#----------------------------------------------------------------------
# Python interface for ISPACK3
# Copyright (C) 2023--2024 Toshiki Matsushima <toshiki@gfd-dennou.org>
#
# This library is free software; you can redistribute it and/or
# modify it under the terms of the GNU Lesser General Public
# License as published by the Free Software Foundation; either
# version 2.1 of the License, or (at your option) any later version.
#
# This library is distributed in the hope that it will be useful,
# but WITHOUT ANY WARRANTY; without even the implied warranty of
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
# Lesser General Public License for more details.
# 
# You should have received a copy of the GNU Lesser General Public
# License along with this library; if not, write to the Free Software
# Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
# 02110-1301 USA.
#----------------------------------------------------------------------
from mpi4py import MPI
import numpy as np
import time
import ispack3 as isp

jm  = 2**10
ntr = 10
mm  = jm-1
im  = jm*2
nm  = mm
nn  = nm

ig=1
ipow=0

comm = MPI.COMM_WORLD
rank = comm.Get_rank()
np_size = comm.Get_size()

jv = isp.syqrjv(jm)
it, t, r = isp.syini1(mm, nm, im, comm)
p, jc = isp.syini2(mm, nm, jm, ig, r, comm)

g_shape = (2, ((jm//jv-1)//np_size+1)*jv, im)
w_shape = (2 * jv * ((jm // jv - 1) // np_size + 1) * (mm // np_size + 1) * np_size * 2 * 2)

s_shape = (2, (mm // np_size + 1) * (2 * (nn + 1) - mm // np_size * np_size), )

G = isp.aligned_array(g_shape, align=64)
W = isp.aligned_array(w_shape, dtype=np.float64, align=64)

S = np.empty(s_shape, dtype=np.float64)

if comm.Get_rank() == 0:
    sall_shape = (2, (2 * nn + 1 - mm) * mm + nn + 1)
else:
    sall_shape = (2, 0)

SALL = np.empty(sall_shape, dtype=np.float64)

np.random.seed(0)
SALL[0,:] = 2 * np.random.rand(SALL.shape[1]) - 1
SALL[1,:] = -SALL[0,:]
SDALL = np.empty(SALL.shape, dtype=np.float64)

for v in range(2):
    isp.syss2s(mm, nn, SALL[v,:], S[v,:], comm)

if(rank==0):
    print("MM=", mm, ", IM=", im, ", JM=", jm, ", JV=", jv, ", NTR=", ntr)
    print("SSE=", isp.mxgcpu())
    print("number of threads =", isp.mxgomp())
    print("number of processes =", np_size)    

rc=(5*im*np.log(im)/np.log(2.0)*0.5*jm+(mm+1)*(mm+1)*jm)*2
    
start_time = time.perf_counter()

for n in range(ntr):    
    isp.syts2v(mm, nm, nm, im, jm, jv, S[0,:], S[1,:], G[0,:,:], G[1,:,:], it, t, p, r, jc, W, ipow, comm)
    
elapsed_time = time.perf_counter() - start_time
    
GFLOPS=rc*ntr/elapsed_time/10**9

if(rank==0):    
    print("S2V:", elapsed_time/ntr, "sec (", GFLOPS, "GFlops)")

start_time = time.perf_counter()

for n in range(ntr):
    isp.sytv2s(mm, nm, nm, im, jm, jv, S[0,:], S[1,:], G[0,:,:], G[1,:,:], it, t, p, r, jc, W, ipow, comm)

elapsed_time = time.perf_counter() - start_time
    
GFLOPS=rc*ntr/elapsed_time/10**9

if(rank==0):    
    print("V2S:", elapsed_time/ntr, "sec (", GFLOPS, "GFlops)")

for v in range(2):
    isp.sygs2s(mm, nn, S[v,:], SDALL[v,:], comm)


if(rank==0):

    l = np.arange(len(SALL[0,:]))
    n, m = isp.sxl2nm(nn, l)

    SL_values = np.zeros_like(SALL, dtype=np.float64)
    
    for v in range(2):
        SL_values[v, m == 0] = np.abs(SDALL[v,m == 0] - SALL[v,m == 0])
        SL_values[v, m > 0] = np.sqrt((SDALL[v,m > 0] - SALL[v,m > 0])**2 + (SDALL[v,m < 0] - SALL[v,m < 0])**2)

    for v in range(2):
        
        SLMAX = np.max(SL_values[v,:])
        LAS = l[np.argmax(SL_values[v,:])]

        SLAMAX = np.sum(SL_values[v,:]**2)
        
        n, m = isp.sxl2nm(mm, LAS)
        
        print("maxerror=", SLMAX, "(n=", n, ", m=", m, ")" )
        print("rmserror=", np.sqrt(SLAMAX/((mm+1)*(mm+2)/2) ) )