DOCSLIB.ORG

Fast Numerical Computation in C++: Expression Templates and Beyond

Total Page：16

File Type：pdf, Size：1020Kb

Download full-text PDF Read full-text
Fast Numerical Computation in C++: Expression Templates and Beyond Numerical computation in C++ Introduction Numerical Fast numerical computation in C++: algorithms, libraries and their Expression Templates and Beyond to performance Interlude: Profiling Lazy Code Generation (LzCG) on Linux Expression template generalities B. Nikolic Lazy code generation – what it is & how it works Cavendish Laboratory/Kavli Institute LzCG example University of Cambridge Summary BoostCon 2011 May 2011 Numerical Overview of ideas computation in C++ Introduction 1. ‘Standard’ rules of C++ lead to inefficient numerical Numerical algorithms, code libraries and their 2. New rules (≡ sub-languages) can be implemented performance Interlude: Profiling using expression templates on Linux 2.1 Types are used confer information about expressions Expression template 2.2 Translated to ‘standard’ C++ at compile-time generalities 3. Makes high-performance numerical C++ libraries Lazy code generation – what possible and successful it is & how it works 4. But is it enough? LzCG example 4.1 Most efficient algorithm not obvious at compile-time Summary 4.2 Convenience/flexibility of generating code in C++ 5. Types retain information about expressions in signatures in object code 5.1 Can re-generate expression template implementations post-compilation-time Numerical Outline computation in C++ Introduction Numerical Introduction algorithms, libraries and their performance Numerical algorithms, libraries and their performance Interlude: Profiling on Linux Expression Interlude: Profiling on Linux template generalities Lazy code Expression template generalities generation – what it is & how it works LzCG example Lazy code generation – what it is & how it works Summary LzCG example Summary Numerical About myself: ALMA telescope computation in C++ Largest ground-based astronomy project in the world Introduction Numerical algorithms, libraries and their performance Interlude: Profiling on Linux Expression template generalities Lazy code generation – what it is & how it works LzCG example Summary Currently being commissioned at altitude of 5000 m in Chile. Will have 66 telescopes separated by up to 15 kms and observed at wavelength between 7 and 0.35 mm. Numerical About myself: Green Bank Telescope computation in C++ Largest steerable telescope in the world Introduction Numerical algorithms, libraries and their performance Interlude: Profiling on Linux Expression template generalities Lazy code generation – what it is & how it works LzCG example Summary Main reflector is 100x110 m in size, total height 160 m. Entire structure is accurate to 0.25 mm. Numerical About myself: Thermal radio emission from computation in C++ Messier 66 Introduction Numerical algorithms, libraries and their performance Interlude: Profiling I Colour scale is on Linux emission from dust at Expression template 0.024 mm wavelegnth generalities Lazy code I Contours represent generation – what emission at 3 mm from it is & how it works hot electron gas LzCG example Summary I Both appear to be powered by recent star formation Numerical General Interests computation in C++ Introduction Numerical algorithms, libraries and their performance Interlude: Profiling I Model optimisation and statistical inference on Linux (maximum-likelihood, Markov Chain Monte Carlo, Expression template Nested Sampling techniques) generalities Pricing and risk-management of derivative contracts Lazy code I generation – what it is & how it works I Remote sensing of Earth’s atmosphere LzCG example I Radiative transfer and other physical simulations Summary ) All very numerically intensive applications... Numerical Aperture synthesis radio-astronomy computation in C++ Introduction Revolutionised the radio view of the universe – Nobel I Numerical