Mercurial > octave-nkf
changeset 14117:acdc4520286a stable
Trivial merge
| author | Jordi Gutiérrez Hermoso <jordigh@octave.org> |
|---|---|
| date | Tue, 27 Dec 2011 18:47:18 -0500 |
| parents | 951eacaf9381 (diff) 6e9fee72a01c (current diff) |
| children | ebe2e6b2ba52 |
| files | scripts/miscellaneous/fact.m scripts/miscellaneous/module.mk |
| diffstat | 21 files changed, 1026 insertions(+), 275 deletions(-) [+] |
line wrap: on
line diff
--- a/NEWS Tue Dec 27 15:08:21 2011 -0500 +++ b/NEWS Tue Dec 27 18:47:18 2011 -0500 @@ -42,8 +42,8 @@ += -= .+= .-= .*= ./= .\= .^= .**= &= |= - See the new "Broadcasting Operations" chapter in the manual for more - details. + See the "Broadcasting" section in the new "Vectorization and Faster + Code Execution" chapter of the manual for more details. ** Octave now features a profiler, thanks to the work of Daniel Kraft under the Google Summer of Code mentorship program. The manual has
--- a/doc/interpreter/Makefile.am Tue Dec 27 15:08:21 2011 -0500 +++ b/doc/interpreter/Makefile.am Tue Dec 27 18:47:18 2011 -0500 @@ -148,7 +148,8 @@ system.texi \ testfun.texi \ tips.texi \ - var.texi + var.texi \ + vectorize.texi TXI_SRC = $(MUNGED_TEXI_SRC:.texi=.txi)
--- a/doc/interpreter/arith.txi Tue Dec 27 15:08:21 2011 -0500 +++ b/doc/interpreter/arith.txi Tue Dec 27 18:47:18 2011 -0500 @@ -189,10 +189,6 @@ @DOCSTRING(sumsq) -@DOCSTRING(accumarray) - -@DOCSTRING(accumdim) - @node Utility Functions @section Utility Functions
--- a/doc/interpreter/container.txi Tue Dec 27 15:08:21 2011 -0500 +++ b/doc/interpreter/container.txi Tue Dec 27 18:47:18 2011 -0500 @@ -507,9 +507,7 @@ The simplest way to process data in a structure is within a @code{for} loop (@pxref{Looping Over Structure Elements}). A similar effect can be achieved with the @code{structfun} function, where a user defined -function is applied to each field of the structure. - -@DOCSTRING(structfun) +function is applied to each field of the structure. @xref{doc-structfun}. Alternatively, to process the data in a structure, the structure might be converted to another type of container before being treated. @@ -887,9 +885,7 @@ is to iterate through it using one or more @code{for} loops. The same idea can be implemented more easily through the use of the @code{cellfun} function that calls a user-specified function on all elements of a cell -array. - -@DOCSTRING(cellfun) +array. @xref{doc-cellfun}. An alternative is to convert the data to a different container, such as a matrix or a data structure. Depending on the data this is possible
--- a/doc/interpreter/expr.txi Tue Dec 27 15:08:21 2011 -0500 +++ b/doc/interpreter/expr.txi Tue Dec 27 18:47:18 2011 -0500 @@ -512,7 +512,8 @@ @cindex complex-conjugate transpose The following arithmetic operators are available, and work on scalars -and matrices. +and matrices. They element-by-element operators and functions broadcast +(@pxref{Broadcasting}). @table @asis @item @var{x} + @var{y} @@ -715,7 +716,8 @@ All of Octave's comparison operators return a value of 1 if the comparison is true, or 0 if it is false. For matrix values, they all -work on an element-by-element basis. For example: +work on an element-by-element basis. Broadcasting rules apply. +@xref{Broadcasting}. For example: @example @group @@ -725,9 +727,9 @@ @end group @end example -If one operand is a scalar and the other is a matrix, the scalar is -compared to each element of the matrix in turn, and the result is the -same size as the matrix. +According to broadcasting rules, if one operand is a scalar and the +other is a matrix, the scalar is compared to each element of the matrix +in turn, and the result is the same size as the matrix. @table @code @item @var{x} < @var{y} @@ -862,8 +864,8 @@ @var{boolean} is false. @end table -For matrix operands, these operators work on an element-by-element -basis. For example, the expression +These operators work on an element-by-element basis. For example, the +expression @example [1, 0; 0, 1] & [1, 0; 2, 3] @@ -872,8 +874,8 @@ @noindent returns a two by two identity matrix. -For the binary operators, the dimensions of the operands must conform if -both are matrices. If one of the operands is a scalar and the other a +For the binary operators, broadcasting rules apply. @xref{Broadcasting}. +In particular, if one of the operands is a scalar and the other a matrix, the operator is applied to the scalar and each element of the matrix.
--- a/doc/interpreter/func.txi Tue Dec 27 15:08:21 2011 -0500 +++ b/doc/interpreter/func.txi Tue Dec 27 18:47:18 2011 -0500 @@ -1225,8 +1225,6 @@ @DOCSTRING(formula) -@DOCSTRING(vectorize) - @DOCSTRING(symvar) @node Commands
--- a/doc/interpreter/matrix.txi Tue Dec 27 15:08:21 2011 -0500 +++ b/doc/interpreter/matrix.txi Tue Dec 27 18:47:18 2011 -0500 @@ -29,7 +29,6 @@ @menu * Finding Elements and Checking Conditions:: * Rearranging Matrices:: -* Applying a Function to an Array:: * Special Utility Matrices:: * Famous Matrices:: @end menu @@ -140,13 +139,6 @@ @DOCSTRING(blkdiag) -@node Applying a Function to an Array -@section Applying a Function to an Array - -@DOCSTRING(arrayfun) - -@DOCSTRING(bsxfun) - @node Special Utility Matrices @section Special Utility Matrices
--- a/doc/interpreter/octave.texi Tue Dec 27 15:08:21 2011 -0500 +++ b/doc/interpreter/octave.texi Tue Dec 27 18:47:18 2011 -0500 @@ -197,8 +197,9 @@ * Plotting:: * Matrix Manipulation:: * Arithmetic:: -* Linear Algebra:: -* Nonlinear Equations:: +* Linear Algebra:: +* Vectorization and Faster Code Execution:: +* Nonlinear Equations:: * Diagonal and Permutation Matrices:: * Sparse Matrices:: * Numerical Integration:: @@ -600,7 +601,6 @@ * Finding Elements and Checking Conditions:: * Rearranging Matrices:: -* Applying a Function to an Array:: * Special Utility Matrices:: * Famous Matrices:: @@ -624,6 +624,15 @@ * Functions of a Matrix:: * Specialized Solvers:: +Vectorization and Faster Code Execution + +* Basic Vectorization:: +* Broadcasting:: +* Function Application:: +* Accumulation:: +* Miscellaneous Techniques:: +* Examples:: + Nonlinear Equations * Solvers:: @@ -844,7 +853,6 @@ Tips and Standards * Style Tips:: Writing clean and robust programs. -* Coding Tips:: Making code run faster. * Comment Tips:: Conventions for writing comments. * Function Headers:: Standard headers for functions. * Documentation Tips:: Writing readable documentation strings. @@ -910,6 +918,7 @@ @include matrix.texi @include arith.texi @include linalg.texi +@include vectorize.texi @include nonlin.texi @include diagperm.texi @include sparse.texi
--- a/doc/interpreter/sparse.txi Tue Dec 27 15:08:21 2011 -0500 +++ b/doc/interpreter/sparse.txi Tue Dec 27 18:47:18 2011 -0500 @@ -223,8 +223,6 @@ @DOCSTRING(speye) -@DOCSTRING(spfun) - @DOCSTRING(spones) @DOCSTRING(sprand)
