Filinski, A. 1996. Controlling eects. Ph.D. thesis, Carnegie Mellon University, School of Computer Science, Pittsburgh, Pennsylvania.  Home/Search   Document Not in Database   Summary   Related Articles   Check

.... infeasible to keep track of probabilities assigned to all events from an infinite discrete domain or a continuous domain, they provide only a binary choice construct (e1 orp e2 in [18] and choose p e1 e2 in [29] As a result, both languages essentially implement the probability monad in [6], which is capable of specifying only probability distributions over finite domains. In this paper, we propose a monadic probabilistic language capable of specifying probability distributions over infinite discrete domains and continuous domains. The mathematical basis for our language is ....

A. Filinski. Controlling E#ects. PhD thesis, School of Computer Science, Carnegie Mellon University, 1996.

....to various interpreters to derive corresponding combinators. For instance, we have already applied it to an interpreter of calculus [24] in ML, and successfully obtained concurrency combinators in ML. 1 We are now working to generalize these results by using the notion of monad reflection [14]. Besides that, the state based let insertion (in Chapter 4) suggests that continuationbased operation can be simulated by state based one in certain cases. It might also be interesting to study when and how such simulation is possible. 1 Interpreter without Interpretive Overheads. Eijiro Sumii. ....

Andrzej Filinski. Controlling Effects. PhD thesis, School of Computer Science, Carnegie Mellon University, 1996. Available at http://www.brics.dk/~andrzej/papers/CE.ps.gz.

....over 30 years of programming language development, suggested that it would be valuable to bring typed call by name (CBN) and typed call by value (CBV) into a common framework. We propose call by push value (CBPV) a new typed paradigm based on Filinski s variant of Moggi s computational calculus [7, 21], as a solution to that problem. We will introduce a CBPV language, and give translations from CBN and CBV languages into it. We claim that, via these translations, CBPV subsumes CBN and CBV. But what does it mean for one language to subsume another After all, there are sound, adequate ....

....the objectives that they were designed to achieve are rather different. We first look at cases in which semantics for the source language does not, so far as we can see, extend along the translation. It is not evident how to provide operational semantics for the monadic target languages of [4, 7] so as to recover standard operational semantics for the source languages. The monad language of [4] do not provide semantics for CBN, because the translation from CBN into it does not preserve the j law for functions. This applies also to the thunking transform from CBN to CBV [9, 34] As ....

[Article contains additional citation context not shown here]

Andrzej Filinski. Controlling Effects. PhD thesis, School of Computer Science, Carnegie Mellon University, Pittsburgh, Pennsylvania, May 1996.

....the entire specification. The disciplined use of monads and monad transformers, as proposed by Moggi and others [133, 134, 40, 88, 87] also provides a systematic approach to achieving good pragmatics of modularity in denotational semantics; it is interesting that monads also support reflection [47]. Recently, a close relationship between action semantics and the monadic style of denotational semantics has been shown [195, 196] B.4 Technical Approach for Options Option 1. A Reasoning Tool for Java For formal interoperation of tools that manipulate, analyze and transform code it is ....

....algebraically specified, and essentially a parameter of action notation. Action semantics has recently been shown to be closely related to monad transformers [195, 196] Moreover, although action notation already provides some support for reflection, the moadic treatment of reflection by Filinski [47] may lead to an enhancement of this aspect. The frameworks of action semantics, uniform semantics, and rewriting logic have considerable potential for synergy. Adopting uniform semantics for defining action notation would make the theory of equivalence developed for uniform semantics applicable ....

Andrzej Filinski. Controlling Effects. PhD thesis, School of Computer Science, Carnegie Mellon University, 1996.

....and then to apply the methods considered here to complete the proof of correspondence between the original program and its cps transformation. The treatment of cps conversion given here invites generalization to an arbitrary syntactically definable monad for the language. Filinski s dissertation [3] is a first step towards a general theory of representation of computational effects. Filinski s work suggests that one could give a fairly general correctness proof along the lines suggested here for a wide variety of definable effects. Acknowledgements We are grateful to John Mitchell and Andy ....

