Joel Seiferas - Academia.edu (original) (raw)
Papers by Joel Seiferas
Symposium on the Theory of Computing, 1981
Any string-matching algorithm requires at least linear time and a constant number of local storag... more Any string-matching algorithm requires at least linear time and a constant number of local storage locations. We design and analyze an algorithm which realizes both asymptotic bounds simultaneously. This can be viewed as completely eliminating the need for the tabulated "failure function" in the linear-time algorithm of Knutb, Morris, and Pratt. It makes possible a completely general implementation as a Fortran subroutine or even as a six-head finite automaton.
... As in the Knuth-Morris-Pratt algori-thm, the text need not be "backed up" or stored... more ... As in the Knuth-Morris-Pratt algori-thm, the text need not be "backed up" or stored at all. ... Like the slow naive algori-thm, this faster algorithm can be implemented as a FOR-TRAN subroutine which receives the text only sequential-ly, say from a card reader. ...
ABSTRACT The marginal utility of the Turing machine computational resources running time and stor... more ABSTRACT The marginal utility of the Turing machine computational resources running time and storage space are studied. A technique is developed which, unlike diagonalization, applies equally well to nondeterministic and deterministic automata. For f, g time or space bounding functions with f(n + 1) small compared to g(n), it is shown that, in terms of word length n, there are languages which are accepted by Turing machines operating with time or space g(n) but which are accepted by no Turing machine operating within time or space f(n). The proof involves use of the recursion theorem together with "padding" or "translational" techniques of formal language theory. Relations between worktape alphabet size, number of worktape heads, number of input heads, and Turing machine storage space are established. Within every common subexponential space bound, it is shown that enlarging the worktape alphabet always increases computing power. A hierarchy of two-way multihead finite automata is obtained even in the nondeterministic case. Results that are only slightly weaker are obtained for Turing machines that accept only languages over a one-letter alphabet. {PB 236-777.AS}
Journal of the ACM, 1978
I > 1} be the set of all nontnvial pahndromes over X A hneartime on-hne recogmtmn algorithm is pr... more I > 1} be the set of all nontnvial pahndromes over X A hneartime on-hne recogmtmn algorithm is presented for P~ ("palstar") on a random-access machine with addmon and umform cost criterion Also presented are a hnear-tlme on-line recognmon algorithm for P~ on a muitltape Turmg machine and a recognition algorithm for Pt 2 on a two-way deterministic pushdown automaton. The correctness of these algorithms is based on new "cancellation iemmas" for the languages P~
Page 1. OBSERVATIONS ON NONDETERMINISTIC MULTIDIMENSIONAL ITERATIVE ARRAYS t Joel I. Seiferas Mas... more Page 1. OBSERVATIONS ON NONDETERMINISTIC MULTIDIMENSIONAL ITERATIVE ARRAYS t Joel I. Seiferas Massachusetts Institute of Technology Cambridge, Massachusetts I. Introduction Let NIA(d) be the family of languages ...
Theoretical Computer Science, 1981
We report a linear-time string-matching algorithm for a random-access machine without dynamic sto... more We report a linear-time string-matching algorithm for a random-access machine without dynamic storage allocation. To do this, we tell how to adapt a cited algorithm to fill its dynamic storage needs by temporarily borrowing some of the space occupied by the input pattern. In automata-theoretic terms, we tell how to adapt the cited algorithm to run on a writing multihead finite automaton with a restricted writing alphabet.
Any string-matching algorithm requires at least linear time and a constant number of local storag... more Any string-matching algorithm requires at least linear time and a constant number of local storage locations. We design and analyze an algorithm which realizes both asymptotic bounds simultaneously. This can be viewed as completely eliminating the need for the tabulated "failure function" in the linear-time algorithm of Knutb, Morris, and Pratt. It makes possible a completely general implementation as a Fortran subroutine or even as a six-head finite automaton.