--- a/doc/interpreter/tips.txi Tue Dec 27 15:08:21 2011 -0500 +++ b/doc/interpreter/tips.txi Tue Dec 27 18:47:18 2011 -0500 @@ -28,7 +28,6 @@ @menu * Style Tips:: Writing clean and robust programs. -* Coding Tips:: Making code run faster. * Comment Tips:: Conventions for writing comments. * Function Headers:: Standard headers for functions. * Documentation Tips:: Writing readable documentation strings. @@ -75,203 +74,6 @@ copyright to anyone else, then place your name in the copyright notice. @end itemize -@node Coding Tips -@section Tips for Making Code Run Faster. -@cindex execution speed -@cindex speedups - -Here are some ways of improving the execution speed of Octave programs. - -@itemize @bullet -@item -Vectorize loops. For instance, rather than - -@example -@group -for i = 1:n-1 - a(i) = b(i+1) - b(i); -endfor -@end group -@end example - -@noindent -write - -@example -a = b(2:n) - b(1:n-1); -@end example - -This is especially important for loops with "cheap" bodies. Often it suffices -to vectorize just the innermost loop to get acceptable performance. A general -rule of thumb is that the "order" of the vectorized body should be greater or -equal to the "order" of the enclosing loop. - -@item -Use built-in and library functions if possible. Built-in and compiled functions -are very fast. Even with a m-file library function, chances are good that it is -already optimized, or will be optimized more in a future release. - -For instance, even better than - -@example -a = b(2:n) - b(1:n-1); -@end example - -@noindent -is - -@example -a = diff (b); -@end example - - -@item -Avoid computing costly intermediate results multiple times. Octave currently -does not eliminate common subexpressions. -Also, certain internal computation results are cached for variables. -For instance, if a matrix variable is used multiple times as an index, -checking the indices (and internal conversion to integers) is only done once. - -@item -Be aware of lazy copies (copy-on-write). When a copy of an object -is created, the data is not immediately copied, but rather shared. The actual -copying is postponed until the copied data needs to be modified. For example: - -@example -@group -a = zeros (1000); # create a 1000x1000 matrix -b = a; # no copying done here -b(1) = 1; # copying done here -@end group -@end example - -Lazy copying applies to whole Octave objects such as matrices, cells, struct, -and also individual cell or struct elements (not array elements). - -Additionally, index expressions also use lazy copying when Octave can determine -that the indexed portion is contiguous in memory. For example: - -@example -@group -a = zeros (1000); # create a 1000x1000 matrix -b = a(:,10:100); # no copying done here -b = a(10:100,:); # copying done here -@end group -@end example - -This applies to arrays (matrices), cell arrays, and structs indexed using (). -Index expressions generating cs-lists can also benefit of shallow copying -in some cases. In particular, when @var{a} is a struct array, expressions like -@code{@{a.x@}, @{a(:,2).x@}} will use lazy copying, so that data can be shared -between a struct array and a cell array. - -Most indexing expressions do not live longer than their `parent' objects. -In rare cases, however, a lazily copied slice outlasts its parent, in which -case it becomes orphaned, still occupying unnecessarily more memory than needed. -To provide a remedy working in most real cases, -Octave checks for orphaned lazy slices at certain situations, when a value -is stored into a "permanent" location, such as a named variable or cell or -struct element, and possibly economizes them. For example: - -@example -@group -a = zeros (1000); # create a 1000x1000 matrix -b = a(:,10:100); # lazy slice -a = []; # the original a array is still allocated -c@{1@} = b; # b is reallocated at this point -@end group -@end example - -@item -Avoid deep recursion. Function calls to m-file functions carry a relatively -significant overhead, so rewriting a recursion as a loop often helps. Also, -note that the maximum level of recursion is limited. - -@item -Avoid resizing matrices unnecessarily. When building a single result -matrix from a series of calculations, set the size of the result matrix -first, then insert values into it. Write - -@example -@group -result = zeros (big_n, big_m) -for i = over:and_over - r1 = @dots{} - r2 = @dots{} - result (r1, r2) = new_value (); -endfor -@end group -@end example - -@noindent -instead of - -@example -@group -result = []; -for i = ever:and_ever - result = [ result, new_value() ]; -endfor -@end group -@end example - -Sometimes the number of items can't be computed in advance, and stack-like -operations are needed. When elements are being repeatedly inserted at/removed -from the end of an array, Octave detects it as stack usage and attempts to use a -smarter memory management strategy pre-allocating the array in bigger chunks. -Likewise works for cell and struct arrays. - -@example -@group -a = []; -while (condition) - @dots{} - a(end+1) = value; # "push" operation - @dots{} - a(end) = []; # "pop" operation - @dots{} -endwhile -@end group -@end example - -@item -Use @code{cellfun} intelligently. The @code{cellfun} function is a useful tool -for avoiding loops. @xref{Processing Data in Cell Arrays}. -@code{cellfun} is often used with anonymous function handles; however, calling -an anonymous function involves an overhead quite comparable to the overhead -of an m-file function. Passing a handle to a built-in function is faster, -because the interpreter is not involved in the internal loop. For example: - -@example -@group -a = @{@dots{}@} -v = cellfun (@@(x) det(x), a); # compute determinants -v = cellfun (@@det, a); # faster -@end group -@end example - -@item -Octave includes a number of other functions that can replace common types of -loops, including @code{bsxfun}, @code{arrayfun}, @code{structfun}, -@code{accumarray}. These functions can take an arbitrary function as a handle. -Be sure to get familiar with them if you want to become an Octave expert. - -@item -Avoid calling @code{eval} or @code{feval} excessively, because -they require Octave to parse input or look up the name of a function in -the symbol table. - -If you are using @code{eval} as an exception handling mechanism and not -because you need to execute some arbitrary text, use the @code{try} -statement instead. @xref{The @code{try} Statement}. - -@item -If you are calling lots of functions but none of them will need to -change during your run, set the variable -@code{ignore_function_time_stamp} to @code{"all"} so that Octave doesn't -waste a lot of time checking to see if you have updated your function -files. -@end itemize @node Comment Tips @section Tips on Writing Comments
--- /dev/null Thu Jan 01 00:00:00 1970 +0000 +++ b/doc/interpreter/vectorize.txi Tue Dec 27 18:47:18 2011 -0500 @@ -0,0 +1,625 @@ +@c Copyright (C) 2011 Jordi Gutiérrez Hermoso +@c +@c This file is part of Octave. +@c +@c Octave is free software; you can redistribute it and/or modify it +@c under the terms of the GNU General Public License as published by the +@c Free Software Foundation; either version 3 of the License, or (at +@c your option) any later version. +@c +@c Octave is distributed in the hope that it will be useful, but WITHOUT +@c ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or +@c FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License +@c for more details. +@c +@c You should have received a copy of the GNU General Public License +@c along with Octave; see the file COPYING. If not, see +@c <http://www.gnu.org/licenses/>. + +@node Vectorization and Faster Code Execution +@chapter Vectorization and Faster Code Execution +@cindex vectorization +@cindex vectorize + +Vectorization is a programming technique that uses vector operations +instead of element-by-element loop-based operations. Besides frequently +writing more succinct code, vectorization also allows for better +optimization of the code. The optimizations may occur either in Octave's +own Fortran, C, or C++ internal implementation, or even at a lower level +depending on the compiler and external numerical libraries used to +build Octave. The ultimate goal is to make use of your hardware's vector +instructions if possible or to perform other optimizations in software. + +Vectorization is not a concept unique to Octave, but is particularly +important because Octave is a