-
Notifications
You must be signed in to change notification settings - Fork 152
/
spaces-cohomology.tex
4535 lines (4113 loc) · 169 KB
/
spaces-cohomology.tex
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
\input{preamble}
% OK, start here.
%
\begin{document}
\title{Cohomology of Algebraic Spaces}
\maketitle
\phantomsection
\label{section-phantom}
\tableofcontents
\section{Introduction}
\label{section-introduction}
\noindent
In this chapter we write about cohomology of algebraic spaces.
Although we prove some results on cohomology of abelian sheaves,
we focus mainly on cohomology of quasi-coherent sheaves, i.e.,
we prove analogues of the results in the chapter ``Cohomology of Schemes''.
Some of the results in this chapter can be found in \cite{Kn}.
\medskip\noindent
An important missing ingredient in this chapter is the
{\it induction principle}, i.e., the analogue for quasi-compact
and quasi-separated algebraic spaces of
Cohomology of Schemes, Lemma \ref{coherent-lemma-induction-principle}.
This is formulated precisely and proved in detail in
Derived Categories of Spaces, Section \ref{spaces-perfect-section-induction}.
Instead of the induction principle, in this chapter we use the
alternating {\v C}ech complex, see
Section \ref{section-alternating-cech}.
It is designed to prove vanishing statements such as
Proposition \ref{proposition-vanishing},
but in some cases the induction principle is a more powerful
and perhaps more ``standard'' tool. We encourage the reader
to take a look at the induction principle
after reading some of the material in this section.
\section{Conventions}
\label{section-conventions}
\noindent
The standing assumption is that all schemes are contained in
a big fppf site $\Sch_{fppf}$. And all rings $A$ considered
have the property that $\Spec(A)$ is (isomorphic) to an
object of this big site.
\medskip\noindent
Let $S$ be a scheme and let $X$ be an algebraic space over $S$.
In this chapter and the following we will write $X \times_S X$
for the product of $X$ with itself (in the category of algebraic
spaces over $S$), instead of $X \times X$.
\section{Higher direct images}
\label{section-higher-direct-image}
\noindent
Let $S$ be a scheme. Let $X$ be a representable algebraic space over $S$.
Let $\mathcal{F}$ be a quasi-coherent module on $X$ (see
Properties of Spaces, Section \ref{spaces-properties-section-quasi-coherent}).
By
Descent, Proposition \ref{descent-proposition-same-cohomology-quasi-coherent}
the cohomology groups $H^i(X, \mathcal{F})$ agree with the usual
cohomology group computed in the Zariski topology of the corresponding
quasi-coherent module on the scheme representing $X$.
\medskip\noindent
More generally, let $f : X \to Y$ be a quasi-compact and quasi-separated
morphism of representable algebraic spaces $X$ and $Y$. Let
$\mathcal{F}$ be a quasi-coherent module on $X$. By
Descent, Lemma \ref{descent-lemma-higher-direct-images-small-etale}
the sheaf $R^if_*\mathcal{F}$ agrees with the
usual higher direct image computed for the Zariski topology
of the quasi-coherent module on the scheme representing $X$
mapping to the scheme representing $Y$.
\medskip\noindent
More generally still, suppose $f : X \to Y$ is a
representable, quasi-compact, and
quasi-separated morphism of algebraic spaces over $S$. Let $V$ be a scheme
and let $V \to Y$ be an \'etale surjective morphism. Let $U = V \times_Y X$
and let $f' : U \to V$ be the base change of $f$. Then for any
quasi-coherent $\mathcal{O}_X$-module $\mathcal{F}$ we have
\begin{equation}
\label{equation-representable-higher-direct-image}
R^if'_*(\mathcal{F}|_U) = (R^if_*\mathcal{F})|_V,
\end{equation}
see
Properties of Spaces,
Lemma \ref{spaces-properties-lemma-pushforward-etale-base-change-modules}.
And because $f' : U \to V$ is a quasi-compact and quasi-separated
morphism of schemes, by the remark of the preceding paragraph we may
compute $R^if'_*(\mathcal{F}|_U)$ by thinking of $\mathcal{F}|_U$ as a
quasi-coherent sheaf on the scheme $U$, and $f'$ as a morphism of schemes.
We will frequently use this without further mention.
\medskip\noindent
Next, we prove that higher direct images of quasi-coherent sheaves are
quasi-coherent for any quasi-compact and quasi-separated morphism of
algebraic spaces. In the proof we use a trick; a ``better'' proof would
use a relative {\v C}ech complex, as discussed in
Sheaves on Stacks, Sections \ref{stacks-sheaves-section-cech} and
\ref{stacks-sheaves-section-sheaf-cech-complex} ff.
\begin{lemma}
\label{lemma-higher-direct-image}
Let $S$ be a scheme. Let $f : X \to Y$ be a morphism of algebraic spaces
over $S$. If $f$ is quasi-compact and quasi-separated, then $R^if_*$
transforms quasi-coherent $\mathcal{O}_X$-modules into
quasi-coherent $\mathcal{O}_Y$-modules.
\end{lemma}
\begin{proof}
Let $V \to Y$ be an \'etale morphism where $V$ is an affine scheme. Set
$U = V \times_Y X$ and denote $f' : U \to V$ the induced morphism.
Let $\mathcal{F}$ be a quasi-coherent $\mathcal{O}_X$-module. By
Properties of Spaces, Lemma
\ref{spaces-properties-lemma-pushforward-etale-base-change-modules}
we have
$R^if'_*(\mathcal{F}|_U) = (R^if_*\mathcal{F})|_V$.
Since the property of being a quasi-coherent module is local in the
\'etale topology on $Y$ (see
Properties of Spaces, Lemma
\ref{spaces-properties-lemma-characterize-quasi-coherent})
we may replace $Y$ by $V$, i.e., we may assume $Y$ is an affine scheme.