prize in 1972 algorithms, libraries and their I Development of the technique closely tied to performance computers: Interlude: Profiling on Linux I Lots of Fourier Transforms Expression I Large quantities of data to be binned, inspected, template generalities discarded if necessary Lazy code I Instruments inherently unstable so calibration is generation – what critical it is & how it works LzCG example I Atacama Large Millimetre Array: eventually 66 antennas, ∼ 20 Mb=s average output data rate: Summary I Computational issues inconvenient, reduce scientist productivity I Square Kilometre Array (SKA): 1000s antennas, wide field of view, ∼ few Gb=s average output data rate: I Computational issues limiting factor in scientific output Numerical Risk management of ‘derivative’ contracts in computation in C++ finance Requirements in just one product line (e.g., credit derivatives) Introduction Numerical algorithms, libraries and their Typically calculations involve either: solving PDEs using performance Interlude: Profiling finite differences; or computing FFTs; or Monte-Carlo (MC) on Linux simulations. Expression template generalities I 2000 nodes × 1 kW=node + 50% aircon cost = Lazy code 3 MW generation – what it is & how it works I 3 MW × 10 p=s × 8500 hr=yr = LzCG example 6 ∼ 2:5 × 10 GBP=yr! Summary I Additional costs / number of nodes: I Installation, maintenance, software licenses (even Excel sometimes!) I Floor-space (in expensive buildings) I Standby backup power generation costs Numerical Numerical performance computation in C++ (Why) does it matter? Introduction Numerical algorithms, Easily parallelisable Difficult to parallelise libraries and their performance Interlude: Profiling I Cost I Feasibility on Linux I Heat, power, floor I Latency Expression template space generalities I User patience Lazy code I Environmental impact generation – what it is & how it works I Time to scale-up LzCG example I Access to capital Summary Parallelisation is usually the most important aspect of high-performance numerical computing I Not directly considering it in this talk although much of the material is relevant Numerical Small problems ≡ simple solutions computation in C++ Many practical scientific and industrial problems can be accelerated a simple way Introduction Numerical algorithms, Listing 1: By-hand coding + SIMD intrinsics libraries and their void add2Vect ( const std :: vector<double> &v1 , performance const std :: vector<double> &v2 , std :: vector<double> &res ) f Interlude: Profiling typedef double v2df a t t r i b u t e ( ( mode(V2DF ) ) ) ; on Linux ∗ ∗ ∗ v2df dest =( v2df )&( res.begin()); Expression const s i z e t n=v1.size(); template ∗ ∗ const v2df src1 =( const v2df )&v1 [ 0 ] ; generalities const v2df ∗src2 =( const v2df ∗)&v2 [ 0 ] ; i f ( n%2==0) Lazy code f generation – what for ( s i z e t i =0; i<n / 2 ; i ++) it is & how it works f dest [ i ]= b u i l t i n i a 3 2 addpd(src1[i],src2[i ]); LzCG example g g Summary else f for ( s i z e t i =0; i<n ; ++ i ) f dest[i]=src1[i]+src2[i ]; g g g Simple problems are common in real life but not really the subject of this talk! Numerical Hand coding unsuitable for large systems computation in C++ Introduction Numerical algorithms, libraries and their performance I Correctness Interlude: Profiling I Maintainability, readability, portability on Linux Expression I Algorithms need adjustment over time template generalities I Experiment with different implementations of Lazy code generation – what algorithms it is & how it works I Approximations: how much precision, what accuracy LzCG example is necessary? Summary ) These can be difficult to achieve with complex hand-crafted code! Numerical Warning! computation in C++ “Don’t try this at home” – try existing libraries first Introduction Numerical algorithms, libraries and their performance Writing numerical libraries is difficult and error prone – Interlude: Profiling on Linux always carefully