{y-\beta }{O(\log (y/\eta ))}\), provided that \(x\ge (1+\varOmega (1))\cdot \beta \), which we may assume (since otherwise the bound is quite useless).
- 14.
Actually, \(\mathtt{CC}_0(f_i)\le (B/2)+1\), via the straightforward protocol in which the first party sends \(v\leftarrow g_i(x)\) to the second party, who replies with the value of \(f_i(x,y)\) that is computed based on v and \(h_i(y)\).
- 15.
The following statement holds under some simplifying assumptions that are listed at the beginning of the proof. Enforcing these assumptions causes an insignificant deterioration in some parameters (i.e., \(B\leftarrow B+3\) and \(\mathtt{Q}_\eta (\epsilon ',\varPi )\) is replaced by \(\mathtt{Q}_\eta (\epsilon ',\varPi )+{\widetilde{O}}(1/\epsilon ')\).
- 16.
Note that if a string is \(\epsilon \)-far from \(\varPi ^{(m)}\), then either the first block is \(\epsilon \)-far from \(\varPi \), or the other blocks are \(\epsilon \)-far from a repetition of the first block. In the first case the original tester rejects (w.p. at least \(1-\eta \)), and in the second case the repetition test rejects (w.p. at least \(1-\eta \)).
- 17.
Specifically, we refer to the use of the encoding of the parties’ strategies by suitable error correcting codes.
- 18.
It is standard to assume that the parties interact by sending single-bit messages and that the first party starts. In such a case, \(A_i^x\) will be defined for strings of length at most \((B-1)/2\), including the empty string, while \(B_i^y\) will be defined for \(\bigcup _{j\in [B/2]}\{0,1\}^j\). In general, the situation may be more complex, but in all cases the length of the description of each of the two strategies is at most \(2^{B-1}\).
- 19.
This uses the assumption that \({\mathtt{P}}:\{0,1\}^{2^B}\rightarrow \{0,1\}\) is onto, which can be justified as in Sect. 7.1. Specifically, we can modify \(A_i^x\) so that it always starts by sending the bit 1, and let \({\mathtt{P}}(a,b)=\sigma \) if \(a(\lambda )=0\) and \(b(1)=\sigma \).
- 20.
The issue is not the specific low level of error, but rather that we have to bound the error away from 1/2 so that we can effectively determine what bit is produced (with probability higher than 1/2) by a pair of residual randomized strategies.
- 21.
Indeed, we may assume (w.l.o.g., cf. [19, Thm. 3.14]) that \(\rho {\mathop {=}\limits ^\mathrm{def}}O(\log (n/\eta ))\), but this is not needed for our argument.
- 22.
Indeed, the straightforward method is to select a random sample of \(O(1/\epsilon ')\) indices \(I\subseteq [2\ell \cdot 2^\rho ]\) and performing an \((\epsilon '/4)\)-test (with error probability \(\mathrm{poly}(\epsilon ')\)) on \(z_i\) for each \(i\in I\).
- 23.
Since \(\epsilon '={\widetilde{\varOmega }}(1/n)\), we do not expect to see pairs that produces the opposite value, which is quite rare (i.e., appears in at most a \(\eta /n\) fraction of the pairs).
- 24.
Note that we may also reject, with very small probability, due to encoutering pairs that produce different values (within a sequence of pairs that safely produces a value). But since the fraction of exceptional pairs is at most \(\eta /n\), this event occurs with very small probability.
- 25.
Indeed, this rough bound neglects the aforementioned additive terms, which are insignificant for constant \(\epsilon >0\).
- 26.
Instead, one reduces the original communication complexity problem to the auxiliary one, and then reduces the auxiliary communication problem to the property testing problem. The first reduction is performed within the setting of communication complexity, whereas the second reduction is the one in which simple combing operators are used.
- 27.
If only a designated subset of the parties obtains the output, then we can emulate only nonadaptive testers (as done in Sect. 5).
- 28.
The following problem differs from the one in [2] in two aspects. Firstly, the computational model is different (i.e., we consider the query complexity of property testing, whereas [2] refers to the space complexity of streaming algorithms). Secondly, the problems are different: We consider an error-correcting encoding (i.e., C(x)) of the information (i.e., x) to which the frequency measure is applied. We stress, however, that the lower bound is not due to the complexity of codeword testing, since codeword testing may be easy for \(\ell (n)=k(n)^{1+o(1)}\) (cf., e.g., [14]).
