%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %% %A options.tex ANUPQ documentation - options Werner Nickel %A Greg Gamble %% %A @(#)$Id: options.tex,v 1.31 2005/07/20 19:38:36 werner Exp $ %% %% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% \Chapter{ANUPQ Options} In this chapter we describe in detail all the options used by functions of the {\ANUPQ} package. Note that by ``options'' we mean {\GAP} options that are passed to functions after the arguments and separated from the arguments by a colon as described in Chapter~"ref:Function Calls" in the Reference Manual. The user is strongly advised to read Section~"Hints and Warnings regarding the use of Options". \>AllANUPQoptions() F lists all the {\GAP} options defined for functions of the {\ANUPQ} package: \beginexample gap> AllANUPQoptions(); [ "AllDescendants", "BasicAlgorithm", "Bounds", "CapableDescendants", "ClassBound", "CustomiseOutput", "Exponent", "Filename", "GroupName", "Identities", "Metabelian", "NumberOfSolubleAutomorphisms", "OrderBound", "OutputLevel", "PcgsAutomorphisms", "PqWorkspace", "Prime", "PrintAutomorphisms", "PrintPermutations", "QueueFactor", "RankInitialSegmentSubgroups", "RedoPcp", "RelativeOrders", "Relators", "SetupFile", "SpaceEfficient", "StandardPresentationFile", "StepSize", "SubList", "TreeDepth", "pQuotient" ] \endexample The following global variable gives a partial breakdown of where the above options are used. \>`ANUPQoptions' V is a record of lists of names of admissible {\ANUPQ} options, such that each field is either the name of a ``key'' {\ANUPQ} function or `other' (for a miscellaneous list of functions) and the corresponding value is the list of option names that are admissible for the function (or miscellaneous list of functions). Also, from within a {\GAP} session, you may use {\GAP}'s help browser (see Chapter~"ref:The Help System" in the {\GAP} Reference Manual); to find out about any particular {\ANUPQ} option, simply type: ```?option <option>''', where <option> is one of the options listed above without any quotes, e.g. \begintt gap> ?option Prime \endtt will display the sections in this manual that describe the `Prime' option. In fact the first 4 are for the functions that have `Prime' as an option and the last actually describes the option. So follow up by choosing \begintt gap> ?5 \endtt This is also the pattern for other options (the last section of the list always describes the option; the other sections are the functions with which the option may be used). In the section following we describe in detail all {\ANUPQ} options. To continue onto the next section on-line using {\GAP}'s help browser, type: \begintt gap> ?> \endtt %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% \Section{Detailed descriptions of ANUPQ Options} \beginitems \>`Prime := <p>'{option Prime}@{option `Prime'|indexit}& Specifies that the $p$-quotient for the prime <p> should be computed. \>`ClassBound := <n>'{option ClassBound}@{option `ClassBound'|indexit}& Specifies that the $p$-quotient to be computed has lower exponent-$p$ class at most <n>. If this option is omitted a default of 63 (which is the maximum possible for the `pq' program) is taken, except for `PqDescendants' (see~"PqDescendants") and in a special case of `PqPCover' (see~"PqPCover"). Let <F> be the argument (or start group of the process in the interactive case) for the function; then for `PqDescendants' the default is `PClassPGroup(<F>) + 1', and for the special case of `PqPCover' the default is `PClassPGroup(<F>)'. \>`pQuotient := <Q>'{option pQuotient}@{option `pQuotient'|indexit}& This option is only available for the standard presentation functions. It specifies that a $p$-quotient of the group argument of the function or group of the process is the pc <p>-group <Q>, where <Q> is of class *less than* the provided (or default) value of `ClassBound'. If `pQuotient' is provided, then the option `Prime' if also provided, is ignored; the prime <p> is discovered by computing `PrimePGroup(<Q>)'. \>`Exponent := <n>'{option Exponent}@{option `Exponent'|indexit}& Specifies that the $p$-quotient to be computed has exponent <n>. For an interactive process, `Exponent' defaults to a previously supplied value for the process. Otherwise (and non-interactively), the default is 0, which means that no exponent law is enforced. \>`Relators := <rels>'{option Relators}@{option `Relators'|indexit}& Specifies that the relators sent to the `pq' program should be <rels> instead of the relators of the argument group <F> (or start group in the interactive case) of the calling function; <rels> should be a list of *strings* in the string representations of the generators of <F>, and <F> must be an *fp group* (even if the calling function accepts a pc group). This option provides a way of giving relators to the `pq' program, without having them pre-expanded by {\GAP}, which can sometimes effect a performance loss of the order of 100 (see Section~"The Relators Option"). *Notes* \beginlist%ordered \item{1.