\medskip\noindent
Assume $Y$ is affine. Since $f$ is quasi-compact we see that $X$
is quasi-compact. Thus we may choose an affine scheme $U$ and a surjective
\'etale morphism $g : U \to X$, see
Properties of Spaces,
Lemma \ref{spaces-properties-lemma-quasi-compact-affine-cover}.
Picture
$$
\xymatrix{
U \ar[r]_g \ar[rd]_{f \circ g} & X \ar[d]^f \\
& Y
}
$$
The morphism $g : U \to X$ is representable, separated
and quasi-compact because $X$ is quasi-separated. Hence the lemma
holds for $g$ (by the discussion above the lemma).
It also holds for $f \circ g : U \to Y$ (as this is a morphism
of affine schemes).
\medskip\noindent
In the situation described in the previous paragraph we will show by
induction on $n$ that $IH_n$: for any quasi-coherent sheaf $\mathcal{F}$
on $X$ the sheaves $R^if\mathcal{F}$
are quasi-coherent for $i \leq n$.
The case $n = 0$ follows from
Morphisms of Spaces, Lemma \ref{spaces-morphisms-lemma-pushforward}.
Assume $IH_n$. In the rest of the proof we show that $IH_{n + 1}$ holds.
\medskip\noindent
Let $\mathcal{H}$ be a quasi-coherent $\mathcal{O}_U$-module.
Consider the Leray spectral sequence
$$
E_2^{p, q} = R^pf_* R^qg_* \mathcal{H}
\Rightarrow
R^{p + q}(f \circ g)_*\mathcal{H}
$$
Cohomology on Sites, Lemma \ref{sites-cohomology-lemma-relative-Leray}.
As $R^qg_*\mathcal{H}$ is quasi-coherent by $IH_n$ all the sheaves
$R^pf_*R^qg_*\mathcal{H}$ are quasi-coherent for $p \leq n$.
The sheaves $R^{p + q}(f \circ g)_*\mathcal{H}$ are all
quasi-coherent (in fact zero for $p + q > 0$ but we do not need this).
Looking in degrees $\leq n + 1$ the only module which we do not
yet know is quasi-coherent is $E_2^{n + 1, 0} = R^{n + 1}f_*g_*\mathcal{H}$.
Moreover, the differentials
$d_r^{n + 1, 0} : E_r^{n + 1, 0} \to E_r^{n + 1 + r, 1 - r}$
are zero as the target is zero. Using that $\QCoh(\mathcal{O}_X)$
is a weak Serre subcategory of $\textit{Mod}(\mathcal{O}_X)$
(Properties of Spaces, Lemma
\ref{spaces-properties-lemma-properties-quasi-coherent}) it
follows that $R^{n + 1}f_*g_*\mathcal{H}$
is quasi-coherent (details omitted).
\medskip\noindent
Let $\mathcal{F}$ be a quasi-coherent $\mathcal{O}_X$-module.
Set $\mathcal{H} = g^*\mathcal{F}$. The adjunction mapping
$\mathcal{F} \to g_*g^*\mathcal{F} = g_*\mathcal{H}$ is injective
as $U \to X$ is surjective \'etale. Consider the exact sequence
$$
0 \to \mathcal{F} \to g_*\mathcal{H} \to \mathcal{G} \to 0
$$
where $\mathcal{G}$ is the cokernel of the first map and in particular
quasi-coherent. Applying the long exact cohomology sequence we obtain
$$
R^nf_*g_*\mathcal{H} \to
R^nf_*\mathcal{G} \to
R^{n + 1}f_*\mathcal{F} \to
R^{n + 1}f_*g_*\mathcal{H} \to
R^{n + 1}f_*\mathcal{G}
$$
The cokernel of the first arrow is quasi-coherent and
we have seen above that $R^{n + 1}f_*g_*\mathcal{H}$ is quasi-coherent.
Thus $R^{n + 1}f_*\mathcal{F}$ has a $2$-step filtration
where the first step is quasi-coherent and the second a submodule of
a quasi-coherent sheaf. Since $\mathcal{F}$ is an arbitrary quasi-coherent
$\mathcal{O}_X$-module, this result also holds for $\mathcal{G}$.
Thus we can choose an exact sequence
$0 \to \mathcal{A} \to R^{n + 1}f_*\mathcal{G} \to \mathcal{B}$
with $\mathcal{A}$, $\mathcal{B}$ quasi-coherent $\mathcal{O}_Y$-modules.
Then the kernel $\mathcal{K}$ of
$R^{n + 1}f_*g_*\mathcal{H} \to R^{n + 1}f_*\mathcal{G}
\to \mathcal{B}$ is quasi-coherent, whereupon we obtain a map
$\mathcal{K} \to \mathcal{A}$ whose kernel $\mathcal{K}'$ is
quasi-coherent too. Hence $R^{n + 1}f_*\mathcal{F}$ sits in an exact
sequence
$$
R^nf_*g_*\mathcal{H} \to
R^nf_*\mathcal{G} \to
R^{n + 1}f_*\mathcal{F} \to \mathcal{K}' \to 0
$$
with all modules quasi-coherent except for possibly $R^{n + 1}f_*\mathcal{F}$.
We conclude that $R^{n + 1}f_*\mathcal{F}$ is quasi-coherent, i.e.,
$IH_{n + 1}$ holds as desired.
\end{proof}
\begin{lemma}
\label{lemma-quasi-coherence-higher-direct-images-application}
Let $S$ be a scheme. Let $f : X \to Y$ be a quasi-separated and quasi-compact
morphism of algebraic spaces over $S$. For any quasi-coherent
$\mathcal{O}_X$-module $\mathcal{F}$ and any affine object $V$ of
$Y_\etale$ we have
$$
H^q(V \times_Y X, \mathcal{F}) = H^0(V, R^qf_*\mathcal{F})
$$
for all $q \in \mathbf{Z}$.
\end{lemma}
\begin{proof}
Since formation of $Rf_*$ commutes with \'etale localization
(Properties of Spaces, Lemma
\ref{spaces-properties-lemma-pushforward-etale-base-change-modules})
we may replace $Y$ by $V$ and assume $Y = V$ is affine.