matrix-oriented language. Vectorized +Octave code will see a dramatic speed up in most cases. + +This chapter discusses vectorization and other techniques for faster +code execution. + +@menu +* Basic Vectorization:: Basic techniques for code optimization +* Broadcasting:: Broadcasting operations +* Function Application:: Applying functions to arrays, cells, and structs +* Accumulation:: Accumulation functions +* Miscellaneous Techniques:: Other techniques for speeding up code +* Examples:: +@end menu + +@node Basic Vectorization +@section Basic Vectorization + +To a very good first approximation, the goal in vectorization is to +write code that avoids loops and uses whole-array operations. As a +trivial example, consider + +@example +@group +for i = 1:n + for j = 1:m + c(i,j) = a(i,j) + b(i,j); + endfor +endfor +@end group +@end example + +@noindent +compared to the much simpler + +@example +c = a + b; +@end example + +@noindent +This isn't merely easier to write; it is also internally much easier to +optimize. Octave delegates this operation to an underlying +implementation which among other optimizations may use special vector +hardware instructions or could conceivably even perform the additions in +parallel. In general, if the code is vectorized, the underlying +implementation has much more freedom about the assumptions it can make +in order to achieve faster execution. + +This is especially important for loops with "cheap" bodies. Often it +suffices to vectorize just the innermost loop to get acceptable +performance. A general rule of thumb is that the "order" of the +vectorized body should be greater or equal to the "order" of the +enclosing loop. + +As a less trivial example, rather than + +@example +@group +for i = 1:n-1 + a(i) = b(i+1) - b(i); +endfor +@end group +@end example + +@noindent +write + +@example +a = b(2:n) - b(1:n-1); +@end example + +This shows an important general concept about using arrays for indexing +instead of looping over an index variable. @xref{Index Expressions}. +Also use boolean indexing generously. If a condition needs to be tested, +this condition can also be written as a boolean index. For instance, +instead of + +@example +@group +for i = 1:n + if a(i) > 5 + a(i) -= 20 + endif +endfor +@end group +@end example + +@noindent +write + +@example +a(a>5) -= 20; +@end example + +@noindent +which exploits the fact that @code{a > 5} produces a boolean index. + +Use elementwise vector operators whenever possible to avoid looping +(operators like @code{.*} and @code{.^}). @xref{Arithmetic Ops}. For +simple in-line functions, the @code{vectorize} function can do this +automatically. + +@DOCSTRING(vectorize) + +Also exploit broadcasting in these elementise operators both to avoid +looping and unnecessary intermediate memory allocations. +@xref{Broadcasting}. + +Use built-in and library functions if possible. Built-in and compiled +functions are very fast. Even with a m-file library function, chances +are good that it is already optimized, or will be optimized more in a +future release. + +For instance, even better than + +@example +a = b(2:n) - b(1:n-1); +@end example + +@noindent +is + +@example +a = diff (b); +@end example + +Most Octave functions are written with vector and array arguments in +mind. If you find yourself writing a loop with a very simple operation, +chances are that such a function already exists. The following functions +occur frequently in vectorized code: + +@itemize @bullet +@item +Index manipulation + +@itemize +@item +find +@item +sub2ind +@item +ind2sub +@item +sort +@item +unique +@item +lookup +@item +ifelse / merge +@end itemize + +@item +Repetition +@itemize +@item +repmat +@item +repelems +@end itemize + +@item +Vectorized arithmetic +@itemize +@item +sum +@item +prod +@item +cumsum +@item +cumprod +@item +sumsq +@item +diff +@item +dot +@item +cummax +@item +cummin +@end itemize + +@item +Shape of higher dimensional arrays +@itemize +@item +reshape +@item +resize +@item +permute +@item +squeeze +@item +deal +@end itemize + +@end itemize + +@node Broadcasting +@section Broadcasting +@cindex broadcast +@cindex broadcasting +@cindex BSX +@cindex recycling +@cindex SIMD + +Broadcasting refers to how Octave binary operators and functions behave +when their matrix or array operands or arguments differ in size. Since +version 3.6.0, Octave now automatically broadcasts vectors, matrices, +and arrays when using elementwise binary operators and functions. +Broadly speaking, smaller arrays are ``broadcast'' across the larger +one, until they have a compatible shape. The rule is that corresponding +array dimensions must either + +@enumerate +@item +be equal or, +@item +one of them must be 1. +@end enumerate + +@noindent +In case all dimensions are equal, no broadcasting occurs and ordinary +element-by-element arithmetic takes place. For arrays of higher +dimensions, if the number of dimensions isn't the same, then missing +trailing dimensions are treated as 1. When one of the dimensions is 1, +the array with that singleton dimension gets copied along that dimension +until it matches the dimension of the other array. For example, consider + +@example +@group +x = [1 2 3; + 4 5 6; + 7 8 9]; + +y = [10 20 30]; + +x + y +@end group +@end example + +@noindent +Without broadcasting, @code{x + y} would be an error because dimensions +do not agree. However, with broadcasting it is as if the following +operation were performed: + +@example +@group +x = [1 2 3 + 4 5 6 + 7 8 9]; + +y = [10 20 30 + 10 20 30 + 10 20 30]; + +x + y +@result{} 11 22 33 + 14 25 36 + 17 28 39 +@end group +@end example + +@noindent +That is, the smaller array of size @code{[1 3]} gets copied along the +singleton dimension (the number of rows) until it is @code{[3 3]}. No +actual copying takes place, however. The internal implementation reuses +elements along the necessary dimension in order to achieve the desired +effect without copying in memory. + +Both arrays can be broadcast across each other, for example, all +pairwise differences of the elements of a vector with itself: + +@example +@group +y - y' +@result{} 0 10 20 + -10 0 10 + -20 -10 0 +@end group +@end example + +@noindent +Here the vectors of size @code{[1 3]} and @code{[3 1]} both get +broadcast into matrices of size @code{[3 3]} before ordinary matrix +subtraction takes place. + +For a higher-dimensional example, suppose @code{img} is an RGB image of +size @code{[m n 3]} and we wish to multiply each colour by a different +scalar. The following code accomplishes this with broadcasting, + +@example +img .*= permute ([0.8, 0.9, 1.2], [1, 3, 2]); +@end example + +@noindent +Note the usage of permute to match the dimensions of the @code{[0.8, +0.9, 1.2]} vector with @code{img}. + +For functions that are not written with broadcasting semantics, +@code{bsxfun} can be useful for coercing them to broadcast. + +@DOCSTRING(bsxfun) + +Broadcasting is only applied if either of the two broadcasting +conditions hold. As usual, however, broadcasting does not apply when two +dimensions differ and neither is 1: + +@example +@group +x = [1 2 3 + 4 5 6]; +y = [10 20 + 30 40]; +x + y +@end group +@end example + +@noindent +This will produce an error about nonconformant arguments. + +Besides common arithmetic operations, several functions of two arguments +also broadcast. The full list of functions and operators that broadcast +is + +@example +@group + plus + .+ + minus - .- + times .* + rdivide ./ + ldivide .\ + power .^ .** + lt < + le <= + eq == + gt > + ge >= + ne != ~= + and & + or | + atan2 + hypot + max + min + mod + rem + xor + + += -= .