SIAM Journal on Computing, Aug 1, 1978
Correcting Counter-Automaton-Recognizable Languages. [SIAM Journal on Computing 7, 357 (1978)]. R... more Correcting Counter-Automaton-Recognizable Languages. [SIAM Journal on Computing 7, 357 (1978)]. Robert A. Wagner, Joel I. Seiferas. Abstract. Correction of a string xxx into a language LLL is the problem of finding a string yinLy \in LyinL to which xxx can be edited at least cost. ...
Journal of Computer and System Sciences, Jun 1, 1983
Any string-matching algorithm requires at least linear time and a constant number of local storag... more Any string-matching algorithm requires at least linear time and a constant number of local storage locations. We design and analyze an algorithm which realizes both asymptotic bounds simultaneously. This can be viewed as completely eliminating the need for the tabulated "failure function" in the linear-time algorithm of Knutb, Morris, and Pratt. It makes possible a completely general implementation as a Fortran subroutine or even as a six-head finite automaton.
Journal of Computer and System Sciences, Feb 1, 1977
Diagonalization, cardinality, and recursive padding arguments are used to separate the Turing mac... more Diagonalization, cardinality, and recursive padding arguments are used to separate the Turing machine space complexity classes obtained by bounding space, number of worktape symbols, and number of worktape heads. Witness languages over a one-letter alphabet are constructed when possible.
SIAM Journal on Computing, Sep 1, 1977
Linear-Time Computation by Nondeterministic Multidimensional Iterative Arrays. [SIAM Journal on C... more Linear-Time Computation by Nondeterministic Multidimensional Iterative Arrays. [SIAM Journal on Computing 6, 487 (1977)]. Joel I. Seiferas. Abstract. It is shown by simulation that every language accepted within time ndn^dnd by ...
Journal of Computer and System Sciences, Feb 1, 1977
Refined Turing machine space complexity classes are defined by limiting all three of the resource... more Refined Turing machine space complexity classes are defined by limiting all three of the resources space, worktape alphabet size, and number of worktape heads. Containment relations among the classes are proved and used to convert a few basic noncontainment relations to noncontainment relations of four more homogeneous types: "less space" (conventional classes), "fewer worktape symbols," "one fewer worktape symbol," and "fewer worktape heads." Noncontainment results for both nondeterrninistic and deterministic classes of both binary and unary languages are obtained. One corollary is a unary language hierarchy theorem for two-way multihead finite automata.
... Zvi Galil Computer Sciences Department IBM Thomas J. Watson Research Center Yorktown Heights,... more ... Zvi Galil Computer Sciences Department IBM Thomas J. Watson Research Center Yorktown Heights, New York 10598 Joel Seiferas Computer Science Department The Pennsylvania State University University Park, Pennsylvania 16802 so that Define a function ...
SIAM Journal on Computing, May 1, 1980
Saving Space in Fast String-Matching. [SIAM Journal on Computing 9, 417 (1980)]. Zvi Galil,Joel S... more Saving Space in Fast String-Matching. [SIAM Journal on Computing 9, 417 (1980)]. Zvi Galil,Joel Seiferas. Abstract. The string-matching problem is to find all instances (as contiguous substrings) of a pattern character string xxx in a longer text" string . ...
17th Annual Symposium on Foundations of Computer Science (sfcs 1976), 1976
... Zvi Galil Computer Sciences Department IBM Thomas J. Watson Research Center Yorktown Heights,... more ... Zvi Galil Computer Sciences Department IBM Thomas J. Watson Research Center Yorktown Heights, New York 10598 Joel Seiferas Computer Science Department The Pennsylvania State University University Park, Pennsylvania 16802 so that Define a function ...
Proceedings of the sixth annual ACM symposium on Theory of computing - STOC '74, 1974
Page 1. OBSERVATIONS ON NONDETERMINISTIC MULTIDIMENSIONAL ITERATIVE ARRAYS t Joel I. Seiferas Mas... more Page 1. OBSERVATIONS ON NONDETERMINISTIC MULTIDIMENSIONAL ITERATIVE ARRAYS t Joel I. Seiferas Massachusetts Institute of Technology Cambridge, Massachusetts I. Introduction Let NIA(d) be the family of languages ...