} The `pq' program does not use `/' to indicate multiplication by an inverse and uses square brackets to represent (left normed) commutators. Also, even though the `pq' program accepts relations, all elements of <rels> *must* be in relator form, i.e.~a relation of form `<w1> = <w2>' must be written as `<w1>*(<w2>)^-1' and then put in a pair of double-quotes to make it a string. See the example below. \item{2.} To ensure there are no syntax errors in <rels>, each relator is parsed for validity via `PqParseWord' (see~"PqParseWord"). If they are ok, a message to say so is `Info'-ed at `InfoANUPQ' level 2. \endlist \>`Metabelian'{option Metabelian}@{option `Metabelian'|indexit}& Specifies that the largest metabelian $p$-quotient subject to any other conditions specified by other options be constructed. By default this restriction is not enforced. \>`GroupName := <name>'{option GroupName}@{option `GroupName'|indexit}& Specifies that the `pq' program should refer to the group by the name <name> (a string). If `GroupName' is not set and the group has been assigned a name via `SetName' (see~"ref:SetName") it is set as the name the `pq' program should use. Otherwise, the ``generic'' name `"[grp]"' is set as a default. \>`Identities := <funcs>'{option Identities}@{option `Identities'|indexit}& Specifies that the pc presentation should satisfy the laws defined by each function in the list <funcs>. This option may be called by `Pq', `PqEpimorphism', or `PqPCover' (see~"Pq"). Each function in the list <funcs> must return a word in its arguments (there may be any number of arguments). Let <identity> be one such function in <funcs>. Then as each lower exponent <p>-class quotient is formed, instances $<identity>(<w1>, \dots, <wn>)$ are added as relators to the pc presentation, where $<w1>, \dots, <wn>$ are words in the pc generators of the quotient. At each class the class and number of pc generators is `Info'-ed at `InfoANUPQ' level 1, the number of instances is `Info'-ed at `InfoANUPQ' level 2, and the instances that are evaluated are `Info'-ed at `InfoANUPQ' level 3. As usual timing information is `Info'-ed at `InfoANUPQ' level 2; and details of the processing of each instance from the `pq' program (which is often quite *voluminous*) is `Info'-ed at `InfoANUPQ' level 3. Try the examples `"B2-4-Id"' and `"11gp-3-Engel-Id"' which demonstrate the usage of the `Identities' option; these are run using `PqExample' (see~"PqExample"). Take note of Note 1.~below in relation to the example `"B2-4-Id"'; the companion example `"B2-4"' generates the same group using the `Exponent' option. These examples are discussed at length in Section~"The Identities Option and PqEvaluateIdentities Function". *Notes* \beginlist%ordered \item{1.} Setting the `InfoANUPQ' level to 3 or more when setting the `Identities' option may slow down the computation considerably, by overloading {\GAP} with io operations. \item{2.} The `Identities' option is implemented at the {\GAP} level. An identity that is just an exponent law should be specified using the `Exponent' option (see~"option Exponent"), which is implemented at the C level and is highly optimised and so is much more efficient. \item{3.} The number of instances of each identity tends to grow combinatorially with the class. So *care* should be exercised in using the `Identities' option, by including other restrictions, e.g.