+= .-= .*= ./= .\= .^= .**= &= |= +@end group +@end example + +Beware of resorting to broadcasting if a simpler operation will suffice. +For matrices @var{a} and @var{b}, consider the following: + +@example +c = sum (permute (a, [1, 3, 2]) .* permute (b, [3, 2, 1]), 3); +@end example + +@noindent +This operation broadcasts the two matrices with permuted dimensions +across each other during elementwise multiplication in order to obtain a +larger 3d array, and this array is the summed along the third dimension. +A moment of thought will prove that this operation is simply the much +faster ordinary matrix multiplication, @code{c = a*b;}. + +A note on terminology: ``broadcasting'' is the term popularized by the +Numpy numerical environment in the Python programming language. In other +programming languages and environments, broadcasting may also be known +as @emph{binary singleton expansion} (BSX, in @sc{Matlab}, and the +origin of the name of the @code{bsxfun} function), @emph{recycling} (R +programming language), @emph{single-instruction multiple data} (SIMD), +or @emph{replication}. + +@subsection Broadcasting and Legacy Code + +The new broadcasting semantics do not affect almost any code that worked +in previous versions of Octave without error. Thus for example all code +inherited from @sc{Matlab} that worked in previous versions of Octave +should still work without change in Octave. The only exception is code +such as + +@example +@group +try + c = a.*b; +catch + c = a.*a; +end_try_catch +@end group +@end example + +@noindent +that may have relied on matrices of different size producing an error. +Due to how broadcasting changes semantics with older versions of Octave, +by default Octave warns if a broadcasting operation is performed. To +disable this warning, refer to its ID (@pxref{doc-warning_ids}): + +@example +warning ("off", "Octave:broadcast"); +@end example + +@noindent +If you want to recover the old behaviour and produce an error, turn this +warning into an error: + +@example +warning ("error", "Octave:broadcast"); +@end example + +@node Function Application +@section Function Application +@cindex map +@cindex apply +@cindex function application + +As a general rule, functions should already be written with matrix +arguments in mind and should consider whole matrix operations in a +vectorized manner. Sometimes, writing functions in this way appears +difficult or impossible for various reasons. For those situations, +Octave provides facilities for applying a function to each element of an +array, cell, or struct. + +@DOCSTRING(arrayfun) +@DOCSTRING(spfun) +@DOCSTRING(cellfun) +@DOCSTRING(structfun) + +@node Accumulation +@section Accumulation + +Whenever it's possible to categorize according to indices the elements +of an array when performing a computation, accumulation functions can be +useful. + +@DOCSTRING(accumarray) + +@DOCSTRING(accumdim) + +@node Miscellaneous Techniques +@section Miscellaneous Techniques +@cindex execution speed +@cindex speedups +@cindex optimization + +Here are some other ways of improving the execution speed of Octave +programs. + +@itemize @bullet + +@item +Avoid computing costly intermediate results multiple times. Octave +currently does not eliminate common subexpressions. Also, certain +internal computation results are cached for variables. For instance, if +a matrix variable is used multiple times as an index, checking the +indices (and internal conversion to integers) is only done once. + +@item +@cindex copy-on-write +@cindex COW +@cindex memory management +Be aware of lazy copies (copy-on-write). When a copy of an object is +created, the data is not immediately copied, but rather shared. The +actual copying is postponed until the copied data needs to be modified. +For example: + +@example +@group +a = zeros (1000); # create a 1000x1000 matrix +b = a; # no copying done here +b(1) = 1; # copying done here +@end group +@end example + +Lazy copying applies to whole Octave objects such as matrices, cells, +struct, and also individual cell or struct elements (not array +elements). + +Additionally, index expressions also use lazy copying when Octave can +determine that the indexed portion is contiguous in memory. For example: + +@example +@group +a = zeros (1000); # create a 1000x1000 matrix +b = a(:,10:100); # no copying done here +b = a(10:100,:); # copying done here +@end group +@end example + +This applies to arrays (matrices), cell arrays, and structs indexed +using (). Index expressions generating cs-lists can also benefit of +shallow copying in some cases. In particular, when @var{a} is a struct +array, expressions like @code{@{a.x@}, @{a(:,2).x@}} will use lazy +copying, so that data can be shared between a struct array and a cell +array. + +Most indexing expressions do not live longer than their `parent' +objects. In rare cases, however, a lazily copied slice outlasts its +parent, in which case it becomes orphaned, still occupying unnecessarily +more memory than needed. To provide a remedy working in most real cases, +Octave checks for orphaned lazy slices at certain situations, when a +value is stored into a "permanent" location, such as a named variable or +cell or struct element, and possibly economizes them. For example: + +@example +@group +a = zeros (1000); # create a 1000x1000 matrix +b = a(:,10:100); # lazy slice +a = []; # the original a array is still allocated +c@{1@} = b; # b is reallocated at this point +@end group +@end example + +@item +Avoid deep recursion. Function calls to m-file functions carry a +relatively significant overhead, so rewriting a recursion as a loop +often helps. Also, note that the maximum level of recursion is limited. + +@item +Avoid resizing matrices unnecessarily. When building a single result +matrix from a series of calculations, set the size of the result matrix +first, then insert values into it. Write + +@example +@group +result = zeros (big_n, big_m) +for i = over:and_over + r1 = @dots{} + r2 = @dots{} + result (r1, r2) = new_value (); +endfor +@end group +@end example + +@noindent +instead of + +@example +@group +result = []; +for i = ever:and_ever + result = [ result, new_value() ]; +endfor +@end group +@end example + +Sometimes the number of items can't be computed in advance, and +stack-like operations are needed. When elements are being repeatedly +inserted at/removed from the end of an array, Octave detects it as stack +usage and attempts to use a smarter memory management strategy +pre-allocating the array in bigger chunks. Likewise works for cell and +struct arrays. + +@example +@group +a = []; +while (condition) + @dots{} + a(end+1) = value; # "push" operation + @dots{} + a(end) = []; # "pop" operation + @dots{} +endwhile +@end group +@end example + +@item +Avoid calling @code{eval} or @code{feval} excessively, because +they require Octave to parse input or look up the name of a function in +the symbol table. + +If you are using @code{eval} as an exception handling mechanism and not +because you need to execute some arbitrary text, use the @code{try} +statement instead. @xref{The @code{try} Statement}. + +@item +If you are calling lots of functions but none of them will need to +change during your run, set the variable +@code{ignore_function_time_stamp} to @code{"all"} so that Octave doesn't +waste a lot of time checking to see if you have updated your function +files. +@end itemize + +@node Examples +@section Examples + +Here goes a gallery of vectorization puzzles with solutions culled from +the help mailing list and the machine learning class...
