Given two finite sets of integers
an id="mmlsi1" class="mathmlsrc">an class="formulatext stixSupport mathImg" data-mathURL="/science?_ob=MathURL&_method=retrieve&_eid=1-s2.0-S0304397516001808&_mathId=si1.gif&_user=111111111&_pii=S0304397516001808&_rdoc=1&_issn=03043975&md5=30b97daafa21c1e91e78a22992c3ccde" title="Click to view the MathML source">S⊆N∖{0}an>an class="mathContainer hidden">an class="mathCode">ath altimg="si1.gif" overflow="scroll">S⊆athvariant="double-struck">N∖alse">{0alse">}ath>an>an>an>
and
an id="mmlsi2" class="mathmlsrc">an class="formulatext stixSupport mathImg" data-mathURL="/science?_ob=MathURL&_method=retrieve&_eid=1-s2.0-S0304397516001808&_mathId=si2.gif&_user=111111111&_pii=S0304397516001808&_rdoc=1&_issn=03043975&md5=3136f4808d7929e2868fbd21d5abe214" title="Click to view the MathML source">D⊆N∖{0,1}an>an class="mathContainer hidden">an class="mathCode">ath altimg="si2.gif" overflow="scroll">D⊆athvariant="double-struck">N∖alse">{0,1alse">}ath>an>an>an>, the imp
arti
al combin
atori
al g
ame
an id="mmlsi3" class="mathmlsrc">an class="formulatext stixSupport mathImg" data-mathURL="/science?_ob=MathURL&_method=retrieve&_eid=1-s2.0-S0304397516001808&_mathId=si3.gif&_user=111111111&_pii=S0304397516001808&_rdoc=1&_issn=03043975&md5=d394395cfc36887c6d73686f026f94c7" title="Click to view the MathML source">i-Mark(S,D)an>an class="mathContainer hidden">an class="mathCode">ath altimg="si3.gif" overflow="scroll">i-all-caps>Markall-caps>alse">(S,Dalse">)ath>an>an>an> is pl
ayed on
a he
ap of tokens. From
a he
ap of
n tokens, e
ach pl
ayer c
an move either to
a he
ap of
an id="mmlsi4" class="mathmlsrc">an class="formulatext stixSupport mathImg" data-mathURL="/science?_ob=MathURL&_method=retrieve&_eid=1-s2.0-S0304397516001808&_mathId=si4.gif&_user=111111111&_pii=S0304397516001808&_rdoc=1&_issn=03043975&md5=cac4228a580e32430b01c9ea3c5ad4f8" title="Click to view the MathML source">n−san>an class="mathContainer hidden">an class="mathCode">ath altimg="si4.gif" overflow="scroll">n−sath>an>an>an> tokens for some
an id="mmlsi5" class="mathmlsrc">an class="formulatext stixSupport mathImg" data-mathURL="/science?_ob=MathURL&_method=retrieve&_eid=1-s2.0-S0304397516001808&_mathId=si5.gif&_user=111111111&_pii=S0304397516001808&_rdoc=1&_issn=03043975&md5=a879e659c11a0d014417eed922109eaf" title="Click to view the MathML source">s∈San>an class="mathContainer hidden">an class="mathCode">ath altimg="si5.gif" overflow="scroll">s∈Sath>an>an>an>,
an id="mmlsi6" class="mathmlsrc">an class="formulatext stixSupport mathImg" data-mathURL="/science?_ob=MathURL&_method=retrieve&_eid=1-s2.0-S0304397516001808&_mathId=si6.gif&_user=111111111&_pii=S0304397516001808&_rdoc=1&_issn=03043975&md5=eddf5265ead9ad8859e4fc02acb0ed8d" title="Click to view the MathML source">s≤nan>an class="mathContainer hidden">an class="mathCode">ath altimg="si6.gif" overflow="scroll">s≤nath>an>an>an>, or to
a he
ap of
an id="mmlsi7" class="mathmlsrc">an class="formulatext stixSupport mathImg" data-mathURL="/science?_ob=MathURL&_method=retrieve&_eid=1-s2.0-S0304397516001808&_mathId=si7.gif&_user=111111111&_pii=S0304397516001808&_rdoc=1&_issn=03043975&md5=fabaf9a482e38b4519853ef4266b2c01" title="Click to view the MathML source">n/dan>an class="mathContainer hidden">an class="mathCode">ath altimg="si7.gif" overflow="scroll">nalse">/dath>an>an>an> tokens for some