~by using the `ClassBound' option (see~"option ClassBound"). \endlist \>`OutputLevel := <n>'{option OutputLevel}@{option `OutputLevel'|indexit}& Specifies the level of ``verbosity'' of the information output by the ANU `pq' program when computing a pc presentation; <n> must be an integer in the range 0 to 3. `OutputLevel := 0' displays at most one line of output and is the default; `OutputLevel := 1' displays (usually) slightly more output and `OutputLevel's of 2 and 3 are two levels of verbose output. To see these messages from the `pq' program, the `InfoANUPQ' level must be set to at least 1 (see~"InfoANUPQ"). See Section~"Hints and Warnings regarding the use of Options" for an example of how `OutputLevel' can be used as a troubleshooting tool. \>`RedoPcp'{option RedoPcp}@{option `RedoPcp'|indexit}& Specifies that the current pc presentation (for an interactive process) stored by the `pq' program be scrapped and clears the current values stored for the options `Prime', `ClassBound', `Exponent' and `Metabelian' and also clears the `pQuotient', `pQepi' and `pCover' fields of the data record of the process. \>`SetupFile := <filename>'{option SetupFile}@{option `SetupFile'|indexit}& Non-interactively, this option directs that `pq' should not be called and that an input file with name <filename> (a string), containing the commands necessary for the ANU `pq' standalone, be constructed. The commands written to <filename> are also `Info'-ed behind a ```ToPQ> ''' prompt at `InfoANUPQ' level 4 (see~"InfoANUPQ"). Except in the case following, the calling function returns `true'. If the calling function is the non-interactive version of one of `Pq', `PqPCover' or `PqEpimorphism' and the group provided as argument is trivial given with an empty set of generators, then no setup file is written and `fail' is returned (the `pq' program cannot do anything useful with such a group). Interactively, `SetupFile' is ignored. *Note:* Since commands emitted to the `pq' program may depend on knowing what the ``current state'' is, to form a setup file some ``close enough guesses'' may sometimes be necessary; when this occurs a warning is `Info'-ed at `InfoANUPQ' or `InfoWarning' level 1. To determine whether the ``close enough guesses'' give an accurate setup file, it is necessary to run the command without the `SetupFile' option, after either setting the `InfoANUPQ' level to at least 4 (the setup file script can then be compared with the ```ToPQ> ''' commands that are `Info'-ed) or setting a `pq' command log file by using `ToPQLog' (see~"ToPQLog"). \>`PqWorkspace := <workspace>'{option PqWorkspace}@{option `PqWorkspace'|indexit}& Non-interactively, this option sets the memory used by the `pq' program. It sets the maximum number of integer-sized elements to allocate in its main storage array. By default, the `pq' program sets this figure to 10000000. Interactively, `PqWorkspace' is ignored; the memory used in this case may be set by giving `PqStart' a second argument (see~"PqStart"). \>`PcgsAutomorphisms'{option PcgsAutomorphisms}% @{option `PcgsAutomorphisms'|indexit} \>`PcgsAutomorphisms := false'{option PcgsAutomorphisms}% @{option `PcgsAutomorphisms'|indexit}& Let <G> be the group associated with the calling function (or associated interactive process). Passing the option `PcgsAutomorphisms' without a value (or equivalently setting it to `true'), specifies that a polycyclic generating sequence for the automorphism group (which must be *soluble*) of <G>, be computed and passed to the `pq' program. This increases the efficiency of the computation; it also prevents the `pq' from calling {\GAP} for orbit-stabilizer calculations. By default, `PcgsAutomorphisms' is set to the value returned by `IsSolvable( AutomorphismGroup( <G> ) )', and uses the package {\AutPGrp} to compute `AutomorphismGroup( <G> )' if it is installed. This flag is set to `true' or `false' in the background according to the above criterion by the function `PqDescendants' (see~"PqDescendants" and~"PqDescendants!interactive"). *Note:* If `PcgsAutomorphisms' is used when the automorphism group of <G> is insoluble, an error message occurs. \>`OrderBound := <n>'{option OrderBound}@{option `OrderBound'|indexit}& Specifies that only descendants of size at most $p^<n>$, where <n> is a non-negative integer, be generated. Note that you cannot set both `OrderBound' and `StepSize'. \>`StepSize := <n>'{option StepSize}@{option `StepSize'} \>`StepSize := <list>'{option StepSize}@{option `StepSize'|indexit}& For a positive integer <n>, `StepSize' specifies that only those immediate descendants which are a factor $p^<n>$ bigger than their parent group be generated. & For a list <list> of positive integers such that the sum of the length of <list> and the exponent-$p$ class of <G> is equal to the class bound defined by the option `ClassBound', `StepSize' specifies that the integers of <list> are the step sizes for each additional class. \>`RankInitialSegmentSubgroups := <n>'{option RankInitialSegmentSubgroups}% @{option `RankInitialSegmentSubgroups'|indexit}& Sets the rank of the initial segment subgroup chosen to be <n>. By default, this has value 0. \>`SpaceEfficient'{option SpaceEfficient}@{option `SpaceEfficient'|indexit}& Specifies that the `pq' program performs certain calculations of $p$-group generation more slowly but with greater space efficiency. This flag is frequently necessary for groups of large Frattini quotient rank. The space saving occurs because only one permutation is stored at any one time. This option is only available if the `PcgsAutomorphisms' flag is set to `true' (see~"option PcgsAutomorphisms"). For an interactive process, `SpaceEfficient' defaults to a previously supplied value for the process. Otherwise (and non-interactively), `SpaceEfficient' is by default `false'. \>`CapableDescendants'{option CapableDescendants}@{option `CapableDescendants'|indexit}& By default, *all* (i.e.~capable and terminal) descendants are computed. If this flag is set, only capable descendants are computed. Setting this option is equivalent to setting `AllDescendants := false' (see~"option AllDescendants"), except if both `CapableDescendants' and `AllDescendants' are passed, `AllDescendants' is essentially ignored. \goodbreak% \>`AllDescendants := false'{option AllDescendants}@{option `AllDescendants'|indexit}& By default, *all* descendants are constructed. If this flag is set to `false', only capable descendants are computed. Passing `AllDescendants' without a value (which is equivalent to setting it to `true') is superfluous. This option is provided only for backward compatibility with the {\GAP} 3 version of the {\ANUPQ} package, where by default `AllDescendants' was set to `false' (rather than `true'). It is preferable to use `CapableDescendants' (see~"option CapableDescendants"). \>`TreeDepth := <class>'{option TreeDepth}@{option `TreeDepth'|indexit}& Specifies that the descendants tree developed by `PqDescendantsTreeCoclassOne' (see~"PqDescendantsTreeCoclassOne") should be extended to class <class>, where <class> is a positive integer. \>`SubList := <sub>'{option SubList}@{option `SubList'|indexit}& Suppose that <L> is the list of descendants generated, then for a list <sub> of integers this option causes `PqDescendants' to return `Sublist( <L>, <sub> )'. If an integer <n> is supplied, `PqDescendants' returns `<L>[<n>]'. \>`NumberOfSolubleAutomorphisms := <n>'{option NumberOfSolubleAutomorphisms}% @{option `NumberOfSolubleAutomorphisms'|indexit}& Specifies that the number of soluble automorphisms of the automorphism group supplied by `PqPGSupplyAutomorphisms' (see~"PqPGSupplyAutomorphisms") in a $p$-group generation calculation is <n>. By default, <n> is taken to be $0$; <n> must be a non-negative integer. If $<n> \ge 0$ then a value for the option `RelativeOrders' (see~"option RelativeOrders") must also be supplied. \>`RelativeOrders := <list>'{option RelativeOrders}% @{option `RelativeOrders'|indexit}& Specifies the relative orders of each soluble automorphism of the automorphism group supplied by `PqPGSupplyAutomorphisms' (see~"PqPGSupplyAutomorphisms") in a $p$-group generation calculation. The list <list> must consist of <n> positive integers, where <n> is the value of the option `NumberOfSolubleAutomorphisms' (see~"option NumberOfSolubleAutomorphisms"). By default <list> is empty. \>`BasicAlgorithm'{option BasicAlgorithm}@{option `BasicAlgorithm'|indexit}& Specifies that an algorithm that the `pq' program calls its ``default'' algorithm be used for $p$-group generation. By default this algorithm is *not* used. If this option is supplied the settings of options `RankInitialSegmentSubgroups', `AllDescendants', `Exponent' and `Metabelian' are ignored. \>`CustomiseOutput := <rec>'{option CustomiseOutput}% @{option `CustomiseOutput'|indexit}& Specifies that fine tuning of the output is desired. The record <rec> should have any subset (or all) of the the following fields: \quad`perm := <list>' & where <list> is a list of booleans which determine whether the permutation group output for the automorphism group should contain: the degree, the extended automorphisms, the automorphism matrices, and the permutations, respectively. \quad`orbit := <list>' & where <list> is a list of booleans which determine whether the orbit output of the