SumTools[DefiniteSum]
SummableSpace
construct the summable space
Calling Sequence
Parameters
Options
Description
Examples
References
Compatibility
SummableSpace[method](reqn, fcn, options)
SummableSpace[method](cert, n, v, options)
method
-
(optional) either Gosper or AccurateSummation; if omitted, Gosper is assumed
reqn
homogeneous linear recurrence
fcn
function name, e.g., v(n)
cert
rational function in n
n
name; the independent variable
v
name; the dependent variable
opts
sequence of optional equations of the form keyword=value. Possible keywords are output, range, or primitive.
Each optional argument is of the form keyword = value. The following options are supported.
'output'
Specifies the desired form of representations of sequences in the summable space. Possible values:
'RESol'
Indicates that the sequences are to be represented by an RESol data structure, of the form RESol⁡reqn,v⁡n,inits, where inits is a set of initial conditions.
'piecewise'
Indicates that the sequences are to be represented by an explicit expression depending on n, which in general is a piecewise expression.
This argument is ignored in the AccurateSummation case, and an RESol data structure is returned always. In the Gosper case, the default is piecewise.
'range'=a..b
Specify an interval R=a..b with integer or infinite bounds (−∞..∞ by default). If this option is given then it is assumed that v⁡n is determined only for n∈R and satisfies reqn for all integers n such that both n and n+1 are in R. Moreover, the discrete Newton-Leibniz formula should be valid for any integers n1,n2∈R.
'primitive'=truefalse
If this option is given, the command returns a pair V,T where V represents the summable space of all v⁡n and T represents the space of all primitives u⁡n. In the Gosper case, both are returned in the form specified by the option 'output'. In the AccurateSummation case, T is returned as an expression in terms of n and v and is typically a piecewise expression. The default is false.
The command SummableSpace(reqn, fcn) or SummableSpace[Gosper](reqn, fcn) constructs the space of all Gosper definite summable sequences v⁡n satisfying the given homogeneous first order linear recurrence reqn with polynomial coefficients, of the form a1⁡n⁢v⁡n+1+a0⁡n⁢v⁡n=0, for all integers n.
The command SummableSpace[AccurateSummation](reqn, fcn) constructs the space of accurate summation definite summable sequences satisfying a given homogeneous linear recurrence reqn of arbitrary order with polynomial coefficients.
The form in which the result is returned is determined by the output option; see below for details. The output may contain placeholders of the form v⁡0,v⁡1,... representing initial conditions or free parameters of the resulting space.
Instead of the recurrence, a certificate cert can be specified, in which case the recurrence is taken as denom⁡cert⁢v⁡n+1−numer⁡cert⁢v⁡n=0.
A sequence satisfying a first order linear recurrence is called hypergeometric. A hypergeometric sequence v⁡n is called Gosper indefinite summable if there is another hypergeometric sequence u⁡n such that v⁡n=u⁡n+1−u⁡n. The sequence u⁡n is called a primitive for v⁡n. A Gosper indefinite summable sequence is called Gosper definite summable if the discrete Newton-Leibniz formula
∑n=n1n2⁡v⁡n=u⁡n2+1−u⁡n1
is valid for any integers n1,n2.
A sequence v⁡n satisfying a homogeneous linear recurrence with polynomial coefficients of order d is called accurate summation indefinite summable if there is a sequence u⁡n such that v⁡n=u⁡n+1−u⁡n and u⁡n satisfies another homogeneous linear recurrence if the same order d. The sequence u⁡n is called a primitive for v⁡n. An accurate summation indefinite summable sequence is called accurate summation definite summable if the discrete Newton-Leibniz formula is valid for any integers n1,n2.
The primitive u⁡n is a linear combination of v⁡n,v⁡n+1,...,v⁡n+d−1 with rational function coefficients, where d is the order of reqn, with the possible exception of finitely many values n. In particular, in the Gosper case the primitive is a rational function multiple of v⁡n.
If no nonzero summable sequences for reqn exist, then the command returns FAIL.
with⁡SumToolsDefiniteSum:
rec≔k⁢v⁡k+1−k+12⁢v⁡k=0
SummableSpace⁡rec,v⁡k,output=RESol
RESol⁡−k2−2⁢k−1⁢v⁡k+k⁢v⁡k+1=0,v⁡k,v⁡−1=v⁡−1,v⁡0=0,v⁡1=0,INFO
V,T≔SummableSpaceGosper⁡rec,v⁡k,output=piecewise,primitive
V,T≔v⁡−1⁢−1−k⁢kΓ⁡−kk≤−100≤k,v⁡−1⁢−1−kΓ⁡−kk≤−100≤k
add⁡eval⁡V,k=i,i=−100..100=eval⁡T,k=101−eval⁡T,k=−100
−v⁡−1933262154439441526816992388562667004907159682643816214685929638952175999932299156089414639761565182862536979208272237582511852109168640000000000000000000000=−v⁡−1933262154439441526816992388562667004907159682643816214685929638952175999932299156089414639761565182862536979208272237582511852109168640000000000000000000000
SummableSpaceGosper⁡rec,v⁡k,range=0..∞
v⁡1⁢Γ⁡k+1⁢k
cert≔kk+2
SummableSpaceGosper⁡cert,k,v
0k≤−2v⁡−1k=−1−v⁡−1k=001≤k
SummableSpaceGosper⁡cert,k,v,range=1..∞
2⁢v⁡1k+1⁢k
SummableSpaceGosper⁡2⁢k2−4⁢k−9⁢v⁡k+1−2⁢k−3⁢k−1⁢k−8⁢v⁡k=0,v⁡k
0k≤−22⁢v⁡1k=−1−3⁢v⁡1k=0v⁡1k=1−8⁢v⁡3⁢Γ⁡k−32⁢k−2⁢k−9π⁢Γ⁡k+22≤k
L≔k−3⁢k−2⁢k+1⁢v⁡k+2−k−3⁢k2−2⁢k−1⁢v⁡k+1−k−22⁢v⁡k=0
SummableSpaceAccurateSummation⁡L,v⁡k,primitive
RESol⁡−k2+4⁢k−4⁢v⁡k+−k3+5⁢k2−5⁢k−3⁢v⁡k+1+k3−4⁢k2+k+6⁢v⁡k+2=0,v⁡k,v⁡2=v⁡2,v⁡3=0,v⁡4=v⁡4,v⁡5=−v⁡44,INFO,v⁡kk−3+k⁢v⁡k+1k≤20k=3v⁡kk−3+k⁢v⁡k+14≤k
SummableSpaceAccurateSummation⁡L,v⁡k,range=4..∞,primitive
RESol⁡−k2+4⁢k−4⁢v⁡k+−k3+5⁢k2−5⁢k−3⁢v⁡k+1+k3−4⁢k2+k+6⁢v⁡k+2=0,v⁡k,v⁡4=v⁡4,v⁡5=v⁡5,INFO,v⁡kk−3+k⁢v⁡k+1
S.A. Abramov. "On the summation of P-recursive sequences." Proc. of ISSAC'06, (2006): 17-22.
The SumTools[DefiniteSum][SummableSpace] command was introduced in Maple 15.
For more information on Maple 15 changes, see Updates in Maple 15.
See Also
OreTools[MathOperations][AccurateIntegration]
SumTools[Hypergeometric][BottomSequence]
SumTools[Hypergeometric][Gosper]
SumTools[IndefiniteSum][AccurateSummation]
Download Help Document