From 36175bfc0f247391ac34af244c97503e140baf39 Mon Sep 17 00:00:00 2001
From: Lutetium-Vanadium <luv.s7000@gmail.com>
Date: Wed, 17 May 2023 14:02:26 +0800
Subject: [PATCH] Added count-sketch

---
 benchmark/torchscripts/CountSketch.pt | Bin 0 -> 2782 bytes
 benchmark/torchscripts/CountSketch.py |  55 ++++++++++++++++++++++++++
 2 files changed, 55 insertions(+)
 create mode 100644 benchmark/torchscripts/CountSketch.pt
 create mode 100644 benchmark/torchscripts/CountSketch.py

diff --git a/benchmark/torchscripts/CountSketch.pt b/benchmark/torchscripts/CountSketch.pt
new file mode 100644
index 0000000000000000000000000000000000000000..71459484bf4c18c130d408d41d0a9c4ce5404b5e
GIT binary patch
literal 2782
zcmbtW2{_c<8Xx<F&@e<vd}N<2S>lF~h8n{tV~IvHW0=V>%wmg>C5*B(zLfctY!#-A
zC0ck0Wl3G#lzj=2y`<1`XTG|9sOP))dG7mv&i|bMIp_VI_nh<o-}i@v3kU)LKp@~J
zkpV~m_LJ!Wv?G3a8ji5r8$-iD0{uuZYf-?_9}>-v^YWsRDL8_cmnf2i!QlyHk~f}$
zAbZnEc+nt$W=y1p8xNjF!4Lz80lq<eiF9lbKLiOE0h$6ZV`khjUO{p#k>OhMvnh%I
z^L3Ws$lmzfKUReVhQX{&)|m9ew*RG($wZ{~=t(8at!hWm@@e4gIlEH9lP|?Sn=U4#
z;S&XZKaKD&DqVUJ!m6%sQE`(H3>ygXVfhL#fp5!xzV;ElosI&&L&?$A0#`Ff9fA76
z;|Z7Jq?-*AEb<;#CC$F!Ovkf^4<Dpk?d>5RgXT-AX;m&iK*W?@Yu(<-GG6@+_hc@-
zFfID2fvsgcr>Ei}>MMb&nhAeev|E{FGAw;C{m}3Z#DcH(pdfnJ*?f-79}`v{l|5o}
z4dey_!=yH?Uzl#<T)G?lFS@z{TeMn6n4ZVkSyy|Ws3RcLXb>tV_!c_U)#^sYW+FvG
z8Uv|OCo@!MJXBQg-bmpyS@zJoF1jFRIB(Q-+&iC|?3HUKZZ)lV%AaMB)W;zSC&*8l
zK{thCx_sFC$D0i9xM|hMeT=sq3C&V6zfc=h_fqhar{+wxbf|7=Z-k}!b?{B&nUdQc
zRv!`C=?!C5PU96*jf~Ur7JTi%c4<O$vr|I}oZva_X)aqbe-YNP90_D7#~q97N>$mm
zHT=CxE}CU!VRQO1PB?BL`j%ZJ$}z?}sAEFwOxKLkVx7CQi%l`K`9oP+OX=9oXCk*_
z+jqE28qAg@6}rs{lM<EoUNj0EX*u77GOeFgdiMxXzG|aUQ?ALhsMe${mLYQ8Q)17N
zhpwEOr;P?771R;z5n%9Fprzh#p8n3aa;nXrgHtTx>z<8HD)C&iWsAl(Fh2kgApCc(
zsSc3+d)I_`<FRyKuRs4A&@Y@c@;sch4JDJX4`)u_HTS{1M!CQ~4p$UU64x}{qWb)?
zrb;C<=8@Yu!=$2xlzln+X`=au!E773X}5VPy8}AOf*??d6*fumsx?*!tRn@w>W3Z+
zWBZ<Xf?yu~ec#uJuTdQ}A+w&$%3-9&;skCgtz*=<!`p@W+F;7JgFR|6Sx{*^opwW6
zqB~>P@MUaUkKh;blzC&1vAvPwVbwejhZ-~ZxP$Zq%8_KnuJEV(?NA|pD8{IC-`?Vv
z=`dre-o!Di@r})i?!`VXOH}zBGI7*3wO#3%aMQhuA;rz*H7NAXLd1$td!9PsxtXN%
zBaNy~`yKsvXOI+BS#Ra5Q=OXQ+AH<*mximetNq>ELkRb=1NxAOh*<~SgdwPiD|RNQ
z>h7D`GyRlGq&>@|FYd@S1+Z(`-Aa*9=(e#-V>!Lg`mLQ1MTx%edzxE1>$f>1Th_R5
z&$XSpnQE07X5gU1vFl7)=}YeKwQ^x)6-nv#!y<bsQhjcv4J#~z92{8TQHP`Wh5DVQ
zzpOwggOnZb2BhQ`Pjow>2A*7OYLeK4@DM>r%(Ww86&4VFn;nf(Z96G72b;Fk_+P_#
z?DaN(Ua{Hl-Pz^GZ>!469s902RF{)7tZSSNb>F58%aR8vDaYVtwiQm6RY%w$feRhW
zY<9=1u7&>X?<Mm-=m$S)ElgKqFh*4y^!;}M?jySGn_3jYIiu*lFm_4_HdQG&z@Os5
z%74>QfQa?dT8fL}zn_Yx&`5j29>S<Gup=R)J#qvag>6$xE@bk8Q+8#>9abt1WD<N7
zo~P@slO4`+K56EzL7r-{F37MK(3y0}&FUMe^Qpaap~Q9GfpcD*HE4q$tZl3Gbym>Y
zlN!Os$}Ana!$BDy>fIZWqZW}avs!%#MGyHSfsoi8^5W~q58>$oT22=eh;cBY*ghDp
zB2%m4Y}GRfg^^qbE&Wx&6gxdN{d<Kb$&W!M4Cq_i=uea2xvTvZXO=oGdK^V=J>RL=
zFo-d;ww>InSR_r$a2Qq`G0HA{m=!4*$w1?reV%F>X*WbWlQNVc76q3L6?~*5duGKl
zgdnwfmWB$LKx2Z3WQMq06jc0$6h3eVxZ=>2D<sS<bt^-+M%lCn)Z(o3@(_W6A<BOK
z&BT=YY?b@11-ei$Q9CxoW+guB^Wl82r0yw|&zIm2@<d}VM-Z=)aP!4(%O>dqpz<ij
z*)LlrqdaFrBYWi=y<>vg_-Dr6s%=8Udg^y-is9NBw3s#q$P?O~^dz~>RW%fhmy3Pv
zjk)pqK=?t4e_V0=WX+<#y1pqd!u7PD#>uUHO0o%VQpDu8V$i)aoD20Zw}~4`iw7Z=
z%pn{zQXIm_^pcs(3_RZWgsb@8FzXz!*ZASsZBRriEt>Bvkp9AE#q`3U-Mm06R2UD5
zVh+hM_lu($K(Wwc`D9yu!~S`(AmjF-T#Z%_&?g2Z4r;6=zQD8-^ZWuCo;*%t){8mi
zPN}7%$2kYsH-l5^SFvM~HY-kh^hOYMQogS~2HX~WDXsemhBu8N@q2XxsQ#I1UqpBz
z(6$+xoyBDbgPXpEz+dwOd4TeU5EwwEVFGB>Z)8F77eQ}LVvgq5{F4dz@80CfT?)V}
zJSOn#wWK#NfnYp^N+bvTANk)<l(*A==J=wxu^zxHNH`y_`oW3TaugZfz&iOAL;8mB
z7nfO^N)*>=Cv(5nBK*W`HbSkv67_XZY6AZS>aR?+affTqUVa^@2A7Pjw{OkeZiHET
zh|=p|m>XceUmFrGEUfc=;tm1bA1XrMTmPv~@O|q5as7}dTs;w99U;nl0|2~XBwUD_
Pr~v>W?!%k^ZSKDT5!ED)