action of the automorphism group should contain: a summary, and a complete listing of orbits, respectively. (It's possible to have *both* a summary and a complete listing.) \quad`group := <list>' & where <list> is a list of booleans which determine whether the group output should contain: the standard matrix of each allowable subgroup, the presentation of reduced $p$-covering groups, the presentation of immediate descendants, the nuclear rank of descendants, and the $p$-multiplicator rank of descendants, respectively. \quad`autgroup := <list>' & where <list> is a list of booleans which determine whether the automorphism group output should contain: the commutator matrix, the automorphism group description of descendants, and the automorphism group order of descendants, respectively. \quad`trace := <val>' & where <val> is a boolean which if `true' specifies algorithm trace data is desired. By default, one does not get algorithm trace data. Not providing a field (or mis-spelling it!), specifies that the default output is desired. As a convenience, `1' is also accepted as `true', and any value that is neither `1' nor `true' is taken as `false'. Also for each <list> above, an unbound list entry is taken as `false'. Thus, for example \begintt CustomiseOutput := rec(group := [,,1], autgroup := [,1]) \endtt specifies for the group output that only the presentation of immediate descendants is desired, for the automorphism group output only the automorphism group description of descendants should be printed, that there should be no algorithm trace data, and that the default output should be provided for the permutation group and orbit output. \>`StandardPresentationFile := <filename>'{option StandardPresentationFile}% @{option `StandardPresentationFile'|indexit}& Specifies that the file to which the standard presentation is written has name <filename>. If the first character of the string <filename> is not `/', <filename> is assumed to be the path of a writable file relative to the directory in which {\GAP} was started. If this option is omitted it is written to the file with the name generated by the command `Filename( ANUPQData.tmpdir, "SPres" );', i.e.~the file with name `"SPres"' in the temporary directory in which the `pq' program executes. \>`QueueFactor := <n>'{option QueueFactor}@{option `QueueFactor'|indexit}& Specifies a queue factor of <n>, where <n> should be a positive integer. This option may be used with `PqNextClass' (see~"PqNextClass"). The queue factor is used when the `pq' program uses automorphisms to close a set of elements of the $p$-multiplicator under their action. The algorithm used is a spinning algorithm: it starts with a set of vectors in echelonized form (elements of the $p$-multiplicator) and closes the span of these vectors under the action of the automorphisms. For this each automorphism is applied to each vector and it is checked if the result is contained in the span computed so far. If not, the span becomes bigger and the vector is put into a queue and the automorphisms are applied to this vector at a later stage. The process terminates when the automorphisms have been applied to all vectors and no new vectors have been produced. For each new vector it is decided, if its processing should be delayed. If the vector contains too many non-zero entries, it is put into a second queue. The elements in this queue are processed only when there are no elements in the first queue left. The queue factor is a percentage figure. A vector is put into the second queue if the percentage of its non-zero entries exceeds the queue factor. \>`Bounds := <list>'{option Bounds}@{option `Bounds'|indexit}& Specifies a lower and upper bound on the indices of a list, where <list> is a pair of positive non-decreasing integers. See~"PqDisplayStructure" and~"PqDisplayAutomorphisms" where this option may be used. \>`PrintAutomorphisms := <list>'{option PrintAutomorphisms}% @{option `PrintAutomorphisms'|indexit}& Specifies that automorphism matrices be printed. \>`PrintPermutations := <list>'{option PrintPermutations}% @{option `PrintPermutations'|indexit}& Specifies that permutations of the subgroups be printed. \>`Filename := <string>'{option Filename}@{option `Filename'|indexit}& Specifies that an output or input file to be written to or read from by the `pq' program should have the name <string>. \enditems %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %% %E