ACM SIGACT News, 1974
Hopcroft and Ullman (problem 3.10 [1]) pose the amusing question of whether the "first third... more Hopcroft and Ullman (problem 3.10 [1]) pose the amusing question of whether the "first third" of a regular language L, FIRST-THIRD(L) = {x| x is a prefix of a member of L of length 3|x|}, is necessarily regular. To see that it is, we can adapt of 1-way deterministic finite-state acceptor (an FA, for short) for L to get a 2-way non-deterministic finite-state acceptor with endmarkers for FIRST-THIRD(L). This acceptor behaves like the FA on x until it reaches the right endmarker, and then it uses another pass over x at half speed to behave like the FA on some nondeterministically chosen continuation of length 2|x|. That such an acceptor accepts a regular language follows from an argument similar to that of Shepherdson for deterministic acceptors [2].
Mathematical Systems Theory, 1977
01977 by SpĒinlp~-Vm'la= New Yock Inc. ... Real-Time Recognition of Substring Repetition and... more 01977 by SpĒinlp~-Vm'la= New Yock Inc. ... Real-Time Recognition of Substring Repetition and Reversal* ... Computer Science Department, The Pennsylvania State University, University Park, Pennsylvania 16802; Department of Mathematical Sciences, Computer Science ...
Proceedings of the thirteenth annual ACM symposium on Theory of computing - STOC '81, 1981
Any string-matching algorithm requires at least linear time and a constant number of local storag... more Any string-matching algorithm requires at least linear time and a constant number of local storage locations. We design and analyze an algorithm which realizes both asymptotic bounds simultaneously. This can be viewed as completely eliminating the need for the tabulated "failure function" in the linear-time algorithm of Knutb, Morris, and Pratt. It makes possible a completely general implementation as a Fortran subroutine or even as a six-head finite automaton.
Symposium on the Theory of Computing, 1981
Any string-matching algorithm requires at least linear time and a constant number of local storag... more Any string-matching algorithm requires at least linear time and a constant number of local storage locations. We design and analyze an algorithm which realizes both asymptotic bounds simultaneously. This can be viewed as completely eliminating the need for the tabulated "failure function" in the linear-time algorithm of Knutb, Morris, and Pratt. It makes possible a completely general implementation as a Fortran subroutine or even as a six-head finite automaton.
... As in the Knuth-Morris-Pratt algori-thm, the text need not be "backed up" or stored... more ... As in the Knuth-Morris-Pratt algori-thm, the text need not be "backed up" or stored at all. ... Like the slow naive algori-thm, this faster algorithm can be implemented as a FOR-TRAN subroutine which receives the text only sequential-ly, say from a card reader. ...
ABSTRACT The marginal utility of the Turing machine computational resources running time and stor... more ABSTRACT The marginal utility of the Turing machine computational resources running time and storage space are studied. A technique is developed which, unlike diagonalization, applies equally well to nondeterministic and deterministic automata. For f, g time or space bounding functions with f(n + 1) small compared to g(n), it is shown that, in terms of word length n, there are languages which are accepted by Turing machines operating with time or space g(n) but which are accepted by no Turing machine operating within time or space f(n). The proof involves use of the recursion theorem together with "padding" or "translational" techniques of formal language theory. Relations between worktape alphabet size, number of worktape heads, number of input heads, and Turing machine storage space are established. Within every common subexponential space bound, it is shown that enlarging the worktape alphabet always increases computing power. A hierarchy of two-way multihead finite automata is obtained even in the nondeterministic case. Results that are only slightly weaker are obtained for Turing machines that accept only languages over a one-letter alphabet. {PB 236-777.AS}
Journal of the ACM, 1978
I > 1} be the set of all nontnvial pahndromes over X A hneartime on-hne recogmtmn algorithm is pr... more I > 1} be the set of all nontnvial pahndromes over X A hneartime on-hne recogmtmn algorithm is presented for P~ ("palstar") on a random-access machine with addmon and umform cost criterion Also presented are a hnear-tlme on-line recognmon algorithm for P~ on a muitltape Turmg machine and a recognition algorithm for Pt 2 on a two-way deterministic pushdown automaton. The correctness of these algorithms is based on new "cancellation iemmas" for the languages P~
Page 1. OBSERVATIONS ON NONDETERMINISTIC MULTIDIMENSIONAL ITERATIVE ARRAYS t Joel I. Seiferas Mas... more Page 1. OBSERVATIONS ON NONDETERMINISTIC MULTIDIMENSIONAL ITERATIVE ARRAYS t Joel I. Seiferas Massachusetts Institute of Technology Cambridge, Massachusetts I. Introduction Let NIA(d) be the family of languages ...