Andrzej Filinski. Controlling Effects. CMU--CS--96--119, School of Computer Science, Carnegie Mellon University, May 1996.

.... call by name with lazy to distinguish this transform from non lazy alternatives that satisfy the full j law. Unfortunately, lazy is sometimes used to mean call by need. A genuinely call by name CPS transform not suffering from laziness can be given by uncurrying all function types. In [4] a related transform is given. But as the target language is a calculus, it is, apart from the thunking of base types, the identity. Definition 2.4 The (uncurrying) call by name CPS transform is defined as follows N LxM(k) khxi N Lx:MM(k) khfiffhx yhi(N LMM(m)fmhgi(gh yhigg N ....

Andrzej Filinski. Controlling Effects. PhD thesis, School of Computer Science, Carnegie Mellon University, 1996.

....the shapes of data structures and the aliasing relationships between them. We plan to investigate how to use these finergrained memory models to increase the flexibility of our type system. A connection can also be drawn between capabilities and monadic type systems. Work relating e#ects to monads [21, 28, 18, 8] has viewed e#ectful functions as pure functions that return state transformers. This might be called an ex post view: the e#ect takes place after the function s execution. In contrast, we take an ex ante view in which the capability to perform the relevant e#ect must be satisfied before the ....

Andrzej Filinski. Controlling E#ects. PhD thesis, Carnegie Mellon University, School of Computer Science, Pittsburgh, Pennsylvania, May 1996.

....a new style of semantics, e.g. Scott semantics, operational semantics, game semantics, continuation semantics etc. we always need to do it twice once for each paradigm. We propose call by push value (CBPV) a new typed paradigm based on Filinski s variant of Moggi s computational calculus [7, 21], as a solution to that problem. We will introduce a CBPV language, and give translations from CBN and CBV languages into it. We claim that, via these translations, CBPV subsumes CBN and CBV. But what does it mean for one language to subsume another After all, there are sound, adequate ....

....Of course, the objectives that they were designed to achieve are different. We first look at cases in which semantics for the source language does not, so far as we can see, extend along the translation. It is not evident how to provide operational semantics for the monadic target languages of [4, 7] so as to recover standard operational semantics for the source languages. The monad language of [4] do not provide semantics for CBN, because the translation from CBN into it does not preserve the j law for functions. This applies also to the thunking transform from CBN to CBV [10, 34] ....

[Article contains additional citation context not shown here]

A. Filinski. Controlling Effects. PhD thesis, School of Computer Science, Carnegie Mellon University, Pittsburgh, Pennsylvania, 1996.

....its sponsoring group, and hence it cannot control tasks running in parallel with it, unless explicitly passed handles to their sponsoring groups. Our semantics measures the resources used by computations according to an independent cost model; our approach generalises the complexity monad [9, 33]. As our semantics is independent of the cost model, one might prefer to have more a realistic cost model, for instance, taking into account memory occupancy (duration or size) or blocked dequeue operations. In order to have a finer control on energy consumption, we noticed that the system could ....

Andrzej Filinski. Controlling Effects. PhD thesis, School of Computer Science. Carnegie Mellon University, May 1996.

....algorithm as one instance of Godelization, and presents a Godelizer for proper combinators in the untyped calculus. An implementation of Berger and Schwichtenberg s algorithm in Standard ML can be found in Filinski s PhD 6 In that work, reify is quote and reflect is unquote . thesis [23]. This implementation handles most of the examples displayed in the present paper, in ML. It is ingenious because as expressed in Figures 1 and 2, type directed partial evaluation requires dependent types. It can be easily translated into Haskell (excluding disjoint sums, of course, for lack of ....

Andrzej Filinski. Controlling Effects. PhD thesis, School of Computer Science, Carnegie Mellon University, Pittsburgh, Pennsylvania, 1995.

....extended to a language with mutable storage (e.g. ML reference types) Both of these extensions are important directions for further research. The treatment of cps conversion given here invites generalization to an arbitrary syntactically definable monad for the language. Filinski s dissertation [5] is a first step towards a general theory of representation of computational effects. Filinski s work suggests that one could give a fairly general correctness proof along the lines suggested here for a wide variety of definable effects. Finally, the relational interpretation of types given in ....