an id="mmlsi396" class="mathmlsrc">an class="formulatext stixSupport mathImg" data-mathURL="/science?_ob=MathURL&_method=retrieve&_eid=1-s2.0-S0304397516001808&_mathId=si396.gif&_user=111111111&_pii=S0304397516001808&_rdoc=1&_issn=03043975&md5=67016a0b0e461e71c530f041ce56520f" title="Click to view the MathML source">d∈Dan>an class="mathContainer hidden">an class="mathCode">ath altimg="si396.gif" overflow="scroll">d∈Dath>an>an>an> if
d divides
n. Such g
ames c
an be considered
as
an integr
al v
ari
ant of
an class="smallcaps">Markan>-type g
ames, introduced by Elwyn Berlek
amp
and Joe Buhler
and studied by Aviezri Fr
aenkel
and Al
an Guo, for which it is
allowed to move from
a he
ap of
n tokens to
a he
ap of
an id="mmlsi9" class="mathmlsrc">an class="formulatext stixSupport mathImg" data-mathURL="/science?_ob=MathURL&_method=retrieve&_eid=1-s2.0-S0304397516001808&_mathId=si9.gif&_user=111111111&_pii=S0304397516001808&_rdoc=1&_issn=03043975&md5=0c5cb95948456a5a7372542aacb62561" title="Click to view the MathML source">⌊n/d⌋an>an class="mathContainer hidden">an class="mathCode">ath altimg="si9.gif" overflow="scroll">alse">⌊nalse">/dalse">⌋ath>an>an>an> tokens for
any
an id="mmlsi396" class="mathmlsrc">an class="formulatext stixSupport mathImg" data-mathURL="/science?_ob=MathURL&_method=retrieve&_eid=1-s2.0-S0304397516001808&_mathId=si396.gif&_user=111111111&_pii=S0304397516001808&_rdoc=1&_issn=03043975&md5=67016a0b0e461e71c530f041ce56520f" title="Click to view the MathML source">d∈Dan>an class="mathContainer hidden">an class="mathCode">ath altimg="si396.gif" overflow="scroll">d∈Dath>an>an>an>.
Under normal convention, it is observed that the Sprague–Grundy sequence of the game an id="mmlsi3" class="mathmlsrc">an class="formulatext stixSupport mathImg" data-mathURL="/science?_ob=MathURL&_method=retrieve&_eid=1-s2.0-S0304397516001808&_mathId=si3.gif&_user=111111111&_pii=S0304397516001808&_rdoc=1&_issn=03043975&md5=d394395cfc36887c6d73686f026f94c7" title="Click to view the MathML source">i-Mark(S,D)an>an class="mathContainer hidden">an class="mathCode">ath altimg="si3.gif" overflow="scroll">i-all-caps>Markall-caps>alse">(S,Dalse">)ath>an>an>an> is aperiodic for any sets S and D. However, we prove that in many cases this sequence is almost periodic and that the sequence of winning positions is periodic. Moreover, in all these cases, the Sprague–Grundy value of a heap of n tokens can be computed in time an id="mmlsi10" class="mathmlsrc">an class="formulatext stixSupport mathImg" data-mathURL="/science?_ob=MathURL&_method=retrieve&_eid=1-s2.0-S0304397516001808&_mathId=si10.gif&_user=111111111&_pii=S0304397516001808&_rdoc=1&_issn=03043975&md5=a167e16b60351b0a8c7d27e7626022a8" title="Click to view the MathML source">O(logn)an>an class="mathContainer hidden">an class="mathCode">ath altimg="si10.gif" overflow="scroll">Oalse">(athvariant="normal">lognalse">)ath>an>an>an>.
We also prove that under misère convention the outcome sequence of these games is purely periodic.