import bisect import copy import decimal import fractions import heapq import itertools import math import random import sys import time from collections import Counter,deque,defaultdict from functools import lru_cache,reduce from heapq import heappush,heappop,heapify,heappushpop,_heappop_max,_heapify_max def _heappush_max(heap,item): heap.append(item) heapq._siftdown_max(heap, 0, len(heap)-1) def _heappushpop_max(heap, item): if heap and item < heap[0]: item, heap[0] = heap[0], item heapq._siftup_max(heap, 0) return item from math import gcd as GCD read=sys.stdin.read readline=sys.stdin.readline readlines=sys.stdin.readlines write=sys.stdout.write def Tonelli_Shanks(N,p): if pow(N,p>>1,p)==p-1: retu=None elif p%4==3: retu=pow(N,(p+1)//4,p) else: for nonresidue in range(1,p): if pow(nonresidue,p>>1,p)==p-1: break pp=p-1 cnt=0 while pp%2==0: pp//=2 cnt+=1 s=pow(N,pp,p) retu=pow(N,(pp+1)//2,p) for i in range(cnt-2,-1,-1): if pow(s,1<>1+i,p) s%=p retu*=pow(nonresidue,p>>2+i,p) retu%=p return retu #mod = 998244353 imag = 911660635 iimag = 86583718 rate2 = (911660635, 509520358, 369330050, 332049552, 983190778, 123842337, 238493703, 975955924, 603855026, 856644456, 131300601, 842657263, 730768835, 942482514, 806263778, 151565301, 510815449, 503497456, 743006876, 741047443, 56250497, 867605899) irate2 = (86583718, 372528824, 373294451, 645684063, 112220581, 692852209, 155456985, 797128860, 90816748, 860285882, 927414960, 354738543, 109331171, 293255632, 535113200, 308540755, 121186627, 608385704, 438932459, 359477183, 824071951, 103369235) rate3 = (372528824, 337190230, 454590761, 816400692, 578227951, 180142363, 83780245, 6597683, 70046822, 623238099, 183021267, 402682409, 631680428, 344509872, 689220186, 365017329, 774342554, 729444058, 102986190, 128751033, 395565204) irate3 = (509520358, 929031873, 170256584, 839780419, 282974284, 395914482, 444904435, 72135471, 638914820, 66769500, 771127074, 985925487, 262319669, 262341272, 625870173, 768022760, 859816005, 914661783, 430819711, 272774365, 530924681) def butterfly(a): n = len(a) h = (n - 1).bit_length() len_ = 0 while len_ < h: if h - len_ == 1: p = 1 << (h - len_ - 1) rot = 1 for s in range(1 << len_): offset = s << (h - len_) for i in range(p): l = a[i + offset] r = a[i + offset + p] * rot % mod a[i + offset] = (l + r) % mod a[i + offset + p] = (l - r) % mod if s + 1 != 1 << len_: rot *= rate2[(~s & -~s).bit_length() - 1] rot %= mod len_ += 1 else: p = 1 << (h - len_ - 2) rot = 1 for s in range(1 << len_): rot2 = rot * rot % mod rot3 = rot2 * rot % mod offset = s << (h - len_) for i in range(p): a0 = a[i + offset] a1 = a[i + offset + p] * rot a2 = a[i + offset + p * 2] * rot2 a3 = a[i + offset + p * 3] * rot3 a1na3imag = (a1 - a3) % mod * imag a[i + offset] = (a0 + a2 + a1 + a3) % mod a[i + offset + p] = (a0 + a2 - a1 - a3) % mod a[i + offset + p * 2] = (a0 - a2 + a1na3imag) % mod a[i + offset + p * 3] = (a0 - a2 - a1na3imag) % mod if s + 1 != 1 << len_: rot *= rate3[(~s & -~s).bit_length() - 1] rot %= mod len_ += 2 def butterfly_inv(a): n = len(a) h = (n - 1).bit_length() len_ = h while len_: if len_ == 1: p = 1 << (h - len_) irot = 1 for s in range(1 << (len_ - 1)): offset = s << (h - len_ + 1) for i in range(p): l = a[i + offset] r = a[i + offset + p] a[i + offset] = (l + r) % mod a[i + offset + p] = (l - r) * irot % mod if s + 1 != (1 << (len_ - 1)): irot *= irate2[(~s & -~s).bit_length() - 1] irot %= mod len_ -= 1 else: p = 1 << (h - len_) irot = 1 for s in range(1 << (len_ - 2)): irot2 = irot * irot % mod irot3 = irot2 * irot % mod offset = s << (h - len_ + 2) for i in range(p): a0 = a[i + offset] a1 = a[i + offset + p] a2 = a[i + offset + p * 2] a3 = a[i + offset + p * 3] a2na3iimag = (a2 - a3) * iimag % mod a[i + offset] = (a0 + a1 + a2 + a3) % mod a[i + offset + p] = (a0 - a1 + a2na3iimag) * irot % mod a[i + offset + p * 2] = (a0 + a1 - a2 - a3) * irot2 % mod a[i + offset + p * 3] = (a0 - a1 - a2na3iimag) * irot3 % mod if s + 1 != (1 << (len_ - 2)): irot *= irate3[(~s & -~s).bit_length() - 1] irot %= mod len_ -= 2 def convolution_naive(a, b): n = len(a) m = len(b) ans = [0] * (n + m - 1) if n < m: for j in range(m): for i in range(n): ans[i + j] = (ans[i + j] + a[i] * b[j]) % mod else: for i in range(n): for j in range(m): ans[i + j] = (ans[i + j] + a[i] * b[j]) % mod return ans def convolution_ntt(a, b): a = a.copy() b = b.copy() n = len(a) m = len(b) z = 1 << (n + m - 2).bit_length() a += [0] * (z - n) butterfly(a) b += [0] * (z - m) butterfly(b) for i in range(z): a[i] = a[i] * b[i] % mod butterfly_inv(a) a = a[:n + m - 1] iz = pow(z, mod - 2, mod) for i in range(n + m - 1): a[i] = a[i] * iz % mod return a def convolution_square(a): a = a.copy() n = len(a) z = 1 << (2 * n - 2).bit_length() a += [0] * (z - n) butterfly(a) for i in range(z): a[i] = a[i] * a[i] % mod butterfly_inv(a) a = a[:2 * n - 1] iz = pow(z, mod - 2, mod) for i in range(2 * n - 1): a[i] = a[i] * iz % mod return a def convolution(a, b): """It calculates (+, x) convolution in mod 998244353. Given two arrays a[0], a[1], ..., a[n - 1] and b[0], b[1], ..., b[m - 1], it calculates the array c of length n + m - 1, defined by > c[i] = sum(a[j] * b[i - j] for j in range(i + 1)) % 998244353. It returns an empty list if at least one of a and b are empty. Complexity ---------- > O(n log n), where n = len(a) + len(b). """ n = len(a) m = len(b) if n == 0 or m == 0: return [] if min(n, m) <= 60: return convolution_naive(a, b) if a is b: return convolution_square(a) return convolution_ntt(a, b) def integrate(a): a=a.copy() n = len(a) assert n > 0 a.pop() a.insert(0, 0) inv = [1, 1] for i in range(2, n): inv.append(-inv[mod%i] * (mod//i) % mod) a[i] = a[i] * inv[i] % mod