「咨询开源硬件」支援高频量化对冲算筹
Opened this issue · 2 comments
Best RAM You Can Buy Today
Patriot Viper Steel DDR4-4400 (2x 8GB) ...
Patriot Viper RGB DDR4-3600 (2x 8GB) ...
Patriot Viper 4 DDR4-3400 (2x 8GB) ...
Corsair Vengeance RGB Pro DDR4-3200 (4x 8GB) ...
Patriot Viper Steel DDR4-3200 (2x 16GB) ...
Patriot Viper Steel DDR4-3600 C18 (2x 32GB) ...
G. ...
Corsair Vengeance LPX DDR4-2666 (2x 8GB)
Oct 20, 2020
Best RAM 2020: Fast, Cheap and RGB | Tom's Hardware
Nvidia GeForce RTX 3080. The best graphics card for PC gaming right now. ...
Nvidia GeForce RTX 2070 Super. The best 4K graphics card for reasonable money. ...
AMD Radeon RX 5700. The best 1440p graphics card… with a little work. ...
AMD Radeon RX 5600 XT. ...
Nvidia GeForce GTX 1660 Super. ...
Nvidia GeForce GTX 1650 Super.
Oct 21, 2020
The best graphics cards in 2020 | PC Gamer
Need to upgraded my equipement to enchance the efficiency...
Originally posted by @englianhu in englianhu/ali-zonghui#9 (comment)
> mts <- smp %>%
+ msts(seasonal.periods = c(1440, nrow(smp)))
>
> mts <- llply(1:ncol(mts), function(i) {
+ y <- mts[,i] %>%
+ tbats %>%
+ forecast(h = 1440) %>%
+ as_tibble
+ names(y)[1] <- names(smp)[i]
+ y
+ }) %>%
+ bind_rows %>%
+ mutate(Model = factor('tbats'), Period = factor('dy.qt'), type = case_when(
+ !is.na(open) ~ 'open',
+ !is.na(high) ~ 'high',
+ !is.na(low) ~ 'low',
+ !is.na(close) ~ 'close')) %>%
+ dlply(.(type, Period), function(x) {
+ x %<>% dplyr::rename(open.Point.Forecast = open,
+ high.Point.Forecast = high,
+ low.Point.Forecast = low,
+ close.Point.Forecast = close)
+ names(x)[str_detect(names(x), '80|95')] <- paste0(x$type[1], '.', names(x)[str_detect(names(x), '80|95')])
+ x[colSums(!is.na(x)) > 0] %>%
+ data.frame %>%
+ as_tibble %>%
+ dplyr::select(-type)
+
+ }) %>%
+ join_all %>%
+ as_tibble
Joining by: Model, Period
Joining by: Model, Period
- Error: cannot allocate vector of size 11.1 Gb
> memory.size()
[1] 11984.01
> gc()
used (Mb) gc trigger (Mb) max used (Mb)
Ncells 3767976 201.3 7315067 390.7 7315067 390.7
Vcells 12814470 97.8 1474157596 11247.0 1534873241 11710.2
> memory.limit()
[1] 16274memory.size() occupied 75% of memory.limit()
> mts <- smp %>%
+ msts(seasonal.periods = c(1440, nrow(smp)))
>
> mts <- llply(1:ncol(mts), function(i) {
+ y <- mts[,i] %>%
+ tbats %>%
+ forecast(h = 1440) %>%
+ as_tibble
+ names(y)[1] <- names(smp)[i]
+ y
+ }) %>%
+ bind_rows %>%
+ mutate(Model = factor('tbats'), Period = factor('dy.qt'), type = case_when(
+ !is.na(open) ~ 'open',
+ !is.na(high) ~ 'high',
+ !is.na(low) ~ 'low',
+ !is.na(close) ~ 'close')) %>%
+ dlply(.(type, Period), function(x) {
+ x %<>% dplyr::rename(open.Point.Forecast = open,
+ high.Point.Forecast = high,
+ low.Point.Forecast = low,
+ close.Point.Forecast = close)
+ names(x)[str_detect(names(x), '80|95')] <- paste0(x$type[1], '.', names(x)[str_detect(names(x), '80|95')])
+ x[colSums(!is.na(x)) > 0] %>%
+ data.frame %>%
+ as_tibble %>%
+ dplyr::select(-type)
+
+ })
> mtss <- mts %>% bind_cols
New names:
* Model -> Model...6
* Period -> Period...7
* Model -> Model...13
* Period -> Period...14
* Model -> Model...20
* ...
> mtss
# A tibble: 1,440 x 28
close.Lo.80 close.Hi.80 close.Lo.95 close.Hi.95 close.Point.For~ Model...6
<dbl> <dbl> <dbl> <dbl> <dbl> <fct>
1 118. 118. 118. 118. 118. tbats
2 118. 118. 118. 118. 118. tbats
3 118. 118. 118. 118. 118. tbats
4 118. 118. 118. 118. 118. tbats
5 118. 118. 118. 118. 118. tbats
6 118. 118. 118. 118. 118. tbats
7 118. 118. 118. 118. 118. tbats
8 118. 118. 118. 118. 118. tbats
9 118. 118. 118. 118. 118. tbats
10 118. 118. 118. 118. 118. tbats
# ... with 1,430 more rows, and 22 more variables: Period...7 <fct>,
# high.Lo.80 <dbl>, high.Hi.80 <dbl>, high.Lo.95 <dbl>, high.Hi.95 <dbl>,
# high.Point.Forecast <dbl>, Model...13 <fct>, Period...14 <fct>,
# low.Lo.80 <dbl>, low.Hi.80 <dbl>, low.Lo.95 <dbl>, low.Hi.95 <dbl>,
# low.Point.Forecast <dbl>, Model...20 <fct>, Period...21 <fct>,
# open.Point.Forecast <dbl>, open.Lo.80 <dbl>, open.Hi.80 <dbl>,
# open.Lo.95 <dbl>, open.Hi.95 <dbl>, Model...27 <fct>, Period...28 <fct>
> mtss[str_detect(names(mtss), 'Model.|Period.')]
# A tibble: 1,440 x 8
Model...6 Period...7 Model...13 Period...14 Model...20 Period...21 Model...27
<fct> <fct> <fct> <fct> <fct> <fct> <fct>
1 tbats dy.qt tbats dy.qt tbats dy.qt tbats
2 tbats dy.qt tbats dy.qt tbats dy.qt tbats
3 tbats dy.qt tbats dy.qt tbats dy.qt tbats
4 tbats dy.qt tbats dy.qt tbats dy.qt tbats
5 tbats dy.qt tbats dy.qt tbats dy.qt tbats
6 tbats dy.qt tbats dy.qt tbats dy.qt tbats
7 tbats dy.qt tbats dy.qt tbats dy.qt tbats
8 tbats dy.qt tbats dy.qt tbats dy.qt tbats
9 tbats dy.qt tbats dy.qt tbats dy.qt tbats
10 tbats dy.qt tbats dy.qt tbats dy.qt tbats
# ... with 1,430 more rows, and 1 more variable: Period...28 <fct>
> mtss[str_detect(names(mtss), 'Model.|Period.')] <- NULL
> mtssI used bind_cols() as alternative method to solve my issue, there has different methods depends on your issue (example: normally matrix will be more efficient than data.frame)...
lineprof in Advanced R by Hadley Wickham
microbenchmark in Advanced R by Hadley Wickham
There has eventually need to use better pc for high efficiency computing...