- 29.
Indeed, this follows from the proof of Theorem A.2, when setting \(m=2\), which correspond to the two-party case, and observing that no-instances are mapped to instances having norm at least \(m^c+(t-1)m>tm=k(n)\). Note that the same lower bound can be proved for \(\varPi '\), by padding the inputs to \(\mathrm DISJ\) with an adequate number of repetitions of some fixed symbol. Note that these arguments rely on the fact that testing \(\varPi \) (or \(\varPi '\)) requires distinguishing codewords that encode information (i.e., x) with a norm below some threshold from codewords that encode information with norm just above that threshold. In contrast, Theorem A.2 refers to a relaxation that captures an approximation of the corresponding norm, and a straightforward adaptation of the reduction from the two-party case does not seem to work here.
- 30.
The result of [2] is actually stronger, since it refers to the case that the no-instances consist of subsets that have pairwise intersections that all equal the same singleton.
- 31.
Actually, we need to generalize Theorem A.1 so that it applies to doubly-relaxed problems. Such a generalization is straightforward.
References
Alon, N., Dar, S., Parnas, M., Ron, D.: Testing of clustering. SIAM J. Discrete Math. 16(3), 393–417 (2003)
Alon, N., Matias, Y., Szegedy, M.: The space complexity of approximating the frequency moments. J. Comput. Syst. Sci. 58(1), 137–147 (1999)
Ben-Sasson, E., Harsha, P., Raskhodnikova, S.: Some 3CNF properties are hard to test. SIAM J. Comput. 35(1), 1–21 (2005)
Blais, E., Brody, J., Matulef, K.: Property testing lower bounds via communication complexity. Comput. Complex. 21(2), 311–358 (2012). https://doi.org/10.1007/s00037-012-0040-x
Blais, E., Raskhodnikova, S., Yaroslavtsev, G.: Lower bounds for testing properties of functions on hypergrid domains. In: ECCC, TR13-036, March 2013
Dodis, Y., Goldreich, O., Lehman, E., Raskhodnikova, S., Ron, D., Samorodnitsky, A.: Improved testing algorithms for monotonicity. In: Hochbaum, D.S., Jansen, K., Rolim, J.D.P., Sinclair, A. (eds.) APPROX/RANDOM 1999. LNCS, vol. 1671, pp. 97–108. Springer, Heidelberg (1999). https://doi.org/10.1007/978-3-540-48413-4_10
Chor, B., Goldreich, O.: Unbiased bits from sources of weak randomness and probabilistic communication complexity. SIAM J. Comput. 17(2), 230–261 (1988)
Goldreich, O. (ed.): Property Testing – Current Research and Surveys. LNCS, vol. 6390. Springer, Heidelberg (2010). https://doi.org/10.1007/978-3-642-16367-8
Goldreich, O.: Short locally testable codes and proofs: a survey in two parts. In: [8]
Goldreich, O.: On testing computability by small width OBDDs. In: Serna, M., Shaltiel, R., Jansen, K., Rolim, J. (eds.) APPROX/RANDOM 2010. LNCS, vol. 6302, pp. 574–587. Springer, Heidelberg (2010). https://doi.org/10.1007/978-3-642-15369-3_43
Goldreich, O.: Introduction to Property Testing. Cambridge University Press, Cambridge (2017)
Goldreich, O., Goldwasser, S., Ron, D.: Property testing and its connection to learning and approximation. J. ACM 45, 653–750 (1998). (Extended abstract in 37th FOCS, 1996)
Goldreich, O., Ron, D.: Property testing in bounded degree graphs. Algorithmica 32, 302–343 (2002). https://doi.org/10.1007/s00453-001-0078-7
Goldreich, O., Sudan, M.: Locally testable codes and PCPs of almost linear length. J. ACM 53(4), 558–655 (2006)
Gur, T., Rothblum, R.: Non-interactive proofs of proximity. In: ECCC, TR13-078, May 2013
Kalyanasundaram, B., Schintger, G.: The probabilistic communication complexity of set intersection. SIAM J. Discrete Math. 5(4), 545–557 (1992)
Katz, J., Trevisan, L.: On the efficiency of local decoding procedures for error-correcting codes. In: Proceedings of the 32nd ACM Symposium on the Theory of Computing, pp. 80–86 (2000)