consider alternatives! Expression template I Can you use standard existing libraries (“C” or “C++”) generalities Lazy code I Are you writing a general purpose library or an generation – what application? it is & how it works LzCG example I Can you, in advance, identify a subset of algorithm Summary which is likely to consume most time but can present a clean, data-only, interface? Numerical Outline computation in C++ Introduction Numerical Introduction algorithms, libraries and their performance Numerical algorithms, libraries and their performance Interlude: Profiling on Linux Expression Interlude: Profiling on Linux template generalities Lazy code Expression template generalities generation – what it is & how it works LzCG example Lazy code generation – what it is & how it works Summary LzCG example Summary Numerical Requirements for good numerical computation in C++ performance Introduction Numerical algorithms, I Maximise parallelism libraries and their performance I Use all of the nodes/processors/cores/execution units Interlude: Profiling I Use Single-Instruction-Multiple-Data (SIMD) on Linux Minimise memory access Expression I template generalities I Keep close data to be processed together I Use algorithms that process small chunks of input Lazy code generation – what data at a time it is & how it works I Avoid temporaries LzCG example I Minimise ‘branching’ Summary I Keep the pipeline and speculative fetches good I But, need enough code at hand to execute I Minimise quantity of transcendental calculations I Includes division in this set I Reducing precision or accuracy makes these faster Numerical Optimisation Challenges I computation in C++ Introduction I Want: to describe the algorithm in simple, readable, Numerical re-usable way algorithms, libraries and their / / This : performance R=A+B+C+D+E; / / Not t h i s : Interlude: Profiling addFiveVect Double Double Double Double
Load more

Recommended publications
  • CSCI 2041: Lazy Evaluation
    CSCI 2041: Lazy Evaluation Chris Kauffman Last Updated: Wed Dec 5 12:32:32 CST 2018 1 Logistics Reading Lab13: Lazy/Streams I Module Lazy on lazy Covers basics of delayed evaluation computation I Module Stream on streams A5: Calculon Lambdas/Closures I Arithmetic language Briefly discuss these as they interpreter pertain Calculon I 2X credit for assignment I 5 Required Problems 100pts Goals I 5 Option Problems 50pts I Eager Evaluation I Milestone due Wed 12/5 I Lazy Evaluation I Final submit Tue 12/11 I Streams 2 Evaluation Strategies Eager Evaluation Lazy Evaluation I Most languages employ I An alternative is lazy eager evaluation evaluation I Execute instructions as I Execute instructions only as control reaches associated expression results are needed code (call by need) I Corresponds closely to I Higher-level idea with actual machine execution advantages and disadvantages I In pure computations, evaluation strategy doesn’t matter: will produce the same results I With side-effects, when code is run matter, particular for I/O which may see different printing orders 3 Exercise: Side-Effects and Evaluation Strategy Most common place to see differences between Eager/Lazy eval is when functions are called I Eager eval: eval argument expressions, call functions with results I Lazy eval: call function with un-evaluated expressions, eval as results are needed Consider the following expression let print_it expr = printf "Printing it\n"; printf "%d\n" expr; ;; print_it (begin printf "Evaluating\n"; 5; end);; Predict results and output for both Eager and Lazy Eval strategies 4 Answers: Side-Effects and Evaluation Strategy let print_it expr = printf "Printing it\n"; printf "%d\n" expr; ;; print_it (begin printf "Evaluating\n"; 5; end);; Evaluation > ocamlc eager_v_lazy.ml > ./a.out Eager Eval # ocaml’s default Evaluating Printing it 5 Lazy Eval Printing it Evaluating 5 5 OCaml and explicit lazy Computations I OCaml’s default model is eager evaluation BUT.