Consider the Leray spectral sequence
$E_2^{p, q} = H^p(Y, R^qf_*\mathcal{F})$
converging to $H^{p + q}(X, \mathcal{F})$, see
Cohomology on Sites, Lemma \ref{sites-cohomology-lemma-Leray}.
By Lemma \ref{lemma-higher-direct-image}
we see that the sheaves $R^qf_*\mathcal{F}$ are quasi-coherent. By
Cohomology of Schemes, Lemma
\ref{coherent-lemma-quasi-coherent-affine-cohomology-zero}
we see that $E_2^{p, q} = 0$ when $p > 0$.
Hence the spectral sequence degenerates at $E_2$ and we win.
\end{proof}
\section{Finite morphisms}
\label{section-finite-morphisms}
\noindent
Here are some results which hold for all abelian sheaves
(in particular also quasi-coherent modules).
We {\bf warn} the reader that these lemmas do not hold for
finite morphisms of schemes and the Zariski topology.
\begin{lemma}
\label{lemma-finite-higher-direct-image-zero}
Let $S$ be a scheme. Let $f : X \to Y$ be an integral (for example finite)
morphism of algebraic spaces. Then
$f_* : \textit{Ab}(X_\etale) \to \textit{Ab}(Y_\etale)$
is an exact functor and $R^pf_* = 0$ for $p > 0$.
\end{lemma}
\begin{proof}
By Properties of Spaces, Lemma
\ref{spaces-properties-lemma-pushforward-etale-base-change}
we may compute the higher direct images on an \'etale cover of $Y$.
Hence we may assume $Y$ is a scheme. This implies that
$X$ is a scheme (Morphisms of Spaces, Lemma
\ref{spaces-morphisms-lemma-integral-local}).
In this case we may apply
\'Etale Cohomology, Lemma \ref{etale-cohomology-lemma-what-integral}.
For the finite case the reader may wish to consult the less technical
\'Etale Cohomology, Proposition
\ref{etale-cohomology-proposition-finite-higher-direct-image-zero}.
\end{proof}
\begin{lemma}
\label{lemma-stalk-push-finite}
Let $S$ be a scheme. Let $f : X \to Y$ be a finite morphism of algebraic
spaces over $S$. Let $\overline{y}$ be a geometric point of $Y$ with
lifts $\overline{x}_1, \ldots, \overline{x}_n$ in $X$. Then
$$
(f_*\mathcal{F})_{\overline{y}} =
\prod\nolimits_{i = 1, \ldots, n}
\mathcal{F}_{\overline{x}_i}
$$
for any sheaf $\mathcal{F}$ on $X_\etale$.
\end{lemma}
\begin{proof}
Choose an \'etale neighbourhood $(V, \overline{v})$ of $\overline{y}$.
Then the stalk $(f_*\mathcal{F})_{\overline{y}}$
is the stalk of $f_*\mathcal{F}|_V$ at $\overline{v}$.
By Properties of Spaces,
Lemma \ref{spaces-properties-lemma-pushforward-etale-base-change}
we may replace $Y$ by $V$ and $X$ by $X \times_Y V$.
Then $Z \to X$ is a finite morphism of schemes and the result is
\'Etale Cohomology, Proposition
\ref{etale-cohomology-proposition-finite-higher-direct-image-zero}.
\end{proof}
\begin{lemma}
\label{lemma-finite-rings}
Let $S$ be a scheme. Let $\pi : X \to Y$ be a finite morphism of algebraic
spaces over $S$. Let $\mathcal{A}$ be a sheaf of rings on $X_\etale$.
Let $\mathcal{B}$ be a sheaf of rings on $Y_\etale$.
Let $\varphi : \mathcal{B} \to \pi_*\mathcal{A}$
be a homomorphism of sheaves of rings so that we obtain a
morphism of ringed topoi
$$
f = (\pi, \varphi) :
(\Sh(X_\etale), \mathcal{A})
\longrightarrow
(\Sh(Y_\etale), \mathcal{B}).
$$
For a sheaf of $\mathcal{A}$-modules $\mathcal{F}$ and a
sheaf of $\mathcal{B}$-modules $\mathcal{G}$ the canonical map
$$
\mathcal{G} \otimes_\mathcal{B} f_*\mathcal{F}
\longrightarrow
f_*(f^*\mathcal{G} \otimes_\mathcal{A} \mathcal{F}).
$$
is an isomorphism.
\end{lemma}
\begin{proof}
The map is the map adjoint to the map
$$
f^*\mathcal{G} \otimes_\mathcal{A}
f^* f_*\mathcal{F} =
f^*(\mathcal{G} \otimes_\mathcal{B} f_*\mathcal{F})
\longrightarrow
f^*\mathcal{G} \otimes_\mathcal{A} \mathcal{F}
$$
coming from $\text{id} : f^*\mathcal{G} \to f^*\mathcal{G}$
and the adjunction map $f^* f_*\mathcal{F} \to \mathcal{F}$.
To see this map is an isomorphism, we may check on stalks
(Properties of Spaces, Theorem
\ref{spaces-properties-theorem-exactness-stalks}).
Let $\overline{y}$ be a geometric point of $Y$ and
let $\overline{x}_1, \ldots, \overline{x}_n$ be the geometric
points of $X$ lying over $\overline{y}$.
Working out what our maps does on stalks, we see that we
have to show
$$
\mathcal{G}_{\overline{y}}
\otimes_{\mathcal{B}_{\overline{y}}}
\left(
\bigoplus\nolimits_{i = 1, \ldots, n} \mathcal{F}_{\overline{x}_i}
\right) =
\bigoplus\nolimits_{i = 1, \ldots, n}
(\mathcal{G}_{\overline{y}}
\otimes_{\mathcal{B}_{\overline{x}}}
\mathcal{A}_{\overline{x}_i}) \otimes_{\mathcal{A}_{\overline{x}_i}}
\mathcal{F}_{\overline{x}_i}
$$
which holds true. Here we have used that
taking tensor products commutes with taking stalks, the
behaviour of stalks under pullback
Properties of Spaces, Lemma \ref{spaces-properties-lemma-stalk-pullback}, and
the behaviour of stalks under pushforward along a closed immersion
Lemma \ref{lemma-stalk-push-finite}.