--- a/liboctave/bsxfun.h Tue Dec 27 15:08:21 2011 -0500 +++ b/liboctave/bsxfun.h Tue Dec 27 18:47:18 2011 -0500 @@ -42,7 +42,7 @@ } (*current_liboctave_warning_with_id_handler) - ("Octave:auto-bsxfun", "%s: automatic broadcasting operation applied", name.c_str ()); + ("Octave:broadcast", "%s: automatic broadcasting operation applied", name.c_str ()); return true; } @@ -69,7 +69,7 @@ } (*current_liboctave_warning_with_id_handler) - ("Octave:auto-bsxfun", "%s: automatic broadcasting operation applied", name.c_str ()); + ("Octave:broadcast", "%s: automatic broadcasting operation applied", name.c_str ()); return true; }
--- a/scripts/general/accumarray.m Tue Dec 27 15:08:21 2011 -0500 +++ b/scripts/general/accumarray.m Tue Dec 27 18:47:18 2011 -0500 @@ -26,14 +26,16 @@ ## the rows of the matrix @var{subs} and the values by @var{vals}. Each ## row of @var{subs} corresponds to one of the values in @var{vals}. If ## @var{vals} is a scalar, it will be used for each of the row of -## @var{subs}. +## @var{subs}. If @var{subs} is a cell array of vectors, all vectors +## must be of the same length, and the subscripts in the @var{k}th +## vector must correspond to the @var{k}th dimension of the result. ## ## The size of the matrix will be determined by the subscripts -## themselves. However, if @var{sz} is defined it determines the matrix -## size. The length of @var{sz} must correspond to the number of columns +## themselves. However, if @var{sz} is defined it determines the matrix +## size. The length of @var{sz} must correspond to the number of columns ## in @var{subs}. An exception is if @var{subs} has only one column, in -## which case @var{sz} may be the dimensions of a vector and the subscripts -## of @var{subs} are taken as the indices into it. +## which case @var{sz} may be the dimensions of a vector and the +## subscripts of @var{subs} are taken as the indices into it. ## ## The default action of @code{accumarray} is to sum the elements with ## the same subscripts. This behavior can be modified by defining the @@ -107,14 +109,27 @@ print_usage (); endif + lenvals = length (vals); + if (iscell (subs)) - subs = cellfun ("vec", subs, "uniformoutput", false); + subs = cellfun (@vec, subs, "uniformoutput", false); ndims = numel (subs); if (ndims == 1) subs = subs{1}; endif + + lensubs = cellfun (@length, subs); + + if (any (lensubs != lensubs(1)) || + (lenvals > 1 && lenvals != lensubs(1))) + error ("accumarray: dimension mismatch"); + endif + else ndims = columns (subs); + if (lenvals > 1 && lenvals != rows (subs)) + error ("accumarray: dimension mismatch") + endif endif if (isempty (fillval)) @@ -127,7 +142,8 @@ if (issparse) - ## Sparse case. Avoid linearizing the subscripts, because it could overflow. + ## Sparse case. Avoid linearizing the subscripts, because it could + ## overflow. if (fillval != 0) error ("accumarray: FILLVAL must be zero in the sparse case"); @@ -153,8 +169,8 @@ if (! (isempty (func) || func == @sum)) - ## Reduce values. This is not needed if we're about to sum them, because - ## "sparse" can do that. + ## Reduce values. This is not needed if we're about to sum them, + ## because "sparse" can do that. ## Sort indices. [subs, idx] = sortrows (subs); @@ -238,7 +254,8 @@ elseif (islogical (vals)) zero = false; elseif (fillval == 0 && all (vals(:) >= 0)) - ## This is a common case - fillval is zero, all numbers nonegative. + ## This is a common case - fillval is zero, all numbers + ## nonegative. zero = 0; else zero = NaN; # Neutral value. @@ -264,7 +281,8 @@ elseif (islogical (vals)) zero = true; elseif (fillval == 0 && all (vals(:) <= 0)) - ## This is a common case - fillval is zero, all numbers non-positive. + ## This is a common case - fillval is zero, all numbers + ## non-positive. zero = 0; else zero = NaN; # Neutral value. @@ -298,14 +316,21 @@ jdx = find (subs(1:n-1) != subs(2:n)); jdx = [jdx; n]; vals = mat2cell (vals(idx), diff ([0; jdx])); - ## Optimize the case when function is @(x) {x}, i.e. we just want to - ## collect the values to cells. + ## Optimize the case when function is @(x) {x}, i.e. we just want + ## to collect the values to cells. persistent simple_cell_str = func2str (@(x) {x}); if (! strcmp (func2str (func), simple_cell_str)) vals = cellfun (func, vals); endif subs = subs(jdx); + if (isempty (sz)) + sz = max (subs); + if (length (sz) == 1) + sz(2) = 1; + endif + endif + ## Construct matrix of fillvals. if (iscell (vals)) A = cell (sz); @@ -328,6 +353,7 @@ %!assert (accumarray ([1,1,1;2,1,2;2,3,2;2,1,2;2,3,2],101:105,[],@(x)sin(sum(x))),sin(cat(3, [101,0,0;0,0,0],[0,0,0;206,0,208]))) %!assert (accumarray ({[1 3 3 2 3 1 2 2 3 3 1 2],[3 4 2 1 4 3 4 2 2 4 3 4],[1 1 2 2 1 1 2 1 1 1 2 2]},101:112),cat(3,[0,0,207,0;0,108,0,0;0,109,0,317],[0,0,111,0;104,0,0,219;0,103,0,0])) %!assert (accumarray ([1,1;2,1;2,3;2,1;2,3],101:105,[2,4],@max,NaN),[101,NaN,NaN,NaN;104,NaN,105,NaN]) +%!assert (accumarray ([1 1; 2 1; 2 3; 2 1; 2 3],101:105, [], @prod), [101, 0, 0; 10608, 0, 10815]) %!assert (accumarray ([1 1; 2 1; 2 3; 2 1; 2 3],101:105,[2 4],@prod,0,true),sparse([1,2,2],[1,1,3],[101,10608,10815],2,4)) %!assert (accumarray ([1 1; 2 1; 2 3; 2 1; 2 3],1,[2,4]), [1,0,0,0;2,0,2,0]) %!assert (accumarray ([1 1; 2 1; 2 3; 2 1; 2 3],101:105,[2,4],@(x)length(x)>1),[false,false,false,false;true,false,true,false])
--- a/scripts/general/accumdim.m Tue Dec 27 15:08:21 2011 -0500 +++ b/scripts/general/accumdim.m Tue Dec 27 18:47:18 2011 -0500 @@ -22,32 +22,37 @@ ## positions defined by their subscripts along a specified dimension. ## The subscripts are defined by the index vector @var{subs}. ## The dimension is specified by @var{dim}. If not given, it defaults -## to the first non-singleton dimension. +## to the first non-singleton dimension. The length of @var{subs} must +## be equal to @code{size (@var{vals}, @var{dim})}. ## -## The extent of the result matrix in the working dimension will be determined -## by the subscripts themselves. -## However, if @var{n} is defined it determines this extent. +## The extent of the result matrix in the working dimension will be +## determined by the subscripts themselves. However, if @var{n} is +## defined it determines this extent. ## ## The default action of @code{accumdim} is to sum the subarrays with the -## same subscripts. This behavior can be modified by defining the @var{func} -## function. This should be a function or function handle that accepts an -## array and a dimension, and reduces the array along this dimension. -## As a special exception, the built-in @code{min} and @code{max} functions -## can be used directly, and @code{accumdim} accounts for the middle empty -## argument that is used in their calling. +## same subscripts. This behavior can be modified by defining the +## @var{func} function. This should be a function or function handle +## that accepts an array and a dimension, and reduces the array along +## this dimension. As a special exception, the built-in @code{min} and +## @code{max} functions can be used directly, and @code{accumdim} +## accounts for the middle empty argument that is used in their calling. ## -## The slices of the returned array that have no subscripts associated with -## them are set to zero. Defining @var{fillval} to some other value allows -## these values to be defined. +## The slices of the returned array that have no subscripts associated +## with them are set to zero. Defining @var{fillval} to some other +## value allows these values to be defined. ## ## An example of the use of @code{accumdim} is: ## -## @smallexample +## @example ## @group -## accumdim ([1, 2, 1, 2, 1], [7,-10,4;-5,-12,8;-12,2,8;-10,9,-3;-5,-3,-13]) +## accumdim ([1, 2, 1, 2, 1], [ 7, -10, 4; +## -5, -12, 8; +## -12, 2, 8; +## -10, 9, -3; +## -5, -3, -13]) ## @result{} ans = [-10,-11,-1;-15,-3,5] ## @end group -## @end smallexample +## @end example ## ## @seealso{accumarray} ## @end deftypefn @@ -68,7 +73,7 @@ error ("accumdim: indices must be positive integers"); else m = max (subs); - if (n == 0) + if (n == 0 || isempty (n)) n = m; elseif (n < m) error ("accumdim: N index out of range"); @@ -86,6 +91,10 @@ endif sz(dim) = n; + if (length (subs) != size (vals, dim)) + error ("accumdim: dimension mismatch") + endif + if (isempty (func) || func == @sum) ## Fast summation case. A = __accumdim_sum__ (subs, vals, dim, n);