    [Show full text]
  • Control Flow
    Control Flow COMS W4115 Prof. Stephen A. Edwards Spring 2002 Columbia University Department of Computer Science Control Flow “Time is Nature’s way of preventing everything from happening at once.” Scott identifies seven manifestations of this: 1. Sequencing foo(); bar(); 2. Selection if (a) foo(); 3. Iteration while (i<10) foo(i); 4. Procedures foo(10,20); 5. Recursion foo(int i) { foo(i-1); } 6. Concurrency foo() jj bar() 7. Nondeterminism do a -> foo(); [] b -> bar(); Ordering Within Expressions What code does a compiler generate for a = b + c + d; Most likely something like tmp = b + c; a = tmp + d; (Assumes left-to-right evaluation of expressions.) Order of Evaluation Why would you care? Expression evaluation can have side-effects. Floating-point numbers don’t behave like numbers. Side-effects int x = 0; int foo() { x += 5; return x; } int a = foo() + x + foo(); What’s the final value of a? Side-effects int x = 0; int foo() { x += 5; return x; } int a = foo() + x + foo(); GCC sets a=25. Sun’s C compiler gave a=20. C says expression evaluation order is implementation-dependent. Side-effects Java prescribes left-to-right evaluation. class Foo { static int x; static int foo() { x += 5; return x; } public static void main(String args[]) { int a = foo() + x + foo(); System.out.println(a); } } Always prints 20. Number Behavior Basic number axioms: a + x = a if and only if x = 0 Additive identity (a + b) + c = a + (b + c) Associative a(b + c) = ab + ac Distributive Misbehaving Floating-Point Numbers 1e20 + 1e-20 = 1e20 1e-20 1e20 (1 + 9e-7) + 9e-7 6= 1 + (9e-7 + 9e-7) 9e-7 1, so it is discarded, however, 1.8e-6 is large enough 1:00001(1:000001 − 1) 6= 1:00001 · 1:000001 − 1:00001 · 1 1:00001 · 1:000001 = 1:00001100001 requires too much intermediate precision.
    [Show full text]
  • 210 CHAPTER 7. NAMES and BINDING Chapter 8
    210 CHAPTER 7. NAMES AND BINDING Chapter 8 Expressions and Evaluation Overview This chapter introduces the concept of the programming environment and the role of expressions in a program. Programs are executed in an environment which is provided by the operating system or the translator. An editor, linker, file system, and compiler form the environment in which the programmer can enter and run programs. Interac- tive language systems, such as APL, FORTH, Prolog, and Smalltalk among others, are embedded in subsystems which replace the operating system in forming the program- development environment. The top-level control structure for these subsystems is the Read-Evaluate-Write cycle. The order of computation within a program is controlled in one of three basic ways: nesting, sequencing, or signaling other processes through the shared environment or streams which are managed by the operating system. Within a program, an expression is a nest of function calls or operators that returns a value. Binary operators written between their arguments are used in infix syntax. Unary and binary operators written before a single argument or a pair of arguments are used in prefix syntax. In postfix syntax, operators (unary or binary) follow their arguments. Parse trees can be developed from expressions that include infix, prefix and postfix operators. Rules for precedence, associativity, and parenthesization determine which operands belong to which operators. The rules that define order of evaluation for elements of a function call are as follows: • Inside-out: Evaluate every argument before beginning to evaluate the function. 211 212 CHAPTER 8. EXPRESSIONS AND EVALUATION • Outside-in: Start evaluating the function body.
    [Show full text]
  • Exam 1 Review: Solutions Capture the Dragons
    CSCI 1101B: Introduction to Computer Science Instructor: Prof. Harmon Exam 1 Review: Solutions Capture the Dragons Boolean Expressions 1. Of the following, which are Boolean expressions? (a)2+3 Not Boolean (b)2>3 Boolean (c) num >= 2 Boolean (d) num1 = num2 Not Boolean 2. If x = 3, and x <= y evaluates to True, what values can y have? y >= 3 3. Simplify each of the following expressions: (a) my_num or True Depends on what the value of my_num is. See rules as in #5. (b) False and "Mary" or True True (c) False and ("Mary" or True) False 4. Is x =< y the same as x <= y? No. Python only recognizes the latter. The former is invalid syntax. 5. If x = 1 and y = 2, explain the following results: (a) >>> x and y # = 2 (b) >>> x or y # = 1 Python uses lazy evaluation. • The expression x and y first evaluates x. If x is equivalent to False (or empty/zero), its value is returned. Otherwise, y is evaluated and the resulting value is returned. • The expression x or y first evaluates x. If x is equivalent to True (or non-empty/nonzero), its value is returned. Otherwise, y is evaluated and the resulting value is returned. 1 CSCI 1101B: Introduction to Computer Science Exam 1 Review: Solutions Conditionals 1. What is indentation, and how is it related to the concept of a code block? Indentation is the placement of text (to the left or right). Python uses indenta- tion to associate code into groups, or blocks. 2. What is the difference between a chained conditional and a nested conditional? Chained => if, elif, else Nested => conditionals exist (are "nested") inside each other 3.