Kearns, M., Ron, D.: Testing problems with sub-learning sample complexity. J. Comput. Syst. Sci. 61(3), 428–456 (2000)
Kushilevitz, E., Nisan, N.: Communication Complexity. Cambridge University Press, Cambridge (1997)
Parnas, M., Ron, D.: Testing the diameter of graphs. Random Struct. Algorithms 20(2), 165–183 (2002)
Ron, D.: Property testing: a learning theory perspective. Found. Trends Mach. Learn. 1(3), 307–402 (2008)
Ron, D.: Algorithmic and analysis techniques in property testing. Found. Trends TCS 5(2), 73–205 (2009)
Rubinfeld, R., Sudan, M.: Robust characterization of polynomials with applications to program testing. SIAM J. Comput. 25(2), 252–271 (1996)
Acknowledgments
Part of this work is based on joint research with Dana Ron, who refused to co-author it. The constructions presented in Sect. 4 are partially inspired by some constructions in [15]. We are grateful to David Woodruff for suggesting Theorem 5.2 and allowing us to present it here. Ditto with respect to Tom Gur for Theorem A.2. We thank Eric Blais and Sofya Raskhodnikova for calling our attention to [5]. Lastly, we thank Tom Gur and Ron Rothblum for comments on an early draft of this paper.
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Appendix: Generalization to Multi-Party Communication
Appendix: Generalization to Multi-Party Communication
The formulation presented in Sect. 3 generalizes easily to the model of multi-party communication. The treatment is quite oblivious of the details of the model; for example, it does not matter if one considers the standard model of “input on the forehead” or to the more natural model in which each party gets a part of the input (with no overlap). (These variations can be captured by the promise problems that the parties wish to solve.) The exact way in which the parties communicate is also not crucial, at least as long as the number of parties (denoted m) is small. For simplicity, we consider here a broadcast model, where in each communication round there is a single designated sender (determined by the transcript of the communication so far).
In light of the above, we consider m-party communication protocols in which the local input of the \(j^\mathrm{th}\) party is denoted \(x^{(j)}\). Let \({\langle {A^{(1)}(x^{(1)}),...,A^{(m)}(x^{(m)})}\rangle }(r)\) denote the (joint) output of the m parties, when the \(j^\mathrm{th}\) party uses strategy \(A^{(j)}\) and gets input \(x^{(j)}\), and all parties have free access to the shared randomness r. Considering promise problems \(\varPsi =(P,S)\) such that \(P,S\subseteq \{0,1\}^{m\cdot n}\), Definition 2.1 extends naturally; that is, the \(\eta \) -error communication complexity of \(\varPsi \), denoted \(\mathtt{CC}_\eta (\varPsi )\), is the minimum communication complexity of all m-protocols that solve \(\varPsi \) with error at most \(\eta \).
Theorem A.1
(Theorem 3.1, generalized to m-party protocols): Let \(\varPsi =(P,S)\) be a promise problem such that \(P,S\subseteq \{0,1\}^{m\cdot n}\), and let \(\varPi \subseteq \{0,1\}^\ell \) be a property, and \(\epsilon ,\eta >0\). Suppose that the mapping \(F:\{0,1\}^{m\cdot n}\rightarrow \{0,1\}^\ell \) satisfies the following two conditions:
-
1.
For every \((x^{(1)},...,x^{(m)})\in P\cap S\), it holds that \(F(x^{(1)},...,x^{(m)})\in \varPi \).
-
2.
For every \((x^{(1)},...,x^{(m)})\in P\setminus S\), it holds that \(F(x^{(1)},...,x^{(m)})\) is \(\epsilon \)-far from \(\varPi \).
Then, \(\mathtt{Q}_\eta (\epsilon ,\varPi )\ge \mathtt{CC}_{2\eta }(\varPsi )/B\), where \(B=\max _{i\in [\ell ]}\{\mathtt{CC}_{\eta /n}(f_i)\}\) and \(f_i(x^{(1)},...,x^{(m)})\) is the \(i^\mathrm{th}\) bit of \(F(x^{(1)},...,x^{(m)})\). Furthermore, if \(B=\max _{i\in [\ell ]}\{\mathtt{CC}_{0}(f_i)\}\), then \(\mathtt{Q}_\eta (\epsilon ,\varPi )\ge \mathtt{CC}_{\eta }(\varPsi )/B\).