\end{proof}
\noindent
We end this section with an insanely general projection formula
for finite morphisms.
\begin{lemma}
\label{lemma-projection-formula-finite}
With $S$, $X$, $Y$, $\pi$, $\mathcal{A}$, $\mathcal{B}$, $\varphi$, and $f$
as in Lemma \ref{lemma-finite-rings} we have
$$
K \otimes_\mathcal{B}^\mathbf{L} Rf_*M =
Rf_*(Lf^*K \otimes_\mathcal{A}^\mathbf{L} M)
$$
in $D(\mathcal{B})$ for any $K \in D(\mathcal{B})$ and
$M \in D(\mathcal{A})$.
\end{lemma}
\begin{proof}
Since $f_*$ is exact (Lemma \ref{lemma-finite-higher-direct-image-zero})
the functor $Rf_*$ is computed by applying $f_*$ to any representative complex.
Choose a complex $\mathcal{K}^\bullet$ of $\mathcal{B}$-modules
representing $K$ which is K-flat with flat terms, see
Cohomology on Sites, Lemma \ref{sites-cohomology-lemma-K-flat-resolution}.
Then $f^*\mathcal{K}^\bullet$ is K-flat with flat terms, see
Cohomology on Sites, Lemma \ref{sites-cohomology-lemma-pullback-K-flat}.
Choose any complex $\mathcal{M}^\bullet$ of $\mathcal{A}$-modules
representing $M$. Then
we have to show
$$
\text{Tot}(\mathcal{K}^\bullet \otimes_\mathcal{B} f_*\mathcal{M}^\bullet)
=
f_*\text{Tot}(f^*\mathcal{K}^\bullet \otimes_\mathcal{A} \mathcal{M}^\bullet)
$$
because by our choices these complexes represent the right and left hand
side of the formula in the lemma.
Since $f_*$ commutes with direct sums
(for example by the description of the stalks in
Lemma \ref{lemma-stalk-push-finite}),
this reduces to the equalities
$$
\mathcal{K}^n \otimes_\mathcal{B} f_*\mathcal{M}^m
=
f_*(f^*\mathcal{K}^n \otimes_\mathcal{A} \mathcal{M}^m)
$$
which are true by Lemma \ref{lemma-finite-rings}.
\end{proof}
\section{Colimits and cohomology}
\label{section-colimits}
\noindent
The following lemma in particular applies to diagrams of quasi-coherent
sheaves.
\begin{lemma}
\label{lemma-colimits}
Let $S$ be a scheme. Let $X$ be an algebraic space over $S$.
If $X$ is quasi-compact and quasi-separated, then
$$
\colim_i H^p(X, \mathcal{F}_i)
\longrightarrow
H^p(X, \colim_i \mathcal{F}_i)
$$
is an isomorphism
for every filtered diagram of abelian sheaves on $X_\etale$.
\end{lemma}
\begin{proof}
This follows from
Cohomology on Sites, Lemma
\ref{sites-cohomology-lemma-colim-works-over-collection}.
Namely, let $\mathcal{B} \subset \Ob(X_{spaces, \etale})$
be the set of quasi-compact and quasi-separated spaces \'etale over $X$.
Note that if $U \in \mathcal{B}$ then, because $U$ is quasi-compact,
the collection of finite coverings $\{U_i \to U\}$ with $U_i \in \mathcal{B}$
is cofinal in the set of coverings of $U$ in $X_{spaces, \etale}$. By
Morphisms of Spaces, Lemma
\ref{spaces-morphisms-lemma-quasi-compact-quasi-separated-permanence}
the set $\mathcal{B}$ satisfies all the assumptions of
Cohomology on Sites, Lemma
\ref{sites-cohomology-lemma-colim-works-over-collection}.
Since $X \in \mathcal{B}$ we win.
\end{proof}
\begin{lemma}
\label{lemma-colimit-cohomology}
\begin{slogan}
Higher direct images of qcqs morphisms commute with filtered colimits
of sheaves.
\end{slogan}
Let $S$ be a scheme. Let $f : X \to Y$ be a quasi-compact and quasi-separated
morphism of algebraic spaces over $S$. Let $\mathcal{F} = \colim \mathcal{F}_i$
be a filtered colimit of abelian sheaves on $X_\etale$.
Then for any $p \geq 0$ we have
$$
R^pf_*\mathcal{F} = \colim R^pf_*\mathcal{F}_i.
$$
\end{lemma}
\begin{proof}
We will use that the morphism of topoi $f_{small} : X_{small} \to Y_{small}$
comes from the morphism of sites
$f_{spaces, \etale} : X_{spaces, \etale} \to Y_{spaces, \etale}$
corresponding to the continuous functor $V \longmapsto X \times_Y V$,
see Properties of Spaces, Lemma
\ref{spaces-properties-lemma-functoriality-etale-site}.
We will apply Cohomology on Sites, Lemma
\ref{sites-cohomology-lemma-higher-direct-image-colimit}
to this morphism of sites. Since every object of $Y_{spaces, \etale}$
has a covering by affine objects, it suffices to show that
for $V$ affine and \'etale over $Y$ we have
$H^p(X \times_Y V, \mathcal{F}) = \colim H^p(X \times_Y V, \mathcal{F}_i)$.
Since $V$ is affine, the algebraic space $X \times_Y V$ is
quasi-compact and quasi-separated.
Hence we can apply Lemma \ref{lemma-colimits}
to conclude.