    [Show full text]
  • Top Functional Programming Languages Based on Sentiment Analysis 2021 11
    POWERED BY: TOP FUNCTIONAL PROGRAMMING LANGUAGES BASED ON SENTIMENT ANALYSIS 2021 Functional Programming helps companies build software that is scalable, and less prone to bugs, which means that software is more reliable and future-proof. It gives developers the opportunity to write code that is clean, elegant, and powerful. Functional Programming is used in demanding industries like eCommerce or streaming services in companies such as Zalando, Netflix, or Airbnb. Developers that work with Functional Programming languages are among the highest paid in the business. I personally fell in love with Functional Programming in Scala, and that’s why Scalac was born. I wanted to encourage both companies, and developers to expect more from their applications, and Scala was the perfect answer, especially for Big Data, Blockchain, and FinTech solutions. I’m glad that my marketing and tech team picked this topic, to prepare the report that is focused on sentiment - because that is what really drives people. All of us want to build effective applications that will help businesses succeed - but still... We want to have some fun along the way, and I believe that the Functional Programming paradigm gives developers exactly that - fun, and a chance to clearly express themselves solving complex challenges in an elegant code. LUKASZ KUCZERA, CEO AT SCALAC 01 Table of contents Introduction 03 What Is Functional Programming? 04 Big Data and the WHY behind the idea of functional programming. 04 Functional Programming Languages Ranking 05 Methodology 06 Brand24
    [Show full text]
  • CS1101S Studio Session Week 11: Stream
    Welcome CS1101S Studio Session Week 11: Stream Niu Yunpeng [email protected] October 30, 2018 Niu Yunpeng CS1101S Studio Week 11 October 30, 2018 1 / 27 Overview 1 Lazy evaluation Computational model Function application 2 Stream Delayed evaluation Stream programming Niu Yunpeng CS1101S Studio Week 11 October 30, 2018 2 / 27 Lazy Evaluation Computational model Computational model is a useful guideline for us to understand how the interpreter works. Computational model may vary depending on programming language and the runtime system used. In CS1101S, we introduce two computational models: substitution model and environment model. What to expect In the coming weeks, these two models will still be valid. Niu Yunpeng CS1101S Studio Week 11 October 30, 2018 3 / 27 Lazy Evaluation Substitution model For stateless programming only: Evaluate all actual arguments; Replace all formal parameters with their actual arguments; Apply each statement in the function body (and get the return value); Repeat the first 3 steps until done. Niu Yunpeng CS1101S Studio Week 11 October 30, 2018 4 / 27 Lazy Evaluation Environment model For stateful programming: Each frame contains a series of bindings of names and values. The value of a variable depends on its environment, a sequence of frames up to the global frame. Each function call will create a new frame and extend its enclosing environment. Niu Yunpeng CS1101S Studio Week 11 October 30, 2018 5 / 27 Lazy Evaluation Function in JavaScript Function in JavaScript is a first-class citizen (object). They have a call method. The call method is triggered when this function is applied. Function application When a function is applied, “this” is prepanded to the list of parameters.
    [Show full text]
  • Monads for Functional Programming
    Monads for functional programming Philip Wadler, University of Glasgow? Department of Computing Science, University of Glasgow, G12 8QQ, Scotland ([email protected]) Abstract. The use of monads to structure functional programs is de- scribed. Monads provide a convenient framework for simulating effects found in other languages, such as global state, exception handling, out- put, or non-determinism. Three case studies are looked at in detail: how monads ease the modification of a simple evaluator; how monads act as the basis of a datatype of arrays subject to in-place update; and how monads can be used to build parsers. 1 Introduction Shall I be pure or impure? The functional programming community divides into two camps. Pure lan- guages, such as Miranda0 and Haskell, are lambda calculus pure and simple. Impure languages, such as Scheme and Standard ML, augment lambda calculus with a number of possible effects, such as assignment, exceptions, or continu- ations. Pure languages are easier to reason about and may benefit from lazy evaluation, while impure languages offer efficiency benefits and sometimes make possible a more compact mode of expression. Recent advances in theoretical computing science, notably in the areas of type theory and category theory, have suggested new approaches that may integrate the benefits of the pure and impure schools. These notes describe one, the use of monads to integrate impure effects into pure functional languages. The concept of a monad, which arises from category theory, has been applied by Moggi to structure the denotational semantics of programming languages [13, 14]. The same technique can be applied to structure functional programs [21, 23].