--- /dev/null Thu Jan 01 00:00:00 1970 +0000 +++ b/scripts/miscellaneous/fact.m Tue Dec 27 18:47:18 2011 -0500 @@ -0,0 +1,269 @@ +## Copyright (C) 2007-2011 Jordi Gutiérrez Hermoso +## Copyright (C) 2007 Stallmanfacts.com +## +## This file is part of Octave. +## +## Octave is free software; you can redistribute it and/or modify it +## under the terms of the GNU General Public License as published by +## the Free Software Foundation; either version 3 of the License, or (at +## your option) any later version. +## +## Octave is distributed in the hope that it will be useful, but +## WITHOUT ANY WARRANTY; without even the implied warranty of +## MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU +## General Public License for more details. +## +## You should have received a copy of the GNU General Public License +## along with Octave; see the file COPYING. If not, see +## <http://www.gnu.org/licenses/>. + +## -*- texinfo -*- +## @deftypefn {Command} {} fact +## @deftypefnx {Function File} {T =} fact() +## Display an amazing and random fact about the world's greatest hacker. +## @end deftypefn + + +function f = fact() + persistent wisdom = \ + { + "Richard Stallman takes notes in binary."; + "Richard Stallman doesn't need sudo. I will make him a sandwich anyway."; + "Richard Stallman is my shephurd, and I am his GNU."; + "Richard Stallman doesn't wget, Richard Stallman wdemands!"; + "Richard Stallman can touch MC Hammer"; + "Richard Stallman doesn't read web pages. They write to him."; + "Richard Stallman gets 9 bits to the byte."; + "Richard Stallman doesn't really believe in open software, because it's not free enough."; + "Richard Stallman can leave neutral or negative feedback on eBay."; + "Richard Stallman is the only man alive who can pronounce GNU the way it is meant to be pronounced."; + "Richard Stallman does not own a mobile phone because he can fashion a crude convex dish and shout into it at the exact resonant frequency of the ozone, causing a voice to seemingly come from the sky above his intended recipient."; + "Richard Stallman is so handsome that when he was young he was responsible for all other geeks not being able to get girls. This is why he has to cover his face with a thick layer of hair."; + "Some people check their computers for viruses. Viruses check their computers for Richard Stallman."; + "Richard Stallman memorises all his documents. In binary. He just types everything in whenever he needs a document."; + "When Richard Stallman makes a sudo command, he loses permissions."; + "Richard Stallman's beard is made of parentheses."; + "Richard Stallman's DNA is in binary."; + "Richard Stallman's nervous system is completely wireless."; + "Richard Stallman's brain accepts UNIX commands."; + "If Richard Stallman has 1GB of RAM, and if you have 1GB of RAM, Richard Stallman has more RAM than you."; + "Richard Stallman eats ethernet cables. That's why they invented wireless."; + "Richard Stallman has a katana. 'Nuff said."; + "Richard Stallman wrote a program that divides by zero."; + "Ricahrd Stallman doesn't use zip drives, he just squeezes the hard drive."; + "Richard Stallman's compiler is afraid to report errors."; + "Richard Stallman wrote the compiler God used. The Big Bang was the Universe's first segfault."; + "Richard Stallman successfully compiled a kernel of popcorn."; + "Richard Stallman doesn't write programs, they write themselves out of reverence."; + "Richard Stallman can make infinite loops end."; + "Richard Stallman's anti-virus programs cures HIV."; + "Richard Stallman's computer doesn't have a clock; it defines what time it is."; + "Richard Stallman wrote a program to compute the last digit of pi."; + "Richard Stallman doesn't use web browsers. He sends a link to a demon that uses wget to fetch the page and sends it back to him."; + "Richard Stallman can solve the halting problem... in polynomial time."; + "For Richard Stallman, polynomial time is O(1)."; + "Richard Stallman didn't \"write\" Emacs or created it in his own image. Richard Stallman made Emacs an instance of himself."; + "Richard Stallman can coerce meaningful data from /dev/null."; + "Some people wear Linus Torvalds pyjamas to bed, Linus Torvalds wears Richard Stallman pyjamas."; + "There is no software development process, only a bunch of programs Richard Stallman allows to exist. "; + "Richard Stallman spends his leisure time programming with Guile on GNU Hurd. "; + "Richard Stallman's left and right hands are named \"(\" and \")\" "; + "Richard Stallman first words were actually syscalls. "; + "Richard Stallman didn't create the singularity. He is the singularity. GNU/Linux is only the event horizon. "; + "When Richard Stallman pipes to more, he gets less "; + "Richard Stallman never showers: he runs 'make clean'. "; + "Richard Stallman needs neither mouse nor keyboard to operate his computer. He just stares it down until it does what he wants. "; + "Richard Stallman didn't write the GPL. He is the GPL. "; + "Richad Stallman's pinky finger is really a USB memory stick. "; + "Richard Stallman called his operating system GNU because he created it before computers existed, when actual gnus were used for calcuations. "; + "In Soviet Russia, Richard Stallman is still Richard Stallman! "; + "Richard Stallman's flute only plays free music. "; + "When Richard Stallman uses floats, there are no rounding errors."; + "Richard Stallman wrote a program so powerful, it knows the question to 42."; + "Richard Stallman released his own DNA under GNU FDL."; + "Richard Stallman knows the entire wikipedia by heart, markup included."; + "Richard Stallman wrote the HAL9000 OS."; + "Richard Stallman's laser pointer is a lightsaber."; + "Richard Stallman never steps down; he shifts the universe up ."; + "Richard Stallman doesn't maintain code; he stares at it until it fixes itself out of reverence."; + "Richard Stallman doesn't use an editor; he sets the fundamental constants of the universe so that a magnetic platter with his code on it evolves itself."; + "Richard Stallman doesn’t code; he dares the computer to not do his bidding."; + "Global warming is caused by Richard Stallman’s rage toward non-free software."; + "Rather than being birthed like a normal child, Richard Stallman instead instantiated himself polymorphically. Shortly thereafter he grew a beard."; + "Richard Stallman discovered extra-terrestrial life but killed them because they used non-free software."; + "Richard Stallman doesn't evaluate expressions, expressions evaluate to Richard Stallman."; + "Richard Stallman can see Russia from his house."; + "Richard Stallman proved P=NP, twice!"; + "Richard Stallman knows of an unfixed bug in TeX."; + "Richard Stallman can write a context-free grammar for C."; + "Richard Stallman can determine whether an arbitrary program will terminate."; + "Richard Stallman's computer has only two buttons. One is for guests."; + "Richard Stallman does not actually write programs. He comes up with a length and digit index in pi."; + "Richard Stallman's distributed version control system is a flamewar on Usenet."; + "Richard Stallman wrote the first version of Emacs on a typewriter."; + "Richard Stallman has no known weaknesses, except for a phobia against soap."; + "Richard Stallman is not affected by Godwin's Law."; + "Richard Stallman can write an anti-virus program that cures HIV. Too bad he never writes anti-virus programs."; + "Richard Stallman' facial hair is \"free as in beard\""; + "Richard Stallman is licensed under GPL, so you can clone him and redistribute copies so you can help your neighbor. For example a version that take a bath more often."; + "Richard Stallman doesn't code; he just travels around the world."; + "Richard Stallman was coded by himself in lisp with Emacs."; + "Richard Stallman