Theorem A.1 is proved by a straightforward generalization of the proof of Theorem 3.1; that is, we merely replace “two” by “m” (and everything goes through). We believe that this generalization further clarifies the ideas underlying the proof of Theorem 3.1 by presenting them in a slightly more abstract form.
Proof:
The following description applies to any communication model in which all parties obtain the output produced by the protocol. Given an \(\epsilon \)-tester with error \(\eta \) for \(\varPi \) and communication protocols for the \(f_i\)’s, we present a protocol for solving \(\varPsi \). The key idea is that, using their shared randomness, the parties (holding the inputs \(x^{(1)},...,x^{(m)}\), respectively) can emulate the execution of the \(\epsilon \)-tester, while providing it with virtual access to \(F(x^{(1)},...,x^{(m)})\). Specifically, when the tester queries the \(i^\mathrm{th}\) bit of the oracle, the parties provide it with the value of \(f_i(x^{(1)},...,x^{(m)})\) by first executing the corresponding communication protocol. Details follow.
The protocol for \(\varPsi \) proceeds as follows: On local input \(x^{(j)}\) and shared randomness \(r=(r_0,r_1,...,r_\ell )\in (\{0,1\}^*)^{\ell +1}\), the \(j^\mathrm{th}\) party invokes the \(\epsilon \)-tester on randomness \(r_0\), and answers the tester’s queries by interacting with the other parties. That is, each of the parties invokes a local copy of the tester’s program, but all copies are invoked on the same randomness, and are fed with identical answers to their (identical) queries. When the tester issues a query \(i\in [\ell ]\), the parties compute the value of \(f_i(x^{(1)},...,x^{(m)})\) by using the corresponding communication protocol, and feed \(f_i(x^{(1)},...,x^{(m)})\) to (their local copy of) the tester. Specifically, denoting the latter protocol (i.e., sequence of strategies) by \((A^{(1)}_i,...,A^{(m)}_i)\), the parties answer with \({\langle {A^{(1)}_i(x^{(1)}),...,A^{(m)}_i(x^{(m)})}\rangle }(r_i)\). When the tester halts, each party outputs the output it has obtained from (its local copy of) the tester.
We stress that the above description is oblivious to the details of the communication model, as long as in this model all parties obtain the output produced by the protocol.Footnote 27 Indeed, the description presented in the proof of Theorem 3.1 is merely a special case (which corresponds to the standard model of two-party computation), and the analysis of the general case (omitted here) is identical to the analysis of the special case presented in the proof of Theorem 3.1. \(\blacksquare \)
On the Potential Usefulness of the Generalization. Tom Gur has pointed out that the generalization to multi-party communication complexity allows additional flexibility for the design of reductions. To illustrate the point, he suggested the proof outlined below, which refers to a multi-party communication complexity model in which parties obtain non-overlapping inputs and communication is by individual point-to-point channels.
Theorem A.2
(a property testing (encoded) version of the frequency moment problem of [2]):Footnote 28 For \(k(n)=n/2\) and \(\ell (n)=n^{1+o(1)}\), let \(\mathcal{F}\) be a finite field of size n, and \(C:\mathcal{F}^{k(n)} \rightarrow \mathcal{F}^{\ell (n)}\) be a \(\mathcal{F}\)-linear code of constant relative distance, denoted \(\delta \). For any sequence \(x=(x_1,...,x_k)\in \mathcal{F}^k\) and \(v\in \mathcal{F}\), let \(\#_v(x)\) denote the number of occurrences of v in x; that is, \(\#_v(x)=|\{i\in [k]:x_i\!=\!v\}|\). For any constant \(c>1\), let
Then, distinguishing inputs in \(\varPi \) from inputs that are \(\delta \)-far from \(\varPi '\) requires \(\varOmega (\ell (n)^{1-(7/c)})\) queries.