\end{proof}
\noindent
The following lemma tells us that finitely presented modules behave
as expected in quasi-compact and quasi-separated algebraic spaces.
\begin{lemma}
\label{lemma-finite-presentation-quasi-compact-colimit}
Let $S$ be a scheme. Let $X$ be a quasi-compact and quasi-separated
algebraic space over $S$. Let $I$ be a directed set and
let $(\mathcal{F}_i, \varphi_{ii'})$ be a system over $I$
of $\mathcal{O}_X$-modules. Let $\mathcal{G}$ be an
$\mathcal{O}_X$-module of finite presentation. Then we have
$$
\colim_i \Hom_X(\mathcal{G}, \mathcal{F}_i)
=
\Hom_X(\mathcal{G}, \colim_i \mathcal{F}_i).
$$
In particular, $\Hom_X(\mathcal{G}, -)$ commutes with filtered
colimits in $\QCoh(\mathcal{O}_X)$.
\end{lemma}
\begin{proof}
The displayed equality is a special case of Modules on Sites, Lemma
\ref{sites-modules-lemma-finite-presentation-quasi-compact-colimit}.
In order to apply it, we need to check the hypotheses of
Sites, Lemma \ref{sites-lemma-directed-colimits-global-sections} part (4)
for the site $X_\etale$.
In order to do this, we will check hypotheses
(2)(a), (2)(b), (2)(c) of
Sites, Remark \ref{sites-remark-stronger-conditions}.
Namely, let $\mathcal{B} \subset \Ob(X_\etale)$ be the set of affine objects.
Then
\begin{enumerate}
\item Since $X$ is quasi-compact, there exists a $U \in \mathcal{B}$
such that $U \to X$ is surjective
(Properties of Spaces, Lemma
\ref{spaces-properties-lemma-quasi-compact-affine-cover}),
hence $h_U^\# \to *$ is surjective.
\item For $U \in \mathcal{B}$ every \'etale covering $\{U_i \to U\}_{i \in I}$
of $U$ can be refined by a finite \'etale covering
$\{U_j \to U\}_{j = 1, \ldots, m}$ with $U_j \in \mathcal{B}$
(Topologies, Lemma \ref{topologies-lemma-etale-affine}).
\item For $U, U' \in \Ob(X_\etale)$ we have
$h_U^\# \times h_{U'}^\# = h_{U \times_X U'}^\#$.
If $U, U' \in \mathcal{B}$, then $U \times_X U'$ is quasi-compact
because $X$ is quasi-separated, see Morphisms of Spaces, Lemma
\ref{spaces-morphisms-lemma-quasi-compact-quasi-separated-permanence}
for example. Hence we can find a surjective \'etale morphism
$U'' \to U \times_X U'$ with $U'' \in \mathcal{B}$
(Properties of Spaces, Lemma
\ref{spaces-properties-lemma-quasi-compact-affine-cover}).
In other words, we have morphisms $U'' \to U$ and $U'' \to U'$
such that the map $h_{U''}^\# \to h_U^\# \times h_{u'}^\#$ is
surjective.
\end{enumerate}
For the final statement, observe that the inclusion functor
$\QCoh(\mathcal{O}_X) \to \textit{Mod}(\mathcal{O}_X)$
commutes with colimits and that finitely presented modules
are quasi-coherent. See
Properties of Spaces, Lemma
\ref{spaces-properties-lemma-properties-quasi-coherent}.
\end{proof}
\section{The alternating {\v C}ech complex}
\label{section-alternating-cech}
\noindent
Let $S$ be a scheme. Let $f : U \to X$ be an \'etale morphism of algebraic
spaces over $S$. The functor
$$
j : U_{spaces, \etale} \longrightarrow X_{spaces, \etale},\quad
V/U \longmapsto V/X
$$
induces an equivalence of $U_{spaces, \etale}$ with the localization
$X_{spaces, \etale}/U$, see
Properties of Spaces, Section \ref{spaces-properties-section-localize}.
Hence there exist functors
$$
f_! : \textit{Ab}(U_\etale) \longrightarrow
\textit{Ab}(X_\etale),\quad
f_! : \textit{Mod}(\mathcal{O}_U) \longrightarrow \textit{Mod}(\mathcal{O}_X),
$$
which are left adjoint to
$$
f^{-1} : \textit{Ab}(X_\etale) \longrightarrow
\textit{Ab}(U_\etale),\quad
f^* : \textit{Mod}(\mathcal{O}_X) \longrightarrow \textit{Mod}(\mathcal{O}_U)
$$
see
Modules on Sites, Section \ref{sites-modules-section-localize}.
Warning: This functor, a priori, has
nothing to do with cohomology with compact supports!
We dubbed this functor ``extension by zero'' in the reference above.
Note that the two versions of $f_!$ agree as $f^* = f^{-1}$ for
sheaves of $\mathcal{O}_X$-modules.
\medskip\noindent
As we are going to use this construction below let us recall some of its
properties. Given an abelian sheaf $\mathcal{G}$ on $U_\etale$
the sheaf $f_!$ is the sheafification of the presheaf
$$
V/X \longmapsto
f_!\mathcal{G}(V) =
\bigoplus\nolimits_{\varphi \in \Mor_X(V, U)}
\mathcal{G}(V \xrightarrow{\varphi} U),
$$
see
Modules on Sites, Lemma \ref{sites-modules-lemma-extension-by-zero}.
Moreover, if $\mathcal{G}$ is an $\mathcal{O}_U$-module, then $f_!\mathcal{G}$
is the sheafification of the exact same presheaf of abelian groups which
is endowed with an $\mathcal{O}_X$-module structure in an obvious way
(see loc.\ cit.). Let $\overline{x} : \Spec(k) \to X$
be a geometric point. Then there is a canonical identification
$$
(f_!\mathcal{G})_{\overline{x}} =
\bigoplus\nolimits_{\overline{u}} \mathcal{G}_{\overline{u}}
$$
where the sum is over all $\overline{u} : \Spec(k) \to U$ such that
$f \circ \overline{u} = \overline{x}$, see
Modules on Sites, Lemma \ref{sites-modules-lemma-stalk-j-shriek}
and
Properties of Spaces, Lemma
\ref{spaces-properties-lemma-points-small-etale-site}.
In the following we are going to study the sheaf $f_!\underline{\mathbf{Z}}$.