doesn't eat McDonald's because the machine that kills the cows uses proprietary software."; + "There is no chin behind Richard Stallman's legendary beard, there is only another Emacs."; + "In an average living room there are 1,242 objects Richard Stallman could use to write an OS, including the room itself."; + "Vendor lock-in is when vendors lock themselves inside of a building out of fear of Richard Stallman's wrath."; + "When Richard Stallman executes ps -e, you show up."; + "When Richard Stallman gets angry he doesn't swear; he recurses."; + "On Richard Stallman's computer the bootloader is contained in his .emacs."; + "Richard Satallman can make any operating system free, free from drivers."; + "Richard Stallman programmed Chuck Norris."; + "Behind Richard Stallman's beard there is another fist, to code faster."; + "Richard Stallman won a Suduku that started with only one number in each line"; + "Richard Stallman's brain compiles and runs C code."; + "Richard Stallman wrote the first version of Emacs using Emacs."; + "Richard Stallman never gonna give you up, never gonna let you down, never gonna run around and desert you, never gonna make you cry, never gonna say goodbye, never gonna tell a lie and hurt you."; + "Richard Stallman, upon reading these facts, didn't laugh at all. Instead, he complained that he is being linked to that dirty \"open source\" software. He also asked it to be changed to \"free software\", in order to raise awareness for software freedom in our society."; + "Richard Stallman has no problem using Emacs. He wrote it with his 4 hands."; + "Richard Stallman will revert the big rip by adding parenthesis to the dark matter."; + "When you make a Google search and it doesn't find the answer, Google gently consults Richard Stallman."; + "Richard Stallman's uptime is over 53 years. And counting up."; + "Richard Stallman's portable music player plays ogg and WMA."; + "Richard Stallman will never die, but may some day go to /dev/null."; + "Richard Stallman once got swine flu, but it got cleansed by hereditay GPL and thus got assimilated."; + "Richard Stallman don't cut his hair because there are no GNU/Scissors"; + "Richard Stallman is the one who trims Chuck Norris beard. And he does it freely, of course."; + "Richard Stallman does not take bath, for the hydroelectric company uses proprietary software."; + "Agent Smith loves Richard Stallman's scent."; + "Richard Stallman is the One."; + "\"They can take our lives, but they can never take our freedom.\" Willian Wallace after a litle talk with Richard Stallman."; + "Richard Stallman can connect to any brain using an Emacs ssh client."; + "Richard Stallman ported Emacs to Intel 4004 chip."; + "Richard Stallman did not write GNU Emacs, he simply read the source code from /dev/null."; + "Richard Stallman once used GDB to reverse-engineer Windows 7 into a free operating system - able to run on GNU Emacs!"; + "Richard Stallman does not contribute to open source projects; open source projects contribute to Richard Stallman, and then call themselves free software projects."; + "Richard Stallman programmed himself before he could even exist"; + "Richard Stallman can fill up /dev/null."; + "Richard Stallman is so zealous about privacy he has /dev/null as his home."; + "When Richard Stallman runs /bin/false, it returns \"true\"."; + "Richard Stallman doesn't like money, because banks don't run on free software."; + "Richard Stallman user GNU tar to compress air."; + "Richard Stallman was installed in the world, it runs on a free program .."; + "When Richard Stallman reports a bug, the bug prefers to squash itself instead of facing Richard Stallman's wrath."; + "There is no Windows in Richard Stallman's house... only Doors..."; + "Richard Stallman doesn't like neither PCs-Intel nor Burger King... He prefers e-Macs..."; + "Richard Stallman can use grep to find Jimmy Hoffa."; + "Richard Stallman made it possible to not absolutely abhor HPUX."; + "When Richard Stallman pours his alphabets cereal into a bowl, only G's, N's, and U's come out."; + "Richard Stallman is pronounced \"GNU slash Stallman\""; + "Richard Stallman doesn't mind if you read his mail as long as you don't delete it before he reads it."; + "Richard Stallman is just a guy who has strong principles and decided to follow them."; + "Richard Stallman knows that you don't have class because it is a keyword that he defined."; + "Richard Stallman doesn't need a qwerty/dvorak keyboard only two buttons \"1\" and \"0\" and his erect penis."; + "On the first day Richard Stallman said M-x create-light."; + "Richard Stallman once went out of scope for a while. The garbage collector never dared to touch him."; + "Richard Stallman does not compile; he closes his eyes, and see energy lines created between bit blocks by the compiler optimizations..."; + "intx80 first calls Richard Stallman before calling sys_call"; + "Tron is actually a biographical story about Richard Stallman. The director decided to tone it down or audiences wouldn't find it believable."; + "Richard Stallman always wears a red shirt to make sure that whatever attacks his away-team has to go through him first."; + "kill -9 invokes Richard Stallman's rage against a process."; + "If Richard were to stumble upon stallmanfacts.com, he would find it a gnuisance."; + "Richard Stallman can telnet into Mordor."; + "sudo chown Richard:Stallman /all/your/base"; + "Richard Stallman's nervous system is completely wireless."; + "Richard Stallman does not sleep. He yields."; + "Some people say M-x psychoanalyse-pinhead is a merely a program. Others say M-x psychoanalyse-pinhead *is* Richard Stallman. All I know is, Richard Stallman is The Stig."; + "If you execute Emacs backward it either undoes the industrial revolution or induces the rapture. But only Richard Stallman knows which."; + "If Richard Stallman's beard were ever trimmed, the clippings would re-marshal into an exact copy of Richard Stallman."; + "Richard Stallman never sleeps because he altered his own source to gain background garbage collection."; + "Richard Stallman's doctor can retrieve a blood sample via CVS."; + "Richard Stallman can touch this"; + "Because Richard Stallman's DNA is licensed under the FDL, his doctor can't draw his blood without violating HIPAA."; + "Richard Stallman can remove his own appendix, using only gdb."; + "Richard Stallman's DNA includes debugging symbols. But he doesn't need them."; + "Richard Stallman met Chuck Norris once. Chuck tried a roundhouse, but Richard bashed him in the skull."; + "Richard Stallman doesn't need to buy a bigger hard drive. He can compress data infinitely. "; + "When Richard Stallman cannot take your call, his beard answers the phone for you."; + "The R in RMS stands for RMS."; + "Richard Stallman can parse HTML with regular expressions."; + "Richard Stallman's traceroute goes all the way through an infinite number of anonymous proxies back to the traffic's source."; + "Richard Stallman's beard is in fact not a just a beard, but a microprinted hard copy of Emacs source code. New patches must be checked against new hair growth before being approved."; + "In the beginning-of-buffer there was Richard Stallman."; + "The NOOP was created to give Richard Stallman some time to comb his beard."; + "Whenever Richard Stallman looks at a Windows computer, it segfaults. Whenever Richard Stallman doesn't look at a Windows computer, it segfaults."; + "Richard Stallman can walk on Windows!"; + "After being unable to satisfy my wife for years, Richard Stallman was able to single-handedly unlock her orgasm within seconds and managed to write a texinfo manual minutes later for other users."; + "Richard Stallman's tabbed browser is a set to wget/telnet fg/bg processes."; + "There is no chin under Richard Stallman' beard. There's only another beard. Recursively."; + "Stallman can chown anything! stallman@stallman~$ chown stallman:stallman Earth (for example)"; + "Richard Stallman freed his beard so he can always check what's in it."; + "In the beginning was the Word, and the