    [Show full text]
  • The Scala Programming Language
    The Scala Programming Language Mooly Sagiv Slides taken from Martin Odersky (EPFL) Donna Malayeri (CMU) Hila Peleg (TAU) Modern Functional Programming • Higher order • Modules • Pattern matching • Statically typed with type inference • Two viable alternatives • Haskel • Pure lazy evaluation and higher order programming leads to Concise programming • Support for domain specific languages • I/O Monads • Type classes • OCaml • Encapsulated side-effects via references Then Why aren’t FP adapted? • Education • Lack of OO support • Subtyping increases the complexity of type inference • Programmers seeks control on the exact implementation • Imperative programming is natural in certain situations Why Scala? (Coming from OCaml) • Runs on the JVM/.NET • Can use any Java code in Scala • Combines functional and imperative programming in a smooth way • Effective library • Inheritance • General modularity mechanisms The Java Programming Language • Designed by Sun 1991-95 • Statically typed and type safe • Clean and Powerful libraries • Clean references and arrays • Object Oriented with single inheritance • Interfaces with multiple inhertitence • Portable with JVM • Effective JIT compilers • Support for concurrency • Useful for Internet Java Critique • Downcasting reduces the effectiveness of static type checking • Many of the interesting errors caught at runtime • Still better than C, C++ • Huge code blowouts • Hard to define domain specific knowledge • A lot of boilerplate code • Sometimes OO stands in our way • Generics only partially helps Why
    [Show full text]
  • Comparison Between Lazy and Strict Evaluation
    View metadata, citation and similar papers at core.ac.uk brought to you by CORE provided by Trepo - Institutional Repository of Tampere University Nayeong Song Comparison between lazy and strict evaluation Bachelor’s thesis Faculty of Information Technology and Communication Sciences (ITC) Examiner: Maarit Harsu Oct 2020 i ABSTRACT Nayeong Song: Comparison between lazy and strict evaluation Bachelor’s thesis Tampere University Bachelor’s Degree Programme in Science and Engineering Oct 2020 Evaluation strategy is the way to defne the order of function call’s argument’s evaluation. Most of the today’s programming languages employ strict programming paradigm, which is focused mostly on call-by-value and call-by-reference. However, call-by-name or call-by-need, which evaluates the expression lazily, is used only in functional programming languages and those are rather unfamiliar to these day’s programmers. In this thesis, the focus is on the comparison between diferent evaluation techniques but mostly focused on strict evaluation and lazy evaluation. First, the theoretical diference is presented in mathematical aspect in lambda calculus, and the practical examples are presented in Python and Haskell. The results show that a program can beneft lazy evaluation when evaluating non- terminable expressions or optimize the performance when the values are cached. However, in case the function has side efects, lazy evaluation cannot be used to- gether and it makes harder to predict the memory usage compared to strict evalu- ation. Keywords: evaluation strategy, lambda calculus, algorithm efciency, strict eval- uation, lazy evaluation, Haskell, Python The originality of this thesis has been checked using the Turnitin Originality Check service.