Here $\underline{\mathbf{Z}}$ denotes the constant sheaf on
$X_\etale$ or $U_\etale$.
\begin{lemma}
\label{lemma-product-is-tensor-product}
Let $S$ be a scheme. Let $f_i : U_i \to X$ be \'etale morphisms
of algebraic spaces over $S$. Then there are isomorphisms
$$
f_{1, !}\underline{\mathbf{Z}} \otimes_{\mathbf{Z}}
f_{2, !}\underline{\mathbf{Z}}
\longrightarrow
f_{12, !}\underline{\mathbf{Z}}
$$
where $f_{12} : U_1 \times_X U_2 \to X$ is the structure morphism
and
$$
(f_1 \amalg f_2)_! \underline{\mathbf{Z}}
\longrightarrow
f_{1, !}\underline{\mathbf{Z}} \oplus
f_{2, !}\underline{\mathbf{Z}}
$$
\end{lemma}
\begin{proof}
Once we have defined the map it will be an isomorphism by our description
of stalks above. To define the map it suffices to work on the level of
presheaves. Thus we have to define a map
$$
\left(\bigoplus\nolimits_{\varphi_1 \in \Mor_X(V, U_1)} \mathbf{Z}\right)
\otimes_{\mathbf{Z}}
\left(\bigoplus\nolimits_{\varphi_2 \in \Mor_X(V, U_2)} \mathbf{Z}\right)
\longrightarrow
\bigoplus\nolimits_{\varphi \in \Mor_X(V, U_1 \times_X U_2)}
\mathbf{Z}
$$
We map the element $1_{\varphi_1} \otimes 1_{\varphi_2}$ to the element
$1_{\varphi_1 \times \varphi_2}$ with obvious notation. We omit the proof
of the second equality.
\end{proof}
\noindent
Another important feature is the trace map
$$
\text{Tr}_f : f_!\underline{\mathbf{Z}} \longrightarrow \underline{\mathbf{Z}}.
$$
The trace map is adjoint to the
map $\mathbf{Z} \to f^{-1}\underline{\mathbf{Z}}$ (which is an isomorphism).
If $\overline{x}$ is above, then $\text{Tr}_f$ on stalks at $\overline{x}$
is the map
$$
(\text{Tr}_f)_{\overline{x}} :
(f_!\underline{\mathbf{Z}})_{\overline{x}} =
\bigoplus\nolimits_{\overline{u}} \mathbf{Z}
\longrightarrow
\mathbf{Z} = \underline{\mathbf{Z}}_{\overline{x}}
$$
which sums the given integers. This is true because it is adjoint to the map
$1 : \mathbf{Z} \to f^{-1}\underline{\mathbf{Z}}$. In particular, if
$f$ is surjective as well as \'etale then $\text{Tr}_f$ is surjective.
\medskip\noindent
Assume that $f : U \to X$ is a surjective \'etale
morphism of algebraic spaces. Consider the {\it Koszul complex}
associated to the trace map we discussed above
$$
\ldots \to \wedge^3f_!\underline{\mathbf{Z}} \to
\wedge^2f_!\underline{\mathbf{Z}} \to f_!\underline{\mathbf{Z}} \to
\underline{\mathbf{Z}} \to 0
$$
Here the exterior powers are over the sheaf of rings $\underline{\mathbf{Z}}$.
The maps are defined by the rule
$$
e_1 \wedge \ldots \wedge e_n \longmapsto
\sum\nolimits_{i = 1, \ldots, n} (-1)^{i + 1}
\text{Tr}_f(e_i)
e_1 \wedge \ldots \wedge \widehat{e_i} \wedge \ldots \wedge e_n
$$
where $e_1, \ldots, e_n$ are local sections of $f_!\underline{\mathbf{Z}}$.
Let $\overline{x}$ be a geometric point of $X$ and set
$M_{\overline{x}} = (f_!\underline{\mathbf{Z}})_{\overline{x}} =
\bigoplus_{\overline{u}} \mathbf{Z}$. Then the stalk of the complex above at
$\overline{x}$ is the complex
$$
\ldots \to \wedge^3 M_{\overline{x}} \to \wedge^2 M_{\overline{x}}
\to M_{\overline{x}} \to \mathbf{Z} \to 0
$$
which is exact because $M_{\overline{x}} \to \mathbf{Z}$ is surjective, see
More on Algebra, Lemma \ref{more-algebra-lemma-homotopy-koszul-abstract}.
Hence if we let $K^\bullet = K^\bullet(f)$ be the complex with
$K^i = \wedge^{i + 1}f_!\underline{\mathbf{Z}}$, then we obtain a
quasi-isomorphism
\begin{equation}
\label{equation-quasi-isomorphism}
K^\bullet \longrightarrow \underline{\mathbf{Z}}[0]
\end{equation}
We use the complex $K^\bullet$ to define what we call
the alternating {\v C}ech complex associated to $f : U \to X$.
\begin{definition}
\label{definition-alternating-cech-complex}
Let $S$ be a scheme. Let $f : U \to X$ be a surjective \'etale morphism
of algebraic spaces over $S$. Let $\mathcal{F}$ be an object of
$\textit{Ab}(X_\etale)$. The
{\it alternating {\v C}ech complex}\footnote{This may be nonstandard notation}
$\check{\mathcal{C}}^\bullet_{alt}(f, \mathcal{F})$
associated to $\mathcal{F}$ and $f$ is the complex
$$
\Hom(K^0, \mathcal{F}) \to \Hom(K^1, \mathcal{F}) \to
\Hom(K^2, \mathcal{F}) \to \ldots
$$
with Hom groups computed in $\textit{Ab}(X_\etale)$.