Theoretical Computer Science, 1981
We report a linear-time string-matching algorithm for a random-access machine without dynamic sto... more We report a linear-time string-matching algorithm for a random-access machine without dynamic storage allocation. To do this, we tell how to adapt a cited algorithm to fill its dynamic storage needs by temporarily borrowing some of the space occupied by the input pattern. In automata-theoretic terms, we tell how to adapt the cited algorithm to run on a writing multihead finite automaton with a restricted writing alphabet.
Any string-matching algorithm requires at least linear time and a constant number of local storag... more Any string-matching algorithm requires at least linear time and a constant number of local storage locations. We design and analyze an algorithm which realizes both asymptotic bounds simultaneously. This can be viewed as completely eliminating the need for the tabulated "failure function" in the linear-time algorithm of Knutb, Morris, and Pratt. It makes possible a completely general implementation as a Fortran subroutine or even as a six-head finite automaton.
SIAM Journal on Computing, Aug 1, 1978
Correcting Counter-Automaton-Recognizable Languages. [SIAM Journal on Computing 7, 357 (1978)]. R... more Correcting Counter-Automaton-Recognizable Languages. [SIAM Journal on Computing 7, 357 (1978)]. Robert A. Wagner, Joel I. Seiferas. Abstract. Correction of a string xxx into a language LLL is the problem of finding a string yinLy \in LyinL to which xxx can be edited at least cost. ...
Journal of Computer and System Sciences, Jun 1, 1983
Any string-matching algorithm requires at least linear time and a constant number of local storag... more Any string-matching algorithm requires at least linear time and a constant number of local storage locations. We design and analyze an algorithm which realizes both asymptotic bounds simultaneously. This can be viewed as completely eliminating the need for the tabulated "failure function" in the linear-time algorithm of Knutb, Morris, and Pratt. It makes possible a completely general implementation as a Fortran subroutine or even as a six-head finite automaton.
Journal of Computer and System Sciences, Feb 1, 1977
Diagonalization, cardinality, and recursive padding arguments are used to separate the Turing mac... more Diagonalization, cardinality, and recursive padding arguments are used to separate the Turing machine space complexity classes obtained by bounding space, number of worktape symbols, and number of worktape heads. Witness languages over a one-letter alphabet are constructed when possible.
SIAM Journal on Computing, Sep 1, 1977
Linear-Time Computation by Nondeterministic Multidimensional Iterative Arrays. [SIAM Journal on C... more Linear-Time Computation by Nondeterministic Multidimensional Iterative Arrays. [SIAM Journal on Computing 6, 487 (1977)]. Joel I. Seiferas. Abstract. It is shown by simulation that every language accepted within time ndn^dnd by ...