return a def differentiate(a): n = len(a) assert n > 0 for i in range(2, n): a[i] = a[i] * i % mod a.pop(0) a.append(0) return a def inverse(a): n = len(a) assert n > 0 and a[0] != 0 res = [pow(a[0], mod - 2, mod)] m = 1 while m < n: f = a[:min(n,2*m)] + [0]*(2*m-min(n,2*m)) g = res + [0]*m butterfly(f) butterfly(g) for i in range(2*m): f[i] = f[i] * g[i] % mod butterfly_inv(f) f = f[m:] + [0]*m butterfly(f) for i in range(2*m): f[i] = f[i] * g[i] % mod butterfly_inv(f) iz = pow(2*m, mod-2, mod) iz = (-iz*iz) % mod for i in range(m): f[i] = f[i] * iz % mod res += f[:m] m <<= 1 return res[:n] def log(a): a = a.copy() n = len(a) assert n > 0 and a[0] == 1 a_inv = inverse(a) a=differentiate(a) a = convolution(a, a_inv)[:n] a=integrate(a) return a def exp(a): a = a.copy() n = len(a) assert n > 0 and a[0] == 0 g = [1] a[0] = 1 h_drv = a.copy() h_drv=differentiate(h_drv) m = 1 while m < n: f_fft = a[:m] + [0] * m butterfly(f_fft) if m > 1: _f = [f_fft[i] * g_fft[i] % mod for i in range(m)] butterfly_inv(_f) _f = _f[m // 2:] + [0] * (m // 2) butterfly(_f) for i in range(m): _f[i] = _f[i] * g_fft[i] % mod butterfly_inv(_f) _f = _f[:m//2] iz = pow(m, mod - 2, mod) iz *= -iz iz %= mod for i in range(m//2): _f[i] = _f[i] * iz % mod g.extend(_f) t = a[:m] t=differentiate(t) r = h_drv[:m - 1] r.append(0) butterfly(r) for i in range(m): r[i] = r[i] * f_fft[i] % mod butterfly_inv(r) im = pow(-m, mod - 2, mod) for i in range(m): r[i] = r[i] * im % mod for i in range(m): t[i] = (t[i] + r[i]) % mod t = [t[-1]] + t[:-1] t += [0] * m butterfly(t) g_fft = g + [0] * (2 * m - len(g)) butterfly(g_fft) for i in range(2 * m): t[i] = t[i] * g_fft[i] % mod butterfly_inv(t) t = t[:m] i2m = pow(2 * m, mod - 2, mod) for i in range(m): t[i] = t[i] * i2m % mod v = a[m:min(n, 2 * m)] v += [0] * (m - len(v)) t = [0] * (m - 1) + t + [0] t=integrate(t) for i in range(m): v[i] = (v[i] - t[m + i]) % mod v += [0] * m butterfly(v) for i in range(2 * m): v[i] = v[i] * f_fft[i] % mod butterfly_inv(v) v = v[:m] i2m = pow(2 * m, mod - 2, mod) for i in range(m): v[i] = v[i] * i2m % mod for i in range(min(n - m, m)): a[m + i] = v[i] m *= 2 return a def power(a,k): n = len(a) assert n>0 if k==0: return [1]+[0]*(n-1) l = 0 while l < len(a) and not a[l]: l += 1 if l * k >= n: return [0] * n ic = pow(a[l], mod - 2, mod) pc = pow(a[l], k, mod) a = log([a[i] * ic % mod for i in range(l, len(a))]) for i in range(len(a)): a[i] = a[i] * k % mod a = exp(a) for i in range(len(a)): a[i] = a[i] * pc % mod a = [0] * (l * k) + a[:n - l * k] return a def sqrt(a): if len(a) == 0: return [] if a[0] == 0: for d in range(1, len(a)): if a[d]: if d & 1: return None if len(a) - 1 < d // 2: break res=sqrt(a[d:]+[0]*(d//2)) if res == None: return None res = [0]*(d//2)+res return res return [0]*len(a) sqr = Tonelli_Shanks(a[0],mod) if sqr == None: return None T = [0] * (len(a)) T[0] = sqr res = T.copy() T[0] = pow(sqr,mod-2,mod) #T:res^{-1} m = 1 two_inv = (mod + 1) // 2 F = [sqr] while m <= len(a) - 1: for i in range(m): F[i] *= F[i] F[i] %= mod butterfly_inv(F) iz = pow(m, mod-2, mod) for i in range(m): F[i] = F[i] * iz % mod delta = [0] * (2 * m) for i in range(m): delta[i + m] = F[i] - a[i] - (a[i + m] if i+m len(a) - 1: break F = res[:2 * m] butterfly(F) eps = [F[i] * G[i] % mod for i in range(2 * m)] butterfly_inv(eps) for i in range(m): eps[i] = 0 iz = pow(2*m, mod-2, mod) for i in range(m,2*m): eps[i] = eps[i] * iz % mod butterfly(eps) for i in range(2 * m): eps[i] *= G[i] eps[i] %= mod butterfly_inv(eps) for i in range(m, 2 * m): T[i] = -eps[i]*iz T[i]%=mod iz = iz*iz % mod m <<= 1 return res def division_modulus(f,g): n=len(f) m=len(g) while m and g[m-1]==0: m-=1 assert m if n>=m: fR=f[::-1][:n-m+1] gR=g[:m][::-1][:n-m+1]+[0]*max(0,n-m+1-m) qR=convolution(fR,inverse(gR))[:n-m+1] q=qR[::-1] r=[(f[i]-x)%mod for i,x in enumerate(convolution(g,q)[:m-1])] while r and r[-1]==0: r.pop() else: q,r=[],f.copy() return q,r def taylor_shift(a,c): a=a.copy() n=len(a) #MD=MOD(mod) #MD.Build_Fact(n-1) for i in range(n): a[i]*=MD.Fact(i) a[i]%=mod C=[1] for i in range(1,n): C.append(C[-1]*c%mod) for i in range(n): C[i]*=MD.Fact_Inve(i) C[i]%=mod a=convolution(a,C[::-1])[n-1:] for i in range(n): a[i]*=MD.Fact_Inve(i) a[i]%=mod return a def multipoint_evaluation(f, x): n = len(x) sz = 1 << (n - 1).bit_length() g = [[1] for _ in range(2 * sz)] for i in range(n): g[i + sz] = [-x[i], 1] for i in range(1, sz)[::-1]: g[i] = convolution(g[2 * i],g[2 * i + 1]) g[1] =division_modulus(f,g[1])[1] for i in range(2, 2 * sz): g[i]=division_modulus(g[i>>1],g[i])[1] res = [g[i + sz][0] if g[i+sz] else 0 for i in range(n)] return res def Chirp_Z_transform(f,q,M): if q==0: if f: return f[0]%mod else: return 0 if M==0: return [] N=len(f) pow_q=[1]+[q]*(N+M-2) inve_q=pow(q,mod-2,mod) pow_inve_q=[1]+[inve_q]*(N+M-2) for _ in range(2): for i in range(1,N+M-1): pow_q[i]*=pow_q[i-1] pow_q[i]%=mod pow_inve_q[i]*=pow_inve_q[i-1] pow_inve_q[i]%=mod a=[f[i]*pow_inve_q[i]%mod for i in range(N-1,-1,-1)] b=pow_q ab=convolution(a,b) return [ab[j+N-1]*pow_inve_q[j]%mod for j in range(M)] T=int(readline()) mod=998244353 for t in range(T): A,N=map(int,readline().split()) d=max(1,int(N**.5)) poly=[0]*(d+1) pow_A=[1,A]+[A*A%mod]*(d-1) for _ in range(2): for j in range(1,d+1): pow_A[j]*=pow_A[j-1] pow_A[j]%=mod for j in range(d): poly[j]=pow(A,j*j,mod) #poly[j]=pow_A[j] S=Chirp_Z_transform(poly,pow(A,2*d,mod),N//d) ans=0 pow_Ad=[1,pow_A[d]]+[pow_A[d]*pow_A[d]%mod]*(N//d-1) for _ in range(2): for j in range(1,N//d+1): pow_Ad[j]*=pow_Ad[j-1] pow_Ad[j]%=mod for i in range(N//d): ans+=pow_Ad[i]*S[i]%mod ans%=mod for i in range(N-N%d,N): ans+=pow(A,i*i,mod) ans%=mod print(ans)