\end{definition}
\noindent
The reader may verify that if $U = \coprod U_i$ and $f|_{U_i} : U_i \to X$
is the open immersion of a subspace, then
$\check{\mathcal{C}}_{alt}^\bullet(f, \mathcal{F})$ agrees with the complex
introduced in
Cohomology, Section \ref{cohomology-section-alternating-cech}
for the Zariski covering $X = \bigcup U_i$ and the restriction
of $\mathcal{F}$ to the Zariski site of $X$. What is more important
however, is to relate the cohomology of the alternating
{\v C}ech complex to the cohomology.
\begin{lemma}
\label{lemma-alternating-cech-to-cohomology}
Let $S$ be a scheme. Let $f : U \to X$ be a surjective \'etale morphism
of algebraic spaces over $S$. Let $\mathcal{F}$ be an object of
$\textit{Ab}(X_\etale)$. There exists a canonical map
$$
\check{\mathcal{C}}^\bullet_{alt}(f, \mathcal{F})
\longrightarrow
R\Gamma(X, \mathcal{F})
$$
in $D(\textit{Ab})$. Moreover, there is a spectral sequence with $E_1$-page
$$
E_1^{p, q} =
\Ext_{\textit{Ab}(X_\etale)}^q(K^p, \mathcal{F})
$$
converging to $H^{p + q}(X, \mathcal{F})$ where
$K^p = \wedge^{p + 1}f_!\underline{\mathbf{Z}}$.
\end{lemma}
\begin{proof}
Recall that we have the quasi-isomorphism
$K^\bullet \to \underline{\mathbf{Z}}[0]$, see
(\ref{equation-quasi-isomorphism}).
Choose an injective resolution $\mathcal{F} \to \mathcal{I}^\bullet$
in $\textit{Ab}(X_\etale)$. Consider the double complex
$\Hom(K^\bullet, \mathcal{I}^\bullet)$ with terms
$\Hom(K^p, \mathcal{I}^q)$. The differential
$d_1^{p, q} : A^{p, q} \to A^{p + 1, q}$
is the one coming from the differential $K^{p + 1} \to K^p$
and the differential $d_2^{p, q} : A^{p, q} \to A^{p, q + 1}$ is the
one coming from the differential
$\mathcal{I}^q \to \mathcal{I}^{q + 1}$.
Denote $\text{Tot}(\Hom(K^\bullet, \mathcal{I}^\bullet))$
the associated total complex, see
Homology, Section \ref{homology-section-double-complexes}.
We will use the two spectral
sequences $({}'E_r, {}'d_r)$ and $({}''E_r, {}''d_r)$
associated to this double complex, see
Homology, Section \ref{homology-section-double-complex}.
\medskip\noindent
Because $K^\bullet$ is a resolution of $\underline{\mathbf{Z}}$
we see that the complexes
$$
\Hom(K^\bullet, \mathcal{I}^q) :
\Hom(K^0, \mathcal{I}^q) \to
\Hom(K^1, \mathcal{I}^q) \to
\Hom(K^2, \mathcal{I}^q) \to \ldots
$$
are acyclic in positive degrees and have $H^0$ equal to
$\Gamma(X, \mathcal{I}^q)$. Hence by
Homology, Lemma \ref{homology-lemma-double-complex-gives-resolution}
the natural map
$$
\mathcal{I}^\bullet(X) \longrightarrow
\text{Tot}(\Hom(K^\bullet, \mathcal{I}^\bullet))
$$
is a quasi-isomorphism of complexes of abelian groups. In particular
we conclude that
$H^n(\text{Tot}(\Hom(K^\bullet, \mathcal{I}^\bullet))) = H^n(X, \mathcal{F})$.
\medskip\noindent
The map $\check{\mathcal{C}}^\bullet_{alt}(f, \mathcal{F}) \to
R\Gamma(X, \mathcal{F})$ of the lemma is the composition of
$\check{\mathcal{C}}^\bullet_{alt}(f, \mathcal{F}) \to
\text{Tot}(\Hom(K^\bullet, \mathcal{I}^\bullet))$
with the inverse of the displayed quasi-isomorphism.
\medskip\noindent
Finally, consider the spectral sequence $({}'E_r, {}'d_r)$.
We have
$$
E_1^{p, q} = q\text{th cohomology of }
\Hom(K^p, \mathcal{I}^0) \to
\Hom(K^p, \mathcal{I}^1) \to
\Hom(K^p, \mathcal{I}^2) \to \ldots
$$
This proves the lemma.
\end{proof}
\noindent
It follows from the lemma that it is important to understand the
ext groups $\Ext_{\textit{Ab}(X_\etale)}(K^p, \mathcal{F})$,
i.e., the right derived functors of
$\mathcal{F} \mapsto \Hom(K^p, \mathcal{F})$.
\begin{lemma}
\label{lemma-compute}
Let $S$ be a scheme. Let $f : U \to X$ be a surjective, \'etale, and separated
morphism of algebraic spaces over $S$. For $p \geq 0$ set
$$
W_p = U \times_X \ldots \times_X U \setminus \text{all diagonals}
$$
where the fibre product has $p + 1$ factors.
There is a free action of $S_{p + 1}$ on $W_p$ over $X$ and
$$
\Hom(K^p, \mathcal{F}) = S_{p + 1}\text{-anti-invariant elements of }
\mathcal{F}(W_p)
$$
functorially in $\mathcal{F}$ where
$K^p = \wedge^{p + 1}f_!\underline{\mathbf{Z}}$.
\end{lemma}
\begin{proof}
Because $U \to X$ is separated the diagonal $U \to U \times_X U$ is a
closed immersion. Since $U \to X$ is \'etale the diagonal
$U \to U \times_X U$ is an open immersion, see
Morphisms of Spaces, Lemmas
\ref{spaces-morphisms-lemma-etale-unramified} and
\ref{spaces-morphisms-lemma-diagonal-unramified-morphism}.
Hence $W_p$ is an open and closed subspace of
$U^{p + 1} = U \times_X \ldots \times_X U$. The action of $S_{p + 1}$
on $W_p$ is free as we've thrown out the fixed points of the action.
By
Lemma \ref{lemma-product-is-tensor-product}
we see that
$$
(f_!\underline{\mathbf{Z}})^{\otimes p + 1} =
f^{p + 1}_!\underline{\mathbf{Z}} = (W_p \to X)_!\underline{\mathbf{Z}}
\oplus Rest
$$
where $f^{p + 1} : U^{p + 1} \to X$ is the structure morphism.