Journal of Computer and System Sciences, Feb 1, 1977
Refined Turing machine space complexity classes are defined by limiting all three of the resource... more Refined Turing machine space complexity classes are defined by limiting all three of the resources space, worktape alphabet size, and number of worktape heads. Containment relations among the classes are proved and used to convert a few basic noncontainment relations to noncontainment relations of four more homogeneous types: "less space" (conventional classes), "fewer worktape symbols," "one fewer worktape symbol," and "fewer worktape heads." Noncontainment results for both nondeterrninistic and deterministic classes of both binary and unary languages are obtained. One corollary is a unary language hierarchy theorem for two-way multihead finite automata.
... Zvi Galil Computer Sciences Department IBM Thomas J. Watson Research Center Yorktown Heights,... more ... Zvi Galil Computer Sciences Department IBM Thomas J. Watson Research Center Yorktown Heights, New York 10598 Joel Seiferas Computer Science Department The Pennsylvania State University University Park, Pennsylvania 16802 so that Define a function ...
SIAM Journal on Computing, May 1, 1980
Saving Space in Fast String-Matching. [SIAM Journal on Computing 9, 417 (1980)]. Zvi Galil,Joel S... more Saving Space in Fast String-Matching. [SIAM Journal on Computing 9, 417 (1980)]. Zvi Galil,Joel Seiferas. Abstract. The string-matching problem is to find all instances (as contiguous substrings) of a pattern character string xxx in a longer text" string . ...
17th Annual Symposium on Foundations of Computer Science (sfcs 1976), 1976
... Zvi Galil Computer Sciences Department IBM Thomas J. Watson Research Center Yorktown Heights,... more ... Zvi Galil Computer Sciences Department IBM Thomas J. Watson Research Center Yorktown Heights, New York 10598 Joel Seiferas Computer Science Department The Pennsylvania State University University Park, Pennsylvania 16802 so that Define a function ...
Proceedings of the sixth annual ACM symposium on Theory of computing - STOC '74, 1974
Page 1. OBSERVATIONS ON NONDETERMINISTIC MULTIDIMENSIONAL ITERATIVE ARRAYS t Joel I. Seiferas Mas... more Page 1. OBSERVATIONS ON NONDETERMINISTIC MULTIDIMENSIONAL ITERATIVE ARRAYS t Joel I. Seiferas Massachusetts Institute of Technology Cambridge, Massachusetts I. Introduction Let NIA(d) be the family of languages ...
ACM SIGACT News, 1974
Hopcroft and Ullman (problem 3.10 [1]) pose the amusing question of whether the "first third... more Hopcroft and Ullman (problem 3.10 [1]) pose the amusing question of whether the "first third" of a regular language L, FIRST-THIRD(L) = {x| x is a prefix of a member of L of length 3|x|}, is necessarily regular. To see that it is, we can adapt of 1-way deterministic finite-state acceptor (an FA, for short) for L to get a 2-way non-deterministic finite-state acceptor with endmarkers for FIRST-THIRD(L). This acceptor behaves like the FA on x until it reaches the right endmarker, and then it uses another pass over x at half speed to behave like the FA on some nondeterministically chosen continuation of length 2|x|. That such an acceptor accepts a regular language follows from an argument similar to that of Shepherdson for deterministic acceptors [2].
Mathematical Systems Theory, 1977
01977 by SpĒinlp~-Vm'la= New Yock Inc. ... Real-Time Recognition of Substring Repetition and... more 01977 by SpĒinlp~-Vm'la= New Yock Inc. ... Real-Time Recognition of Substring Repetition and Reversal* ... Computer Science Department, The Pennsylvania State University, University Park, Pennsylvania 16802; Department of Mathematical Sciences, Computer Science ...
Proceedings of the thirteenth annual ACM symposium on Theory of computing - STOC '81, 1981
Any string-matching algorithm requires at least linear time and a constant number of local storag... more Any string-matching algorithm requires at least linear time and a constant number of local storage locations. We design and analyze an algorithm which realizes both asymptotic bounds simultaneously. This can be viewed as completely eliminating the need for the tabulated "failure function" in the linear-time algorithm of Knutb, Morris, and Pratt. It makes possible a completely general implementation as a Fortran subroutine or even as a six-head finite automaton.