    [Show full text]
  • CS 251: Programming Languages Fall 2015 ML Summary, Part 5
    CS 251: Programming Languages Fall 2015 ML Summary, Part 5 These notes contain material adapted from notes for CSE 341 at the University of Washington by Dan Grossman. They have been converted to use SML instead of Racket and extended with some additional material by Ben Wood. Contents Introduction to Delayed Evaluation and Thunks .......................... 1 Lazy Evaluation with Delay and Force ................................. 2 Streams .................................................... 4 Memoization ................................................. 6 Integrating Lazy Evaluation into the Language ........................... 11 Introduction to Delayed Evaluation and Thunks A key semantic issue for a language construct is when are its subexpressions evaluated. For example, in ML (and similarly in Racket and most but not all programming languages), given e1 e2 ... en we evaluate the function arguments e2, ..., en once before we execute the function body and given a function fn ... => ... we do not evaluate the body until the function is called. This rule (\evaluate arguments in advance") goes by many names, including eager evaluation and call-by-value. (There is a family of names for parameter-passing styles referred to with call-by-.... Many are not particularly illuminating or downright confusing names, but we note them here for reference.) We can contrast eager evaluation with how if e1 then e2 else e3 works: we do not evaluate both e2 and e3. This is why: fun iffy x y z = if x then y else z is a function that cannot be used wherever
    [Show full text]
  • Programming Language Concepts Subroutines
    Programming Language Concepts Subroutines Janyl Jumadinova 23 February, 2017 Arguments are also called \actual parameters" (as opposed to \formal parameters" in the function definition) Arguments, Parameters 2/20 Arguments, Parameters Arguments are also called \actual parameters" (as opposed to \formal parameters" in the function definition) 2/20 Examples of Function Calls (Python) def f(a,b,c): ... return 100*a+10*b+c 3/20 I Named Arguments Examples of Function Calls (Python) def f(a=1,b=1,c=1): ... return 100*a+10*b+c I Default Parameter Values 4/20 Examples of Function Calls (Python) def f(a=1,b=1,c=1): ... return 100*a+10*b+c I Default Parameter Values I Named Arguments 4/20 Examples of Function Calls (Haskell) 5/20 Parameter Evaluation \Applicative Order" evaluation: arguments evaluated before the function call. \Eager" evaluation. 6/20 Parameter Evaluation \Normal Order" evaluation: arguments are not evaluated until they are needed (possibly never) 7/20 Parameter Evaluation \Lazy" evaluation: I arguments are evaluated at most once (possibly never). I Even though we used a Haskell example to illustrate \normal order", it is more accurate to call Haskell's evaluation order \lazy." I In normal order, a parameter could be evaluated more than once (i.e., evaluated each time it appears in the function). 8/20 Closures I In languages that support “first-class functions," a function may be a return value from another function; a function may be assigned to a variable. I This raises some issues regarding scope. 9/20 Closures 10/20 Another solution (used by JavaScript and most other statically-scoped languages): bind the variables that are in the environment where the function is defined.
    [Show full text]
  • Lazy and Eager Evaluation
    Functional Programming and Specification Lecture Note 1, 4 February 2011 Lazy and Eager Evaluation Recall that evaluation is “call by value” in ML: before performing function application, we have to evaluate the argument. This happens even if the argument isn’t needed for the result, as in fun f(y) = 1 Because of this, ML is called an eager language. Another terminology for essentially the same thing is that ML is strict; this means that applying f or any other function to an expression exp having no value (because exp fails to terminate) gives an expression that has no value. Some functional languages use lazy evaluation, also known as “call by need”, meaning that an expression will not be evaluated unless its value is actually required in order to produce the result. In a lazy language, f(exp) = 1, even if exp fails to terminate — the argument to f doesn’t need to be evaluated to determine the result. This isn’t done by analyzing the definition of f, but rather by delaying evaluation of the argument until its value is required. One can’t analyze functions in general to see if they use their arguments — one reason (not the only one!) is that in ML, any function might be the outcome of a complicated computation rather than just text. Not evaluating an unused argument saves time. If its evaluation doesn’t terminate, not evaluating it saves a lot of time! Here is another example: fun g(x,y,z) = if x<2 then y+3 else z+6 Depending on the value of x, either the value of y or the value of z will be required, but not both.
    [Show full text]