Looking at stalks over a geometric point $\overline{x}$ of $X$
we see that
$$
\left(
\bigoplus\nolimits_{\overline{u} \mapsto \overline{x}} \mathbf{Z}
\right)^{\otimes p + 1}
\longrightarrow
(W_p \to X)_!\underline{\mathbf{Z}}_{\overline{x}}
$$
is the quotient whose kernel is generated by all tensors
$1_{\overline{u}_0} \otimes \ldots \otimes 1_{\overline{u}_p}$
where $\overline{u}_i = \overline{u}_j$ for some $i \not = j$.
Thus the quotient map
$$
(f_!\underline{\mathbf{Z}})^{\otimes p + 1}
\longrightarrow
\wedge^{p + 1}f_!\underline{\mathbf{Z}}
$$
factors through $(W_p \to X)_!\underline{\mathbf{Z}}$, i.e., we get
$$
(f_!\underline{\mathbf{Z}})^{\otimes p + 1}
\longrightarrow
(W_p \to X)_!\underline{\mathbf{Z}}
\longrightarrow
\wedge^{p + 1}f_!\underline{\mathbf{Z}}
$$
This already proves that $\Hom(K^p, \mathcal{F})$ is (functorially) a
subgroup of
$$
\Hom((W_p \to X)_!\underline{\mathbf{Z}}, \mathcal{F}) = \mathcal{F}(W_p)
$$
To identify it with the $S_{p + 1}$-anti-invariants we have to prove that
the surjection $(W_p \to X)_!\underline{\mathbf{Z}}
\to \wedge^{p + 1}f_!\underline{\mathbf{Z}}$ is the maximal
$S_{p + 1}$-anti-invariant quotient. In other words, we have to show that
$\wedge^{p + 1}f_!\underline{\mathbf{Z}}$ is the quotient of
$(W_p \to X)_!\underline{\mathbf{Z}}$ by the subsheaf generated by
the local sections $s - \text{sign}(\sigma)\sigma(s)$ where $s$ is
a local section of $(W_p \to X)_!\underline{\mathbf{Z}}$.
This can be checked on the stalks, where it is clear.
\end{proof}
\begin{lemma}
\label{lemma-twist}
Let $S$ be a scheme. Let $W$ be an algebraic space over $S$.
Let $G$ be a finite group acting freely on $W$.
Let $U = W/G$, see
Properties of Spaces, Lemma \ref{spaces-properties-lemma-quotient}.
Let $\chi : G \to \{+1, -1\}$ be a character.
Then there exists a rank 1 locally free sheaf of $\mathbf{Z}$-modules
$\underline{\mathbf{Z}}(\chi)$ on $U_\etale$ such that for every
abelian sheaf $\mathcal{F}$ on $U_\etale$ we have
$$
H^0(W, \mathcal{F}|_W)^\chi =
H^0(U, \mathcal{F} \otimes_{\mathbf{Z}} \underline{\mathbf{Z}}(\chi))
$$
\end{lemma}
\begin{proof}
The quotient morphism $q : W \to U$ is a $G$-torsor, i.e., there exists
a surjective \'etale morphism $U' \to U$ such that
$W \times_U U' = \coprod_{g \in G} U'$ as spaces with $G$-action over $U'$.
(Namely, $U' = W$ works.) Hence $q_*\underline{\mathbf{Z}}$ is a finite
locally free $\mathbf{Z}$-module with an action of $G$. For any
geometric point $\overline{u}$ of $U$, then we get $G$-equivariant
isomorphisms
$$
(q_*\underline{\mathbf{Z}})_{\overline{u}}
= \bigoplus\nolimits_{\overline{w} \mapsto \overline{u}} \mathbf{Z}
= \bigoplus\nolimits_{g \in G} \mathbf{Z} = \mathbf{Z}[G]
$$
where the second $=$ uses a geometric point
$\overline{w}_0$ lying over $\overline{u}$ and
maps the summand corresponding to $g \in G$ to the summand
corresponding to $g(\overline{w}_0)$. We have
$$
H^0(W, \mathcal{F}|_W) =
H^0(U, \mathcal{F} \otimes_\mathbf{Z} q_*\underline{\mathbf{Z}})
$$
because
$q_*\mathcal{F}|_W = \mathcal{F} \otimes_\mathbf{Z} q_*\underline{\mathbf{Z}}$
as one can check by restricting to $U'$. Let
$$
\underline{\mathbf{Z}}(\chi) =
(q_*\underline{\mathbf{Z}})^\chi \subset
q_*\underline{\mathbf{Z}}
$$
be the subsheaf of sections that transform according to $\chi$. For
any geometric point $\overline{u}$ of $U$ we have
$$
\underline{\mathbf{Z}}(\chi)_{\overline{u}} =
\mathbf{Z} \cdot \sum\nolimits_g \chi(g) g
\subset
\mathbf{Z}[G] = (q_*\underline{\mathbf{Z}})_{\overline{u}}
$$
It follows that $\underline{\mathbf{Z}}(\chi)$ is locally free of
rank 1 (more precisely, this should be checked after restricting to $U'$).
Note that for any $\mathbf{Z}$-module $M$ the $\chi$-semi-invariants
of $M[G]$ are the elements of the form $m \cdot \sum\nolimits_g \chi(g) g$.
Thus we see that for any abelian sheaf $\mathcal{F}$ on $U$ we have
$$
\left(\mathcal{F} \otimes_\mathbf{Z} q_*\underline{\mathbf{Z}}\right)^\chi
=
\mathcal{F} \otimes_\mathbf{Z} \underline{\mathbf{Z}}(\chi)
$$
because we have equality at all stalks. The result of the lemma follows by
taking global sections.
\end{proof}
\noindent
Now we can put everything together and obtain the following
pleasing result.
\begin{lemma}
\label{lemma-alternating-spectral-sequence}