Indeed, it follows that testing \(\varPi \) requires query complexity \(\varOmega (\ell (n)^{1-(7/c)})\), but this (and, in fact, a stronger \(\varOmega (n/\log n)\) lower bound) can be proved by reduction from a two-party communication complexity problem (i.e., \(\mathrm DISJ\)).Footnote 29 In contrast, Theorem A.2 refers to a doubly-relaxed decision problem, where one level of relaxation is the approximation of the norm (captured by the gap between \(\varPi \) and \(\varPi '\)) and the second is the standard property testing relaxation (captured by the gap between \(\varPi '\) and \(\delta \)-far from \(\varPi '\)). Such doubly-relaxed problems have been often considered in the property testing literature (cf., e.g., [1, 20]), starting with [18]. The following proof, which adapts a proof of [2] (which in turn refers to streaming algorithms), relies on a reduction from a multi-party communication problem. As is the case with its streaming original [2], it is not clear whether Theorem A.2 can be proved by reduction from a two-party communication problem.
Proof:
We shall use a reduction from the following multi-party communication problem, denoted (m, t)-\(\mathrm{DISJ}_n\). In this problem, there are m parties, each holding a t-subset of [n], and the problem is to distinguish the case that the subsets are pairwise disjoint from the case that the intersection of all subsets is non-empty. By [2], if \(n \ge 2mt-m+1\), then the communication complexity of (m, t)-\(\mathrm{DISJ}_n\) (in the point-to-point channels model) is \(\varOmega (t/m^3)\).Footnote 30
We set \(m=n^{1/c}\) and \(t=n/2m\) (so that \(n=2mt\)), and represent the input of the \(j^\mathrm{th}\) party by a sequence \(x^{(j)}\in \mathcal{F}^t\). Recall that \(|\mathcal{F}|=n\) and \(k(n)=n/2=mt\). Now, we let \(F(x^{(1)},...,x^{(m)}) = C(x^{(1)}\cdots x^{(m)})\), which equals \(\sum _{j\in [m]} C(0^{(j-1)t}x^{(j)}0^{(m-j)t})\) by the \(\mathcal{F}\)-linearity. Hence, each bit of \(F(x^{(1)},...,x^{(m)})\) can be computed (in this communication model) by communicating \(m^2\log _2n\) bits (i.e., each party sends a single field elements to each of the other parties). Note that if \(x=(x^{(1)},...,x^{(m)})\) is a yes-instance of (m, t)-\(\mathrm{DISJ}_n\) then \(\sum _{v\in \mathcal{F}}\#_v(x)^c = mt = k(n)\), since each element that occurs in x occurs in it exactly once (i.e., in one of the \(x^{(j)}\)’s), which means that \(F(x^{(1)},...,x^{(m)})\) is in \(\varPi \). On the other hand, if \((x^{(1)},...,x^{(m)})\) is a no-instance of (m, t)-\(\mathrm{DISJ}_n\) then \(\sum _{v\in \mathcal{F}}\#_v(x)^c > m^c = n = 2k(n)\), since at least one element occurs m times (i.e., in all the \(x^{(j)}\)’s), which means that \(F(x^{(1)},...,x^{(m)})\) is not in \(\varPi '\), and so it is \(\delta \)-far from any codeword in \(\varPi '\) (since it is itself a codeword).
Applying Theorem A.1,Footnote 31 it follows that the query complexity of the promise problem of distinguishing \(\varPi \) from the set of \(\ell (n)\)-long sequences that are \(\delta \)-far from \(\varPi '\) is lower bounded by \(\varOmega (t/m^3)/(m^2\log n)\), which equals \(\varOmega (n/(m^6\log n)) = \varOmega (n^{1-6(1+o(1))/c})\). Using \(n=\ell (n)^{1/(1+o(1))}\), the claim follows. \(\blacksquare \)
Rights and permissions
Copyright information
© 2020 Springer Nature Switzerland AG
About this chapter
Cite this chapter
Goldreich, O. (2020). On the Communication Complexity Methodology for Proving Lower Bounds on the Query Complexity of Property Testing. In: Goldreich, O. (eds) Computational Complexity and Property Testing. Lecture Notes in Computer Science(), vol 12050. Springer, Cham. https://doi.org/10.1007/978-3-030-43662-9_7
Download citation
DOI: https://doi.org/10.1007/978-3-030-43662-9_7
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-030-43661-2
Online ISBN: 978-3-030-43662-9
eBook Packages: Computer ScienceComputer Science (R0)