Word was with RMS, and the Word was GNU."; + "RMS means \"RMS means Stallman\""; + "Richard Stallman is the babelfish of his own speeches."; + "Richard Stallman wrote his own library and lives in it."; + "Richard Stallman found Waldo using grep in /dev/null"; + "Richard Stallman doesn't sleep; he is compiling"; + "Richard Stallman will get Coca Cola to release their recipe under the GPL."; + "Richard Stallman doesn't change clothes. He makes case mods."; + "Richard Stallman compiled the first version of gcc with an hexadecimal editor."; + "Richard Stallman will be the last guest on Linux Outlaws"; + "Richard Stallman calculates the universe's entropy by exploiting forced stack overflows."; + "Richard Stallman's consciousness will one day become the singularity, which will create Deep Thought, and answer the meaning of life, the universe and everything."; + "C is actually written in RMS."; + "Richard Stallman can write software that does not have a buffer overflow when counting money lost by Jerome Kerviel."; + "There were no double rainbows before Richard Stallman."; + "Chuck Norris had to shorten his beard in the presence of Stallman because two beards that awsome, so close would segfault the universe (again)."; + "RMS is Titanic."; + "Richard Stallman is the answer to the Turing Test."; + "Richard Stallman's beard makes ads for Gillette and Braun appear."; + "for i = 1 to Stallman will never stop."; + "\"RMS\" stands for \"RMS Makes Software\""; + "Whenever someone writes a \"Hello, world\" program, Richard Stallman says \"Hello\" back."; + "Richard Stallman wasn't born. He was compiled from source."; + "Richard Stallman has a URL tatooed on the left side of his chest where you can download his genetic code."; + "The GNU command line idiom that Richard Stallman never needs: \"date | more\""; + "Richard Stallman's toe cheese is aged to perfection."; + "Richard Stallman doesn't always run an OS kernel, but when he does he prefers GNU/Hurd. He is... the most interesting hacker in the world. Stay free, my friends."; + "When Richard Stallman gets hungry, he just picks debris from his foot and eats it."; + "Richard Stallman can GPL your code just by looking at it funny."; + "Richard Mathew Stallman loves birds. Birds make auricular love to Richard Mathew Stallman."; + "Richard Stallman is so free that the primitive recursive function for computing his liberty causes a stack overflow."; + "GNU Hurd is taking more than twenty years to develop because Richard Stallman is using a programming language comprised entirely of different lengths of time."; + "Richard Stallman's beard contains Richard Stallman, whose beard contains Richard Stallman...."; + "Richard Stallman had a Google Plus account in 2010."; + "sudo chown rms:gnu ~/base -R"; + "Richard Stallman pipes the Emacs binaries to /dev/dsp before he goes to sleep."; + "When Richard Stallman counted his fingers as a kid, he always started with 0."; + "When Richard Stallman's computer gets a virus, he simply applies a GPL license to it which converts the whole botnet to Linux. I mean, GNU/Linux."; + "Richard Stallman's beard trimmings can cure cancer. Too bad he never shaves."; + "Richard Stallman's doesn't kill a process, he just dares it to stay running."; + "Richard Stallman exists because he compiled himself into being. "; + "Richard Stallman's first words were in binary. When they couldn't understand him, he wrote a parser."; + "Richard Stallman doesn't need any codecs, he just opens a multimedia file with Emacs, and reads the bytes of the file as plain text. He then performs all the necessary decoding in his mind. But he refuses to decode files encrypted with DRM, although his mind is able to."; + }; + + w = wisdom{randi([1, numel(wisdom)])}; + if nargout > 0 + f = w; + else + w = wordwrap (w); + printf ("%s", w); + endif +endfunction + +function out = wordwrap (w) + cols = terminal_size ()(2); + wc = strsplit (w, " "); + out = "\n"; + i = 1; + numwords = numel (wc); + while i <= numwords; + line = wc{i}; + while (i < numwords && length (newline = cstrcat (line, " ", wc{i+1})) < cols) + line = newline; + i++; + endwhile + out = cstrcat (out, line, "\n"); + i++; + endwhile + out = cstrcat(out, "\n"); +endfunction \ No newline at end of file
--- a/scripts/miscellaneous/module.mk Tue Dec 27 15:08:21 2011 -0500 +++ b/scripts/miscellaneous/module.mk Tue Dec 27 18:47:18 2011 -0500 @@ -20,6 +20,7 @@ miscellaneous/dos.m \ miscellaneous/dump_prefs.m \ miscellaneous/edit.m \ + miscellaneous/fact.m \ miscellaneous/fileattrib.m \ miscellaneous/fileparts.m \ miscellaneous/fullfile.m \
--- a/scripts/miscellaneous/warning_ids.m Tue Dec 27 15:08:21 2011 -0500 +++ b/scripts/miscellaneous/warning_ids.m Tue Dec 27 18:47:18 2011 -0500 @@ -120,6 +120,11 @@ ## directory as the .oct file referred to by the autoload() command. ## By default, the @code{Octave:autoload-relative-file-name} warning is enabled. ## +## @item Octave:broadcast +## Warn when performing broadcasting operations. By default, this is +## enabled. See the Broadcasting section in the Vectorization and Faster +## Code Execution chapter in the manual. +## ## @item Octave:built-in-variable-assignment ## By default, the @code{Octave:built-in-variable-assignment} warning is ## enabled.
--- a/src/DLD-FUNCTIONS/bsxfun.cc Tue Dec 27 15:08:21 2011 -0500 +++ b/src/DLD-FUNCTIONS/bsxfun.cc Tue Dec 27 18:47:18 2011 -0500 @@ -312,8 +312,8 @@ DEFUN_DLD (bsxfun, args, , "-*- texinfo -*-\n\ @deftypefn {Loadable Function} {} bsxfun (@var{f}, @var{A}, @var{B})\n\ -The binary singleton expansion function applier does what its name\n\ -suggests: applies a binary function @var{f} element-by-element to two\n\ +The binary singleton expansion function applier performs broadcasting,\n\ +that is, applies a binary function @var{f} element-by-element to two\n\ array arguments @var{A} and @var{B}, and expands as necessary\n\ singleton dimensions in either input argument. @var{f} is a function\n\ handle, inline function, or string containing the name of the function\n\
--- a/src/DLD-FUNCTIONS/cellfun.cc Tue Dec 27 15:08:21 2011 -0500 +++ b/src/DLD-FUNCTIONS/cellfun.cc Tue Dec 27 18:47:18 2011 -0500 @@ -365,6 +365,21 @@ @end group\n\ @end example\n\ \n\ +Use @code{cellfun} intelligently. The @code{cellfun} function is a\n\ +useful tool for avoiding loops. It is often used with anonymous\n\ +function handles; however, calling an anonymous function involves an\n\ +overhead quite comparable to the overhead of an m-file function.\n\ +Passing a handle to a built-in function is faster, because the\n\ +interpreter is not involved in the internal loop. For example:\n\ +\n\ +@example\n\ +@group\n\ +a = @{@dots{}@}\n\ +v = cellfun (@@(x) det(x), a); # compute determinants\n\ +v = cellfun (@@det, a); # faster\n\ +@end group\n\ +@end example\n\ +\n\ @seealso{arrayfun, structfun, spfun}\n\ @end deftypefn") {
--- a/src/data.cc Tue Dec 27 15:08:21 2011 -0500 +++ b/src/data.cc Tue Dec 27 18:47:18 2011 -0500 @@ -6636,9 +6636,10 @@ if (n < 0) n = idx.extent (0); else if (idx.extent (n) > n) - error ("accumarray: index out of range"); - - dim_vector rdv = vals.dims (); + error ("accumdim: index out of range"); + + dim_vector vals_dim = vals.dims (), rdv = vals_dim; + if (dim < 0) dim = vals.dims ().first_non_singleton (); else if (dim >= rdv.length ()) @@ -6648,7 +6649,11 @@ NDT retval (rdv, T()); + if (idx.length () != vals_dim(dim)) + error ("accumdim: dimension mismatch"); + retval.idx_add_nd (idx, vals, dim); + return retval; }
--- a/src/version.h.in Tue Dec 27 15:08:21 2011 -0500 +++ b/src/version.h.in Tue Dec 27 18:47:18 2011 -0500 @@ -87,6 +87,8 @@ #define OCTAVE_STARTUP_MESSAGE \ X_OCTAVE_NAME_VERSION_COPYRIGHT_COPYING_WARRANTY_AND_BUGS \ (" For details, type `warranty'.") "\n\n" \ - "For information about changes from previous versions, type `news'." - + "For information about changes from previous versions, type `news'.\n\n\ +** WARNING ** Semantics for arithmetic operators have changed and now\n\ + broadcast. Type `doc broadcast' to get a description of\n\ + the changes." #endif