literal 0
HcmV?d00001

diff --git a/benchmark/torchscripts/CountSketch.py b/benchmark/torchscripts/CountSketch.py
new file mode 100644
index 00000000..84685928
--- /dev/null
+++ b/benchmark/torchscripts/CountSketch.py
@@ -0,0 +1,55 @@
+import torch
+import time
+import os
+import math
+
+
+@torch.jit.script
+def CountSketch(A: torch.Tensor, B: torch.Tensor, s: int):
+    m1, n = A.shape
+    n, m2 = B.shape
+    # initialize sketch matrices
+    Ca = torch.zeros((m1, s))
+    Cb = torch.zeros((s, m2))
+
+    L = torch.randint(s, (n, ))
+    G = torch.randint(2, (n, ))
+
+    # modify the random column with random sign
+    for i in range(0, n):
+        if G[i] == 1:
+            Ca[:, L[i]] += A[:, i]
+            Cb[L[i], :] += B[i, :]
+        else:
+            Ca[:, L[i]] -= A[:, i]
+            Cb[L[i], :] -= B[i, :]
+
+    return torch.matmul(Ca, Cb)
+
+
+def main():
+    width = 1000
+    A = torch.rand(10000, width)
+    B = torch.rand(width, 5000)
+
+    t = time.time()
+
+    aResult = CountSketch(A, B, 500)
+    print("approximate: " + str(time.time() - t) + "s")
+
+    print(aResult)
+
+    # exact result
+    t = time.time()
+    eResult = torch.matmul(A, B)
+    print("\nExact: " + str(time.time() - t) + "s")
+
+    print(eResult)
+
+    print("\nerror: " + str(torch.norm(aResult - eResult, p='fro').item()))
+
+    CountSketch_script = CountSketch.save("CountSketch.pt")
+
+
+if __name__ == '__main__':
+    main()