Andrzej Filinski. Controlling Effects. CMU--CS--96--119, School of Computer Science, Carnegie Mellon University, May 1996.

....care of by subsumption, since T T . One may extend the theory to a full type theory of monads by dispensing with T T , adding an explicit unit operator and adding rules comprising the monadic equality laws. This would provide a powerful mechanism for abstractly managing effects in general [34], not just the nontermination effects managed by the partial types. I have not done so in this type theory, preferring instead to keep the type theory closely tied to its operational semantics, thereby providing as much structure as possible at the expense of abstraction (recall Section 3.2.1) ....

Andrzej Filinski. Controlling Effects. Ph.D. dissertation, Carnegie Mellon University, School of Computer Science, Pittsburgh, Pennsylvania, May 1996.

....shapes of data structures and the aliasing relationships between them. We plan to investigate how to use these finergrained memory models to increase the flexibility of our type system. A connection can also be drawn between capabilities and monadic type systems. Work relating effects to monads [21, 28, 18, 8] has viewed effectful functions as pure functions that return state transformers. This might be called an ex post view: the effect takes place after the function s execution. In contrast, we take an ex ante view in which the capability to perform the relevant effect must be satisfied before the ....

Andrzej Filinski. Controlling Effects. PhD thesis, Carnegie Mellon University, School of Computer Science, Pittsburgh, Pennsylvania, May 1996.

....that an extra argument be passed to each function referring to the dynamic state . As discussed in Section 2, such a code is prone to errors, and it also hampers scalability. 8. Semantics of Exceptions First class continuations and state can be used to implement exception handling mechanisms [18]. We show here that the same is true for first class continuations and dynamic binding. In Standard ML, exceptions are raised by an operator raise, and are caught by handlers installed with handle. In the semantics of ML [6, 40] a raised exception returns an exceptional value, distinct from a ....

Andrzej Filinski. Controlling Effects. PhD thesis, School of Computer Science. Carnegie Mellon University, May 1996.

....cannot distinguish. As a result, v 6ae =d , and using Felleisen s definition of expressiveness [8, Thm 3.14] we conclude that: Proposition 1. v cannot macro express dynamic binding relative to d . 6 Semantics of Exceptions First class continuations and state can simulate exceptions [13]. We show here that exceptions can be defined in terms of first class continuations and dynamic binding. In the semantics of ML [26] a raised exception returns an exceptional value, distinct from a normal value, which has the effect to prune its evaluation context until a handler is able to deal ....

Andrzej Filinski. Controlling Effects. PhD thesis, School of Computer Science. Carnegie Mellon University, May 1996.

No context found.

Andrzej Filinski. Controlling E#ects. PhD thesis, Carnegie Mellon University, School of Computer Science, Pittsburgh, Pennsylvania, May 1996.

....shapes of data structures and the aliasing relationships between them. We plan to investigate how to use these finergrained memory models to increase the flexibility of our type system. A connection can also be drawn between capabilities and monadic type systems. Work relating e#ects to monads [21, 28, 18, 8] has viewed e#ectful functions as pure functions that return state transformers. This might be called an ex post view: the e#ect takes place after the function s execution. In contrast, we take an ex ante view in which the capability to perform the relevant e#ect must be satisfied before the ....

Andrzej Filinski. Controlling E#ects. PhD thesis, Carnegie Mellon University, School of Computer Science, Pittsburgh, Pennsylvania, May 1996.

No context found.

Filinski, A. 1996. Controlling eects. Ph.D. thesis, Carnegie Mellon University, School of Computer Science, Pittsburgh, Pennsylvania.

No context found.

Andrzej Filinski. Controlling E#ects. PhD thesis, School of Computer Science, Carnegie Mellon University, Pittsburgh, Pennsylvania, USA, May 1996.

No context found.

Andrzej Filinski. Controlling Effects. PhD thesis, School of Computer Science, Carnegie Mellon University, 1996. 165

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC